Welcome to The Progressive Corporation first quarter investor event. The company will now make detailed comments related to quarterly results in addition to those provided in its quarterly report on Form 10-Q and the letter to shareholders, which have been posted to the company's website and will use this event to respond to questions after a prepared presentation by the company. This event is available via a moderated conference call line and a live webcast with a brief delay. Webcast participants will be able to view the presentation slides live or download them from the webcast site. Participants on the phone can access the slides from the Events page at investors.progressive.com. In the event we encounter any technical difficulty with the webcast transmission, webcast participants can connect through the conference call line. The dial-in information and passcode are available on the Events page at investors.progressive.com.
Acting as a moderator for the event will be Ms. Julia Hornick. At this time, I will turn the event over to Ms. Hornick. Go ahead.
Thank you, Victor, good afternoon to all. Today, we will begin with a presentation on operational efficiency and claims accuracy. Our presentation will be followed by Q&A with our CEO, Tricia Griffith, our CFO, John Sauerland, and our guest speakers today, Mike Esposito and Kia Berglund. Our Chief Investment Officer, Bill Cody, will join us by telephone for Q&A. As always, discussions in this event may include forward-looking statements. These statements are based on management's current expectations and are subject to many risks and uncertainties that could cause actual events and results to differ materially from those discussed during the event. Additional information concerning those risks and uncertainties is available on our 2017 annual report on Form 10-K , where you can find discussions of the risk factors affecting our businesses, safe harbor statements related to forward-looking statements, and other discussions of the challenges we face.
These documents can be found via the investors page of our website, progressive.com. It's now my pleasure to turn it over to our CEO, Tricia Griffith.
Good afternoon. Welcome to Progressive's first quarter webcast. Thanks for joining us. Last time we met, we outlined a construct that we call our four cornerstones. It's really how we win and win in the right way, who we are, our core values, why we are here, our purpose statement, where we are headed, our vision, and how we are going to get there, our strategy. We focused on the strategy last session. We will continue to do so in this session. Very specifically, we focused last time on maintaining a brand. We had Dan Witalec come from our acquisitions group, he leads that group, really talk about the science of acquisition, understanding how to be efficient with how much money we spend on our marketing and our branding.
We also had Kat Kalogi come in from our marketing group to talk about more of the art around it, the right messages to the right consumer at the right time, whether you're a Sam, Diane, Wright, or Robinson. Those two together with art and science are really maintaining a great brand. This is really important, especially as you've probably likely read in the Q that we've spent over 30% this quarter as compared to last year on marketing. We are extraordinarily pleased with the results. As you can see by our results, it's working. We are growing new apps. We're able to retain those customers. It's really a great story. As long as we can be efficient, we will continue to spend in that area. We look at spend in terms of efficiency of cost per sale being lower than our allowable spend.
We call that here targeted acquisition cost, TAC. As long as our cost per sale is less than TAC, we will continue to spend because we see that as efficient. We're really excited about the results in going forward. Today, we're going to focus on offering competitive prices. That's another one of the pillars within our strategy. We're going to focus very particularly on operational efficiency. Operational efficiency is clearly important in a price-sensitive industry. It's not just about that. We want to make sure that we care about meeting our customer's need, creating value for our shareholders, and making this an engaged place to work so our workforce knows where we're headed so we can ultimately achieve our vision of becoming consumers' number one choice. Really, efficiency and competitive prices are complementary to industry-leading segmentation and claims accuracy.
Let me break down the components of Progressive expenses. We break them down roughly in a third, a third, a third. The first third is acquisition expenses, which Dan Witalec talked about in detail during the last webcast. The other portion of our overall expense ratio is what we call non-acquisition expense ratio. I think that acronym is specific to Progressive. When you hear it today, you'll hear us say NAER, or N-A-E-R. That's what we're talking about. Mike Esposito, our PL controller, will talk about that from that perspective. The other third is loss adjustment expense. We have Kia Berglund, our claims controller, who will talk about LAE in conjunction with all other three guiding principles and how important it is to make sure we have a near-perfect balance of all of those guiding principles. We call this our efficiency ratio graph.
If you take a look at it, you'll see the top 10 private passenger auto insurers plotted on the graph with the industry average. On the Y-axis, you'll see a 5-year compounded annual growth rate of net written premium, and on the X-axis is the efficiency ratio. One minus ER plus LAE. Clearly, there's a correlation of those companies that do better from an efficiency perspective than industry as correlated to their growth. We're excited to be one of the very few companies that are better than the industry. In fact, if you look at 2017 data, and frankly, the last 10 years, we have about a 3.5 point advantage based on our efficiency ratio compared to the industry. In a really price-sensitive industry, that advantage is significant. We want to continue to maintain that. That's why operational efficiency is critical to our business model.
This chart shows where we pay the most out of the $ premium that comes in. It's losses plus LAE. This is data from 2013 to the end of last year. The blue line is Progressive and the orange line is the industry. You can see, we've continued to and actually grown the gap between us and the industry. That's why it's so important to not just care about efficiency, you have to care about accuracy. You can have a process change, whether it's technology related or process related, and if you don't test it and make sure it doesn't have negative second-order effects on the accuracy or the indemnity $, things can go not so well. We really pride ourselves on making sure we test every change.
We want to be very efficient on the LAE side, but only if it also works on the indemnity side. We're very proud of these results. Lastly, I think you've seen this chart before, this is really, again, the top 10 private passenger auto carriers. These are statutory results based on growth and profitability. You can see, it's one thing to grow, it's one thing to be profitable, but growing profitably is really tough, and there's very few companies in this area. The combined ratio is the X-axis, and it's inverted. You want to be the company in the top right-hand quadrant of this graph. We happen to be there, and we've been there for quite a while. Oftentimes, internally, we say that's a really lonely piece of real estate, and we'd like to keep it that way.
If you plotted 2018, the first quarter, we'd actually be a little bit higher and to the right. Again, extraordinary results that we'd love to share with you today. That said, we also know that our competitors have been taking rates and a lot of them have a tailwind from the tax reform. There's a lot of pressure on us to make sure we maintain this gap, and we actually appreciate that pressure. We think of pressure as a great challenge for us to continue to think about our cornerstones and how to build better models and be more efficient while serving our customers and our shareholders. With that, I'd like to invite Mike Esposito up to talk about what he's doing on the Personal Lines side in terms of efficiency.
Thank you, Tricia. Just to round out the setup for this section, when we talk about non-acquisition expense ratio, we exclude costs associated with our direct media and agent commissions. We include costs associated with our customer-facing sales and service processes and all the activities we undertake to create and support our products and services. As you can see, over a 10-year period, we've been able to take out over two points from our non-acquisition expense ratio, or NAER, as we call it, which, as Tricia noted, puts us in a very advantageous position in the market. What I'd like to go over today is how we've achieved these results and what we expect to achieve going forward. I'd like to do that by talking about four levers of non-acquisition expense advantage. First, economies of scale. Second, products and experiences that we design for efficiency. Third, improving retention.
Fourth is running more preferred business, including Robinsons, which is our Destination Era strategy. I'll start by talking about economies of scale. We, like other large companies, gain economies of scale or spread fixed costs through things like real estate and IT. In our case, we build large campus contact centers that are more efficient than having numerous smaller offices. In IT, we build systems that can handle an increasing number of transactions without increasing costs at the same rate. What I'm showing here on this slide is those real estate costs as a % of earned premium indexed over time, and the same thing for IT. You can see that we've reduced these real estate expenses by 51% over this period and these IT expenses by 22%.
These gains are great. There is more to this story than just growing into a fixed cost base. For example, in real estate, in addition to scale, we practice active cost management, which means we optimize the space we need. For the space we use, we negotiate better rates. In addition, we enable fixed cost leverage through the implementation of a home-based phone consultant model in our customer relationship management or CRM organization. This model gives us greater access to the talent we need to support our business and provides flexibility that employees value. In addition to these benefits, this model enables us to grow our staff to support our growing business without a commensurate increase in space.
You can see in the chart on the right that at the end of 2017, over 40% of our CRM phone consultants were home-based. That proportion has more than doubled since 2013. In IT, scalable systems do matter. It's important to note also that a significant portion of IT expenses are labor, the professionals that develop and support our systems. What's been happening in IT is representative of what's been happening all across Progressive, and that is vigilance on headcount growth. To that end, we recently segmented our workforce into two categories. One is volume driven, which is headcount that we manage to a service level. Therefore, we expect to increase as our business grows. The second is non-volume driven, which is headcount that we intend to hold flat even as our business grows. Volume driven would be in organizations such as CRM and claims.
Non-volume driven would be in organizations like IT. You can see that since we implemented this segmentation at the end of 2016, we have indeed held our non-volume driven headcount flat to slightly declining, while volume driven headcount has increased significantly. This has enabled us continue a long-term trend of a decreasing proportion of premium going to salaries, which as you can see, has come down by 15% since 2009. To be clear, we continue to invest in IT to develop the products and services our customers value. Over this period, since 2006, our IT investment has increased by 38%. The message here is that even as we increase that investment, we intend to be efficient with that investment, like we are with every investment we make.
I'd like to transition from this macro level to the micro level and talk about how we achieve customer-level cost reductions through efficient products and services. I'd like to do that by talking about Brea, who's a real customer who demonstrates how we achieve efficiency by reducing the need for service, by enabling self-service, and through automation. What I'm showing here is a timeline of Brea's notable interactions with Progressive. You can see that she purchased an auto policy at the end of 2012, and about a year later, she informed us that she had moved and become a homeowner. At that time, she purchased a home insurance policy through Progressive Home Advantage, and therefore had moved from a Diane in our consumer segmentation to a Robinson. In 2014, she had a chat session, and in 2016, she downloaded our mobile app.
In 2017, she became more active using our online services, our mobile app, and chat. These are not an exhaustive list of Brea's interactions with Progressive, but they do demonstrate how we're able to meet customers' needs while being efficient. I'd like to take you deeper into those interactions to demonstrate how they impact NER. First, Brea's auto policy purchase. We invest in our quoting and buying processes in efficient workflows and high conversion for do-it-yourself quoters and for agents, regardless of whether you think of agents as Progressive employees or one of the many thousands of independent agents who represent Progressive. We invest in reading accuracy at point of sale to reduce post-sale work. We optimize our use of external data, leveraging our analytical capabilities. We promote low-cost options using our pricing segmentation and user experience skills.
In Brea's case, she completed her application using our electronic signature or e-sign process. She chose an EFT or electronic funds transfer bill plan, and she chose paperless for her policy, meaning to receive communication from us electronically. In addition to a lower rate for Brea, these choices reduce cost during the application process and set her up to be low cost going forward. You can see in the chart on the right that Brea is not alone. Almost 70% of our auto policies in force are paperless. About three-quarters of our policies in force are either paid in full bill plan, which is another billing option that reduces service cost, or EFT. The next set of interactions are where Brea had a servicing need, and she chose to take advantage of our self-service options.
You can see a couple of times where Brea downloaded her ID card, and she used online servicing to make a coverage change. Brea is not alone. About 80% of our ID cards are delivered digitally. Endorsements, like coverage changes through our interactive digital channels, which include web browsers and mobile apps, while not the majority, are increasing over time and currently stand about 24%. Brea also makes good use of our mobile app. She downloaded the app in 2016. In 2017, she used it to request roadside assistance, to check her billing history, and to move an EFT withdrawal date. Brea is representative of a trend to increasing digital interactions with Progressive, particularly through our mobile app. What I'm showing here is an index of inbound interaction counts over a recent three-year period. Inbound interactions are ID cards, endorsements, and payments.
You can see that over this period, which was a period where we're increasing our policy in force count, that we had very moderate changes in inbound interaction counts through web browser, through phone calls, and through speech self-service. While over this period, the interaction counts through our mobile app increased by almost eight times. We want to make things easy for our customers, and our mobile app is a great way to accomplish that goal. In addition to self-service, Brea took advantage of our chat offerings. I'd like to show you her first interaction. In this case, Brea had a question about her policy payments, which had decreased.
Brea was able to get verification that indeed the payment amount was correct. The happy surprise was able to use chat to request a coverage change. She used chat to get help for a coverage change, which she was then able to complete on her own. Chat is becoming an increasingly preferred mode of interaction for our customers. In addition to meeting customer needs, chat provides an additional potential benefit. We've recently developed an artificial intelligence experience in chat, which internally we call Chai, which automates the initial response to intents, as they're called, or categories of service requests, thereby reducing the need for human interaction and making chats more efficient. I'd like to show you what that looks like. In this case, the customer types their question or service request into a dialog box.
Chai determines if the intent is in scope and provides an automated initial response in addition to links with related content. If the customer's request is not fulfilled with Chai, they always have the option to request interaction with a live representative. We recently tested Chai in a three-month period in a 50/50 A/B test. For the intents and scope, we were able to satisfy the customer's need with an automated response, thereby reducing the need for human intervention, which I'm showing here in this chart. The intents and scope are the orange line, the 50% that went to Chai, the gray line. Roughly half of that time, the blue line, we were able to respond with Chai. With the success of the 50/50 period, we increased the test portion to 90%. You can see the gray line go up. We had similar effectiveness.
We'll continue to test and expand the use of Chai to make these interactions more efficient. These successes on more efficient products and experience at the customer level, combined with system changes we make behind the scenes that enable more efficient workflows, have resulted in decreases in the drivers of cost to serve our customers, which we classify as operations expense in NAER. I'm showing two examples on that. On the left, you have workload in our processing group. This is the group in CRM who handles incoming documents, emails, and chats. I'm showing workload indexed over time on a policy in force basis. You can see that in this period, we've decreased that workload per policy in force by almost 20%. On the right, I'm showing postage.
You can see similarly, we've decreased the postage per policy in force count in this period over 40% through features such as paperless and electronic ID cards. At the customer level, our success in avoiding the need for service, enabling self-service and automation and operational efficiency has decreased our cost to serve our customers on a unit basis. I'm showing that here on the blue line. You can see from 2006 to 2008, we had a significant decrease in our operations cost per policy in force, and that was in large part driven by an organizational change we made where we combined what were formerly channel-specific activities in order to increase efficiency. We were able to continue that downward trend in operations cost per policy in force over that period to 2017.
With the benefit of a growing premium base, we were able to decrease operations as a percent of earned premium by 30% in this period. To be clear, we continue to hold high standards for customer service. We know our customers have options. You've heard about our recent gains in retention, and that's one of the things we continue to look at to make sure that we are meeting customer needs while we pursue efficiency. We have goals to continue to reduce NAER, and we have actions underway to achieve those goals. For example, we're in the process of upgrading our online servicing system, which we expect to result in many more customer-initiated self-service transactions to be successfully completed online, avoiding the need for a costlier phone call.
We are expanding our use of automation and artificial intelligence, like we have for chat, but focused on the incoming documents we receive. Lastly, we're developing the next generation of our interactive voice response system, which we expect to enable more speech self-service transactions, and in the cases where expert help is needed from our consultants, to more efficiently route those calls using natural language understanding and machine learning capabilities. As valuable as it is to lower cost and therefore lower rates overall, it's even more valuable to offer lower rates to lower cost customers. We enable that by including operations pricing in our highly segmented pricing model. What I'm showing here is a representative set of factors in our latest product model. As you would expect from us, this is a multivariate equation, so this is a combination of five different variables categorized into deciles.
You can see that the factors vary from the lowest to the highest by more than four times. We're not alone in this. Our competitors also price using operations. We believe our use in pricing, like our pricing for loss cost, is more sophisticated than our competitors. We'll continue to advance the science as we continue to achieve cost reductions. I'd like to transition from the micro level back up to the macro level and talk briefly about the impact of retention and our Destination Era strategy. In retention, you're probably familiar with the fact that loss ratios tend to be higher in new business periods, and the same phenomena exists for NAER.
You recall that I said that NAER excludes costs associated with direct media and agent commissions, and even excluding those costs, you can see in the chart on the left that renewal NAER is about 35% less than new business NAER. What that means is as we increase retention and as we have more low-cost renewal terms in our mix of business, we will have lower NAER. In a similar vein, as we write more preferred business, we will also lower NAER. In part, that's because more preferred business does retain longer, and therefore, we get the benefit of having those more numerous renewal terms. Even looking only at renewal terms, which I'm showing here in the chart on the right, you can see that the relativity between Sams and Dianes and Wrights and Robinsons is about 10 points, with Robinsons being the lowest.
What this means is that our Destination Era strategy, in addition to expanding our addressable market and fueling growth, will help us attract lower cost customers. I'd like to take us back to the 2006 to 2017 view. What I'm showing here is a decomposition of our index of NAER over time, broken into 2 components. One is cost per policy in force, the other is earned premium in policy in force. You can see that from the period from 2006 to 2011, we were in a deflationary period for premium. You can also see, and that's the blue line, by the way, you can also see through the RFZ that we're able to decrease cost faster than the reduction in premium, thereby achieving a moderate decrease in NAER.
You can also see that after that period in 2011, we continued a trend of decreasing cost per policy in force until just recently. The uptick that you see is driven by 2 things. One is there are costs within NAER that are proportional to premium, such as premium tax and credit card fees. Two, our recently strong new business growth does put upward pressure on NAER. You'll also note since 2011 that premium trend has increased significantly and that we've been able to keep our cost trend lower than premium trend. Going forward, we expect that we'll be able to continue the long-term trend of decreasing cost per policy in force and that regardless of the direction of premium, we'll be able to keep our cost trend lower, thereby decreasing NAER and positioning us well for any market conditions.
In summary, our strong growth enables fixed cost leverage economies of scale while we continually search for efficiencies. Our strong capabilities in segmentation, technology, analysis, and user experience allow us to develop products and services that meet customer needs while being efficient. Our segmented pricing allows us to offer lower rates to lower cost customers. Increasing retention will result in more numerous lower cost tenure periods in our books of business. Our Destination Era strategy, in addition to expanding our addressable market, will help us attract lower cost customers. All these elements combined help us achieve a competitive advantage and enable us to put a larger share of premium towards what customers value the most, and that is settling claims. With that, I will hand it off to Kia.
Thanks, Mike. In claims, we contribute to Progressive's growth and profitability by resolving claims quickly, efficiently, and paying the right amount every time. As you know, the greatest portion of our premium dollars comes from claims, $0.75 on the dollar approximately. Naturally, it's of paramount importance that we get that right. After all, this is the product we sell and the promise we make with our customers. We believe we can excel by focusing on the four guiding principles, accuracy, efficiency, work environment, and customer service, and over time, that focus has served us extremely well. This chart should look familiar to many of you, as we've shared it previously. The Y-axis represents loss cost plus LAE, and the X-axis represents LAE.
For those of you who haven't seen this before, if we don't spend the right amount of time determining how much to pay, the accuracy or quality of our claims handling will deteriorate. If we spend too much time on non-value-added activities, our cost structure will increase, and again, we will not achieve the right outcome. Our goal is to find the optimal balance between accuracy and efficiency, i.e. movement down the curve, while at the same time looking for ways to push the curve downward through either process or technology changes. We're constantly in the details, looking for ways to deliver higher quality service at a lower cost structure. When we find those opportunities, our claims organization can execute, as our track record demonstrates. This chart shows the blue line shows LAE as a % of net earned premium, and the gray bars are net written premium.
You can see that since 2006, we've been able to lower LAE by almost two points. I'll walk through some of the ways in which we've been able to do this. Through 2014, we made significant efficiency gains through a series of process and organizational changes focused on three key areas. First, increased segmentation of our claims handling to better match claims complexity with rep skill level. Secondly, by centralizing aspects of our claims handling to achieve greater economies of scale. Third, by creating more streamlined and greater accountability through the introduction of our file owner and managed repair rep roles. More recently, we've made the intentional decision to allow LAE to increase to better position ourselves for expected policy and future growth.
Beginning in 2015, we began to add staff ahead of growth, while at the same time making investments in technology to support work from home, fraud investigations, and the introduction of photo estimating. This trade-off was the right decision to ensure that the accuracy of our claims handling did not deteriorate during a period of high growth. Bottom line is we have found ways to become more lean and increase productivity while still optimizing along the curve. This has then had a direct impact on our ability to offer more competitive prices. Our efforts to find that right balance between accuracy and efficiency have paid off, as you can see with this orange line. This line reflects the % of files that meet our quality standards, as assessed by an independent group which reports up through our Chief Financial Officer, John Sauerland.
This independent assessment is a valuable control for us, but it's not the only way that we measure quality. We also do audits and diagnostic file reviews across our claims organization. There are a lot of eyes on quality, and it is something we take very seriously, especially when we are growing and adding staff like we are now. In addition to monitoring LAE, we also look at the cost of handling claims on a per feature basis. Tracking both LAE and cost per feature allows us to stay focused on cost containment even during periods of high written premium growth. This chart shows cost per feature adjusted for average paid severity. That adjustment allows us to account for mix shifts over time in the complexity of claims or type of claims we handle.
Here, you can see from this chart that we have made strong progress in reducing cost per feature over time, with the recent upward trend we discussed earlier. As I mentioned before, quality is not something we're willing to compromise on. Making sure we have the right person on the right claim is really important. Again, we're very comfortable making a deliberate decision to add staff ahead of expected growth, and that investment has paid off. As you can see here, again, with the orange line representing quality, you can see that we've maintained very strong quality results even during the more recent period of growth. The chart on the right helps to further illustrate the recent growth in our claims organization. The line represents the % of individuals with less than one year job tenure. From 2014 to 2016, this percentage increased by almost 12 points.
This increase in hiring requires tight coordination across our recruiting, onboarding, and training organizations. The increase in less tenured reps was anticipated and planned for, through appropriate leadership oversight, process and training support, along with monitoring mechanisms to ensure the quality of our claims handling did not deteriorate. Handling a claim is our moment of truth. It's the most tangible aspect of our product. We will continue to have a laser focus on our ability to deliver high-quality outcomes and ensure we have the right staff in place to do so. Overall, we believe this track record of LAE and cost per feature improvement really highlights the effectiveness of our strategy and the ability of our claims organization to drive for results and deliver superior outcomes.
Now, I'd like to look ahead and shift our focus to the future and share our plans for improvement across all four guiding principles. A couple of years ago, Andrew Quigg spoke to you all about the power of the virtuous cycle of data and our broader efforts at The Progressive Corporation to capture more data. Similar to Andrew, I started out as a product manager and witnessed the power of this cycle firsthand as it relates to pricing and segmentation. When we write new business, we know a little bit about our customers. We amass more data through our experience with them. That internal data set can then be used to derive insights that lead to pricing segmentation advancements. That then helps us to offer more competitive prices, which leads to growth. That growth then leads to more data.
This is the virtuous cycle in pricing. It's something that we manage extremely well at The Progressive Corporation. This concept can also be applied to claims, where the advantages from more data are similar. More relationships and experiences translates to more data. Insights from that data can then be used to improve our claims handling across all four guiding principles. That then supports The Progressive Corporation's ability to offer more competitive prices, providing a benefit back to the customer, while also generating more growth, either through new business acquisition or by extending the retention of existing customers. That then generates more data and so on. You can see that the cycle is very much relevant to claims as well.
Add to that an explosion of data from connected cars and advancements in analytics and artificial intelligence in the marketplace. You have a very exciting landscape of opportunities for The Progressive Corporation in an area where we have traditionally excelled. Looking forward, we see tremendous opportunity to apply our competitive strengths in data and analytics to manage this dynamic virtuous cycle of data and explore applications for advanced analytics and artificial intelligence in claims handling. Couple that with our proven track record of execution and incredibly talented claims professionals. We believe we will be able to move the LAE plus loss curve significantly in the future to our advantage. Let's take a look at how this applies across our four guiding principles so that I can share how these themes come together in the future. We'll start with accuracy.
We see enormous potential to capitalize the rapidly expanding set of data to improve the accuracy of our claims handling. Internally, we like to refer to this as a tsunami of data coming from connected cars, mobile devices, policy data, and other sources. There's no doubt that this data and our proven ability to leverage data throughout our workflows will provide significant competitive advantages. Real-time access to this data will better inform and automate our loss triage process. This means getting the claim in the right hands faster, further compressing our report to assignment times, and jumpstarting claims handling and ultimate resolution. This is a big deal for us. Reducing cycle time can lead to significant savings in rental and storage fees, among other cost drivers.
Data and more accurate triage then helps to expedite accurate investigations, including the ability to identify and investigate fraud, promote more precise damage evaluations, and most certainly ensure we continue to pay the right amount in the settlement of valid claim exposures. Let me share a compelling example from an actual claim to help illustrate the power of utilizing data in a real-time fashion to help improve the accuracy of our claims investigations. Here's what our rep knew at the time of the loss. This is a new business policy, and our customer had signed up for the Snapshot mobile app. It's a single-car policy with an older vehicle and no prior insurance. A couple of weeks prior to the incident, our insured added physical damage.
When our customer called to report the loss, they stated that they swerved to avoid an unknown claimant vehicle and hit a bridge abutment with significant force. You can see from the damage in the photos. There was no police report. Collectively, these data points raise flags about the authenticity of the loss and suggest the need for an additional investigation. In this case, the outcome of that investigation was to afford coverage for our insured. However, if our rep had been able to access trip data from the Snapshot mobile app, it probably would have altered the course of the investigation. Now, I want to quickly mention, we don't currently use data from our Snapshot mobile app in the handling of claims without permission from our insured.
However, that may change in the future in light of the research we're doing, and we'll be very transparent and open at that time. Let's take a look at the trip data. We'll start with the trips leading up to the accident. I think you'll see that it raises some questions. To begin with, the red dot shows the location of the accident. You can see a picture of the bridge they hit as well. In the morning, prior to the trip resulting in the loss, our insured logs the trip beginning in this warehouse area, travels northwest, circling the area where the accident later occurs, and then arriving as a residential location. Just a short while later, we see another trip starting in a commercial area. A little research reveals some salvage yards in the vicinity.
Once again, the insured travels north, stopping briefly in the same residential area from the prior trip, and then passing the accident location once before finally stopping at the scene of the accident. There could be a reasonable explanation for these trips, but especially in conjunction with the other flags, it raises questions. Now, let's take a closer look at the final trip resulting in the loss. If we recall from the insured's statement, we would expect to see a very abrupt stop as they lose control of the vehicle, swerving to avoid an unknown claimant vehicle, and then hitting the bridge abutment with enough force to do significant damage to the vehicle. Yet when we look at the trip data in the chart to the left, we don't see the sudden deceleration that signifies an accident where we would expect to see it.
The only significant deceleration we see is when the vehicle comes to a stop at this intersection. Note the yellow line. If we follow the rest of the trip, we don't see any more stops or yellow lines. There's a brief stop at the same residential complex. Then we see the vehicle come to a more gradual stop at the scene of the accident. For the story to hold up, we'd expect to see another yellow line in the scene of the accident where the crash occurs. When you look at the data in its totality, this accident looks less and less authentic and more like a staged accident with a previously damaged vehicle, potentially even towed to the scene of the loss. Hindsight's always 20/20. This could look like really low-hanging fruit, but personally, I find that really exciting.
Even with the results we are delivering today, we find ways to improve in very meaningful manners. This trip's data is also a glimpse into the world of connected car data, which could come from other external sources. This is a very complex landscape and one that will certainly evolve over time. For now, we think we're really well-positioned to learn from this data, given our unique history with Snapshot and the tremendous amount of data we have compiled over time. Let's now shift our focus to efficiency. We see great potential to leverage technology and advanced analytics to increase productivity and claims handling segmentation along with accuracy. Here again, we can use a broad dataset, including connected car data, to better triage claims and identify the right subset of claims for a low to no-touch claims handling process. Let's walk through another example to help illustrate this opportunity.
Take the case of Sally. She's a mobile Snapshot customer who purchased her policy online, signing up for e-sign, paperless, and is comfortable using online self-service. She's a married renter who's also a busy working mom. She starts her policy on a Tuesday. A day after starting her policy, Sally has a minor accident after dropping her son off at daycare and driving to work. It's a rainy day. She swerved to avoid another vehicle and hit a barrier, resulting in damage to her hood, bumper, and license plate. A busy life just got a lot more complicated, something a lot of us can relate to. That same day, Sally calls us to report her claim. Since her car is drivable and not severely damaged, we offer her the convenient option of customer photo estimate, which she readily accepts.
She sends us pictures of her vehicle within two hours. By the next day, our rep has an estimate back to her. So far, so good. By eliminating the time it takes to drive to a customer to complete an estimate, we improve productivity, all while offering the customer a convenient self-service option they desire. While this actual experience was fast, in fact, at least five days faster than had we scheduled a field inspection, in the future, we see ample opportunity to expedite this even further. The challenge is that we have not yet verified coverage. The fact that she bought her policy a day prior to the accident raises a concern. Our special investigative unit rep initiates a coverage investigation, which for this particular claim takes seven days and requires additional documentation.
This is a very reasonable course of action given the concern that this claim could be from damage which occurred prior to the policy's inception. In this case, the investigation confirms coverage. We issue a payment to the customer who receives it about seven days later. Let's consider what this could look like with real-time access to data and advanced analytics. In the future, we'll be able to offer a more automated, expedited process that meets customer needs without sacrificing accuracy. Going back to Sally, she's already indicated a preference for self-service and a comfort with technology. Her claim involving minor damage is also a great candidate for automation. In that more automated process, Sally will be able to submit photos through a mobile app at the time of loss. Connected car data and other facts of loss could then be used to confirm coverage.
Recall that Sally had enrolled in the mobile Snapshot app. In her case, the trip data does in fact confirm the time and location of the loss. In other instances, connected car data or other data could be a factor in triaging the claim for additional investigation. In her case, it helped support a faster resolution. After clearing coverage, we could then digest the submitted photos and leverage artificial intelligence and machine vision to triage the estimate for either additional review, or in the case of Sally's claim involving a simple repair, create a fully automated estimate. It's important to note here that we have a very large database of audited physical damage estimates and 1 billion images. We believe this sizable history provides us with a competitive advantage as we move forward in the machine vision and estimatics space.
We're also developing the ability to provide a digital payment option, shortening that final piece of the process as well. In the case of Sally's claim, these advancements would take a two-week process and turn it into a two-day process. All in all, we see the potential to improve really across all four guiding principles, and in particular, achieve significant efficiency savings for the right subset of claims. I want to highlight here that there's a lot of discussion in the marketplace about these themes, and in particular, the use of photo estimating. Our approach here will be measured and methodical. That's how we operate. We'll be careful to understand the technology, spend the appropriate R&D time, roll out each phase in a test-and-learn approach. That way, we can ensure that we don't compromise across any of our four guiding principles.
While our pace may not be the fastest, we believe that the quality of our execution and the ability to leverage our strengths in data and analytics will result in superior outcomes. When we think about improvements in accuracy and efficiency, they can't be at the expense of work environment or customer service, our two remaining guiding principles. Let's take a look at how work environment and customer service will also benefit from the greater use of data and analytics. As you know, our culture and work environment have long been strengths of Progressive. You've heard Tricia speak many times about the importance of our core values. In claims, we talk about the pillars of pride, purpose, and connection to building engagement. People want to feel good about the work they're doing, know that they are making an impact, and feel like they're part of a broader community.
We know from Gallup research that more engaged employees exhibit higher levels of productivity. In our first year of using the Gallup survey, we were really pleased with our overall results. In fact, Gallup noted to us that many other companies work for years to achieve comparable low levels of active disengagement. We're proud of Progressive's overall results, but to see claims score above is really exciting because that's a really tough job and engagement is so critical. How can greater use of data and analytics enhance both this tenet of engagement and our level of service? In several key ways. First, by reducing unnecessary points of friction with customers. We've already talked about how data can help us better segment claims handling and the need for investigations.
That also has direct benefit to our employees and our customers by allowing us to focus outreach or requests when it's most necessary or warranted. Contributing then to more purposeful work and driving engagement. Secondly, by focusing the human touch where it really adds value versus tasks that could be automated or handled through self-service. Our reps excel at helping our customers during their time of need. You've heard Tricia highlight this in our CAT handling, where the energy and enthusiasm to get customers back on their feet is contagious and inspiring. That's true in the day-to-day handling of claims as well. Future efforts to increase self-service options or even introduce artificial intelligence-enabled chat technology will allow our reps to focus their interactions with customers in more meaningful steps across the claims handling process.
Moments where, for example, the rep can really focus on the customer needs versus repair scheduling logistics or gathering information which could be automated. Together, these two elements enable faster resolution through fewer to no customer touch points where that aligns with customer preference. At the end of the day, using data and analytics to triage the right claims for the right customers into a low to no-touch claims handling process benefits all four guiding principles, including customer service and work environment. We can use data to predict which customers would prefer this more expedited, automated process. In summary, our growth enables us to leverage fixed costs and economies of scale across the enterprise. You can now see how the virtuous cycle of data applies to claims as well. Greater use of data analytics and automation in claims handling will enable significant improvements across all four guiding principles.
We have very strong capabilities in data and analytics and a proven track record in claims handling execution. Together, these should provide ample reason to believe that we will deliver on our strategy and extend our competitive advantage. Thank you. Please take a minute while we prepare for Q&A.
Thank you. If you are a webcast participant, you may ask a question by pressing "ask a question" in the upper right corner of your browser window. If you are a telephone participant, you may press star one on your telephone. You may press pound to withdraw your question. After pressing star one, please be advised that you'll be momentarily blocked from hearing the live call. Additionally, there is a brief lag between the webcast audio and the telephone audio. If you're participating through both mediums, please mute your webcast audio while asking your question. To allow us to respond to as many participants as possible, you'll be limited to one question and one follow-up question. To the extent you have additional questions, you may place yourself back in queue by pressing star one on your telephone.
With those instructions, I will turn it over to Victor, who's going to introduce our first question from the conference call line.
Yes, ma'am. Our first question comes the line of Amit Kumar from Buckingham Research. You may begin.
Thanks, good afternoon, and thanks for the details on the webcast today. One question and one quick follow-up. Going back to the discussion in the 10Q, when you talked about the decrease in frequency, had frequency been in line with expectations? Is there a way to quantify, I guess, the margin impact versus what was reported? I mean, put another way, how much did favorable frequency benefit results relative to your expectations?
It's really tough to actually quantify specifics to frequency because so many things can happen. When we look at our overall goal for profit is to make sure that we have the high-level goal that we talk about all the time, 96%, grow as fast as you can. We build in that $0.04, at least that $0.04 of underwriting profit. Now, when you think of, take last year quarter 2017 over this year quarter, a couple things came into play. Reduction in frequency, which we don't predict because there's so many things that can happen. Our cat losses were about a point less, maybe a little bit more than a point less than last year at this time. Our expense ratio was about a half a point less than last year at this time. We had less unfavorable reserves.
There's a lot of things that go into our margins. Again, our focus really is to grow as fast as we can at a 96% as long as we take care of our customers, and if our costs are lower, then we enjoy more favorable margins. Do you want to add anything?
That's great.
Just following up on that, Tricia, you talked about, I guess, the quadrant at the start of the presentation where you are in terms of growth versus combined ratio. Do you believe you're at, I guess, the optimal growth trajectory given the profitability, or there could be some desire to maybe sacrifice the margin and grow even faster based on the market environment? Thanks.
It's a great question. Obviously, when you looked at the two by two that I shared today, we haven't reached nirvana because we're actually even higher than that in the first quarter of 2018. We're really proud of that. That's why we continue to spend more on advertising, because we want to grow as fast as we possibly can. Of course, the caveat being taking care of our customers, which Kia outlined as well. We will continue on that. If we see that the growth stalls or it's not to our expectations, and we think that we need to make changes, we'll do things. We really, again, grow as fast as we can. The margin, of course, has to be $0.04. If it's greater, that's great, because that means we've done some great things on cost, and efficiency likely. You can't tell the future.
It's not that we're standing still. Like I said, when I talked about that quadrant, there's a lot of rate getting into the system from our competition. This is a very competitive industry, so we do what we can, and I love the fact that we have had for over 40 years, this sort of true north of 96 grow as fast as you can, because it allows you just to continue that. You never want to grow at the just throw away margin. You want to make sure you have both that balance. Again, can't predict it. We are running as fast as we can, making sure, like Kia said, all eyes on quality, that accuracy, that indemnity piece is a big part of our results, and we are all over that, making sure that we continue to have great results from a loss perspective. We're excited.
We're going to continue, and we'll react if things change and things will change. I mean, things always do. We never know what's ahead of us, but that's the fun part about being in this business. I talk about this a lot with catastrophes. We don't know when they're going to come. We don't know how severe they will be. We don't know how they'll impact us. It's how we react to those, and that's what we're really proud of.
All right. Thank you. Victor, we'll take the next question from the conference call line, please.
Our next question comes from the line of Sarah DeWitt from J.P. Morgan. You may begin.
Hi. Good afternoon. Thank you for the presentation. That was very helpful. Pulling it all together, given the claims initiatives you are working on, how much do you think you can reduce the expense ratio and LAE ratio over time? Specifically, I would think quick photo claim would be pretty meaningful for the LAE, but I am not really sure how to quantify that. Any help on that would be appreciated.
We believe it is meaningful. We want to make sure that we have the right photos and the videos that do not require then a lot of supplemental movement. It is really about that balance of cost and quality or LAE and indemnity. If you have great photos and great reps that can write those photos, not being sight of car, but writing as good as if you are sight of car, then it is a win. That is why we are very slow to roll out to make sure we will take one cohort of damaged vehicles, achieve success, and then move on and on. We think it is significant. John and I were both in the roles of claims president and personal lines president when we were over 12 LAE in expense ratio. We care deeply about efficiency.
It was a lot of work to get both of those down two points. We as a team, my team and I have a goal to get to a certain percentage both on expense ratio and LAE by next year. We believe it is doable, but we also want to make sure we invest in the business and we are not short-sighted. I will share that goal with you, and things can change, we want to invest, but we would like to try to get to nine and nine expense ratio and LAE. We will not do that if we cannot service our customers, if we cannot invest in the long-term business of Progressive, and if we see any negative effect on indemnity.
Sorry, what's nine and nine and nine points on the LAE and?
Nine LAE, nine expense ratio in 2019.
That's the Non-Acquisition Expense Ratio.
Yeah. I'm sorry.
Oh, got it.
Yeah. Thank you.
Okay. Got it.
That's the one-third of the expense ratio.
Right. Okay. Thank you. Secondly, your mid-teens auto PIF growth has been very impressive, and a lot of that is obviously company specific, but I would think a portion of that is because of some of the dislocation we've seen in the industry. How long do you think that pace can persist? Are you seeing any signs of the competition getting a little hungrier after a few years of price increases?
I mean, a lot of it is the competitive environment. There's been a lot of rate inserted into the system, it's always a competitive environment. I think we really leveraged on this timeframe where we could grow because we had the right price in the system. We've always talked about being ahead of rates and having smaller bites of the apple and not have to take huge increases that cause rate shock. I'm being very competitive in the market. Having a competitive product out there has been huge, and we continue to evolve our product. Just rolled out our latest model in one state, Iowa called 8.5, and early signs that looks great as well. We continue to move. The market is always moving in terms of our competition.
A key part of our success has really been our continued improvement on the retention side, that's very important. That's always been our Holy Grail. We want to make sure when we spend this amount of money to acquire customers, that we nurture them and keep them. That means not shocking them with rate, that means helping them with their experiences, both on the CRM side and the claims side. Seeing that retention is really important. In addition to that, with the acquisition of ASI, now Progressive Home, we have a different mix of customers, we're continuing to see more preferred customers that we know are stickier.
A lot of things in play, we just have to continue, like I said earlier in the beginning of the presentation, make sure we accept that challenge, that we can't stand still, that we have to continue to put pressure on ourselves to fire on all cylinders to continue to grow, both in units, as you say.
Okay, great. Thank you for the answers.
Thanks, Sarah.
Victor, can we take the next
Our next question comes from the line of Elyse Greenspan from Wells Fargo. You may begin.
Hi, thank you. My first question is going back to the frequency conversation. What do you think drove the decline in frequency in the first quarter? I know we went through a period of a couple of years where most folks in the industry were seeing elevated frequency. You guys were obviously seeing better trends. We continued in your 10-Q pointed to unexpected favorable frequency in the first quarter. What do you think has been driving that for the industry at least to start the year?
I would say, speaking for ourselves, we've talked about it a little bit. I think it's becoming a little bit level in the industry, but we have a different mix of customers. Vehicle miles driven, we're down as we've seen that. It's becoming a little bit more level at that 1.5%. There's a couple of different things that come into play, but for us, we see it as our mix of customers and vehicle miles traveled as a couple inputs.
Okay. In terms of retention, in your Q, you highlighted some pretty strong growth metrics in your PL, both on a trailing 12-month and three-month basis. Can you just update us? How do those trends compare to your targets? You think about maybe some metrics in terms of the percentage of Robinsons in terms of your cohort and just where you think that these retention metrics, can they keep going higher, as you think about going forward from here?
We set a goal. Actually, I stood here and talked to all of you in 2015. I set a goal of increasing overall retention, every consumer segment by 50%. We talked a little bit about how we're doing really well in that plan. Again, we target very surgically different segments. We then sub-segment those targets. We'll have separate types of SAMs that we believe are going to last longer. As long as we make our 96 calendar year target each year, we're happy with that. We know that a lot of it has to do with nature and nurture. Again, all these things come into play. Having the right product that gives discounts to more preferred customers. Those preferred customers, if they're happy with our service and they can add more products, stay with us longer.
Obviously you want that to continue on that trajectory. There's a lot of things that go into that. Again, I talked about pricing and getting ahead small bites of the apple and not having to rate shock our customers. We've been doing that for a couple of years. We watch all of that closely. For us, we have internal goals, very specific with each segment. We don't share them publicly, but we feel we're doing very well and specifically in the Robinson cohort. Again, what's most exciting about that is that we have so much more to grow in that segment. As you know, not too long back, we rolled out a product called HomeQuote Explorer. People being able to go online on their mobile devices to get a home quote.
We've grown our Progressive Advantage Agency on the direct side if you want to call in to buy a home and auto in that bundle. Of course, we have our platinum agents across the country that are doing that. We're firing on all cylinders there, but there's a lot of opportunity there. That's one where we feel really positive because we know Robinsons stay longer as long as we can service them and do the right thing, and we believe we can, and we have a lot of trajectory and some room to grow in that cohort.
Thank you.
Thanks, Elyse.
Our next question comes from the line of Mike Zaremski from Credit Suisse. You may begin.
Hey, thanks for taking my question. First question pertains to the helpful presentation. Does the expense ratio differ materially for an existing customer who mostly engages with you guys using your app versus a customer who doesn't use the app?
Yeah. Without knowing the very specifics on that, clearly self-service is a much better expense on our side from that perspective, when you're talking about Non-Acquisition Expense Ratio. Again, I want to be clear on we have overall expense ratio, then acquisition, non-acquisition. Yeah, when people self-service, you're not picking up the phone and talking to them and every second at a call center costs money. Absolutely.
Okay. Got it. Are you seeing more people engaged with the app?
Yes, we're seeing more people engaged with the app and even buying. The biggest portion of our customers on the direct side are through the mobile app.
Okay. My second question was regarding Snapshot users. Does the retention rate for someone who used Snapshot differ from the non-Snapshot customers? Maybe you could do the same regarding that customer's loss ratio.
I think it depends on if they get a surcharge or not. We get the Snapshot information, we have your driving behavior, and then we give you the final rate, and some people get discounts, and the discounts vary, and some people get surcharges. Oftentimes, if you're getting a surcharge and you want a discount, you might shop a little bit. We're okay with that. That's sort of adverse selection as we call it in insurance. That's an okay thing. What was the second question?
Loss ratio.
The loss ratio curve.
Yeah, they vary. I don't have the specifics with me now. Do you have?
I would just say that we price all of our products to the same margin. The Snapshot customers are being priced to that 96 combined ratio, as are the non-Snapshot customers. As Tricia noted, some customers we find to be really good drivers, they're getting the bigger discounts. Some are not as good drivers, and we're giving them surcharges today. We have gradually been flattening that loss ratio curve, if you will, over time across the best to the least good drivers, as we like to say. That loss ratio curve is flattening. In aggregate, the Snapshot program is priced to the same 96 combined ratio as the non-Snapshot.
I mean, I guess I'll ask another way. Is Snapshot having an impact on your financials materially in any way? I know it's giving you data, but is it benefiting Progressive?
Well, we see Snapshot as an incredible rating variable. Yeah, I think that it's a really important part and has been for a long time. To be able to understand real driving needs is incredible. We're evolving. We're doing a lot of R&D that we're not using for either rating or claims, but we're learning a lot. We continue to understand people's driving behavior, people's behavior when they're on the phone, when they're on an app. Yeah, it's incredible. With our mobile device that we've evolved to, you have the option to still have the dongle in your car or a mobile device. You also have the Snapshot information at a lower cost because you don't have to pay for that dongle. Yeah, it's a big part of how we think about really understanding driving behavior.
Well, it wasn't specifically part of your question. Those customers who we give the biggest discounts to who are the best drivers stay the longest. Their retention is longer than the customers on the end of the spectrum who we are surcharging, many of whom, as Tricia noted, are shopping because we have increased their rate. To the extent they're finding a rate with a competitor that is not inclusive of that knowledge that we have with the Snapshot device, it's very possible that they are underpriced at the competitor. Net, that adverse selection allows us to continue to have a rate level that's competitive, we believe, and has an adverse effect on the competitor's loss ratio.
Thank you.
I'm actually going to take a question now from the webcast participants. This question is, appreciating the company's strong growth and great current underwriting, how do past periods or cycles of strong PIF growth and subsequent claims experience inform your current comfort level with regards to the amount of growth that is prudent and sustainable?
Yeah. We are, like I said, extraordinarily happy with our results. I would say John and I are extraordinarily paranoid. We look at data all the time to see what's happening and what's going on. I grew up in claims, so in past periods, there were times where maybe we grew and didn't quite keep our eye on that indemnity, on the accuracy. We are all over that, and I can tell you, I read every audit which is done from John Sauerland's group to Michael Sieger, our President of Claims' group, to make sure we're not off on that, especially with all the new hires. We've made a very diligent effort, one, to hire in advance of need on both claims and CRM, but also to widen our ratios.
To make sure that we don't have, I guess I should say narrow them, to have less people reporting to a new supervisor to make sure they really understand it. I feel extraordinarily bullish on where we are on all things going. Again, we are always looking to see what could go wrong, what's around the corner, how will we react, what will we do. I think having that balance of all four guiding principles in claims, because that's the majority of the money that we pay out on both LAE and indemnity, just having a surgical focus on that is critical, and it's one of the biggest things we do.
Thank you. Victor, can we take a question from the conference call line, please?
Yes, ma'am. Our next question comes from the line of Meyer Shields from KBW. You may begin.
Thanks. I want to go back to the frequency issue again, if that's all right. Tricia, when you talk about the mix of customers explaining some of the unexpected frequency, does that mean that these customers were, quote, "overcharged"?
When we have the data for a customer, again, you're going state by state, product by product, venue by venue. We have so much knowledge. That's why when we've talked often about whether this is expense ratio or the loss ratio of new customers, you don't know a lot. You know what you have at the time of application. Whether it's that state or that person. They evolve over time, and we know more about them, and we rate accordingly. When we talk about mix of customers, just in terms of frequency, we know that when we believe that we have more preferred customers, that they're going to likely have lower loss ratios, and that's what I talk about when I'm thinking of mix of customers.
Okay, that's helpful. I'm trying to think the right way to ask this too. You've observed better-than-expected frequency in the first quarter of 2018. Given that you don't project frequency, and I'm probably phrasing that poorly, is your expectation for frequency in the first quarter of 2019 now one and a half points better than it was at the beginning of this year?
Again, we don't predict it, so I can't say that. All I would say is the industry is leveling out, and we've had fairly persistent low frequency in the last year or so. Again, it's not extraordinary. We price for severity trends, and we react to frequency trends.
Just to add a little bit on that. When we say it's lower than expected, we are constantly adjusting prices at the state level, at the product level. When we do so, our pricing team is forced to make projections around loss costs. We are pricing a product today for a cost of goods sold we don't truly know yet and won't know for at least six months, and in some cases, far longer than that. You have to make assumptions or projections, if you will, around frequency and severity. It's very difficult to know. There's a lot that goes into it. It's the customer set for sure, but there are tons of environmental factors that go into those selects as well. Severity is a little easier to determine, especially on physical damage. Generally speaking, you see parts prices go up, you see labor rates go up.
Some of the injury severity numbers are a little more difficult to assess. There's medical inflation in there that is far more difficult to know for sure than physical damage, the frequency numbers are driven by a lot of environmental factors that we can't specifically determine. Those pricers at each state product combination are making those determinations. We might believe in one state, the frequency is actually on the rise. In another state, we might project that in aggregate, it will go down. In both cases, we will be wrong. I am certain of it. Our challenge is to be right enough in that combination, again, to stay below that 96 as we set those rates for the future periods.
Okay, thank you.
Victor, can we take the next caller from the conference call line, please?
Yes, ma'am. Our next question comes on the line of Marcos Holanda from Raymond James. You may begin.
Good afternoon, thanks for taking my call. Thanks for making insurance a little cooler, too, as I really do enjoy your presentations. My question is in regards to your investment income results for the quarter. They were up almost 30%. I think that's the highest I've seen, at least for the last couple of years. Can you talk about what drove that performance in the quarter?
I will let Bill Cody follow up with that, but I want to thank you for calling us cool because not normally do you get cool when you're auto insurance in Cleveland, you made my day.
Dude.
Yeah. I'm going to let Bill Cody talk. He's on the phone from Progressive Capital Management to talk about investment results and how we think about investments in terms of protecting the balance sheet of the operating company. Bill?
Sure. Thanks, Tricia. Our results from a total return basis were not particularly wonderful in the first quarter. We lost about 30 basis points in fixed income and in equities, but our yield went up a bit, which is generally a function of a couple things. One is higher absolute yields in the market. Treasury yields have gone up. We've also seen a little bit of a rise in spreads. It gives us an opportunity to put new cash to work at somewhat better levels. The other thing that's helped us is we have a relatively short duration, so a fair bit of our portfolio runs off on a regular basis, and we don't have the same marked-to-market hit that you do have if you had a longer duration.
The third thing, which is relatively significant for us too, is we've had really strong, profitable growth from the underwriting business and the new cash that comes in from our operations on a daily and weekly basis, we get to put to work at better levels. Those are the key aspects that have pushed our overall yields a bit higher.
I would say our cash from operations has been up almost 60% year-over-year. That's hugely helpful to go into the portfolio.
As we model this out for 2018 and 2019, we should expect, I wouldn't say similar amounts, but just higher level of investment income coming from those high-yielding securities and the operational cash flow?
I hope so. I hope yields continue to move higher. That's what we've been waiting for for a while.
The market doesn't like that too much.
Right. Yeah.
All right.
Yeah.
Go ahead.
I'm sorry. I was going to say, we haven't changed our strategy. We're not going to stretch for yield.
We will increase our duration modestly as time goes on and as yields rise a little bit. Our strategy will stay the same to protect the operating business.
Got it. Thank you. Tricia, perhaps just one more for you. A big theme in every call is commercial auto results. You guys are a large commercial auto writer. I know it's mainly for small businesses. Can you maybe talk about your experiences there and what are your thoughts are on results for the overall industry?
We ran into some profitability issues late 2016, took a lot of rate, which takes a little bit more time to earn in because the majority are annual policies. We also had some underwriting restrictions to make sure we got to our profitability goal. We feel good where we are now. We're investing a lot in that business, and it's a more volatile business, so we watch it more carefully, and it has higher limits. We feel positive from both a growth And especially new app growth and a profitability perspective on commercial auto. We feel like we're in a good position. Again, we sub-segment every single part of our business, whether it's for hire or tow, et cetera.
As needed, again, same thing with venues by state, by product, and we react quickly if we see a trend that causes us any concern. We're happy with the commercial auto business for sure.
Just finally, is the Smart Haul product doing a good job into Snapshot? How's adoption for that vis-a-vis the Snapshot product?
The electronic logging device. It's pretty early on. It was supposed to go into effect in January. They gave a little lenience, so really it's been in effect for sure in April. We have a lot of data because we've been doing research on it for a couple of years. More to come on that because it'll take some time to understand. We give a 3% discount for participation, and then we watch over time. You can get up to an additional 15%, so up to 18% for the Smart Haul program. We don't have a lot yet because really, in earnest, it started in April for the truckers. Much more to come on that. We'll be watching that closely, and we can let you know probably next quarter or the quarter after that.
Okay. Thank you. Just one more thing. How has the Uber experience been so far?
How has what?
The Uber experience.
Good. We've extended our relationship with them. It's obviously a learning thing on the commercial side. We took a long pilot, like we do in claims in Texas to really understand the model. They have higher limits. Clearly earlier this year, we added Arizona, Colorado, and Florida. We're excited about our relationship and continue to learn about the loss experience for those specific drivers.
Thanks for the questions. Can we take another caller from the line, please, Victor?
Our next question comes from the line of Yaron Kinar from Goldman Sachs. You may begin.
Hi, good afternoon. One question with regards to the targeting of nine and nine in the NAER in LAE. Can you help us understand how that then flows through the 96% combined ratio? I guess, what component of the combined ratio do you see growing?
First we got to get there. That's our goal, and we've set forth sort of some strategies around to get there. Once we believe it's sustainable and we can do it without affecting other things, then we bake it into the pricing like we did years ago with the four-point decrease in LAE and expense ratio. We have a ways to go. Right now, we've been investing a lot in the company, investing a lot in hiring in advance of need. We want to continue, and hiring, I should say, in technology efficiencies like our artificial intelligence, et cetera, that we think will help get us there. We don't want to get out in front of that's a goal. I'll likely not talk about that a whole lot anymore because it's something internally that we do.
If you start to see it go down, and we believe that it's going to be able to stay down at long term, we'll bake it into our pricing.
Maybe if I understand you correctly, there could well be a period over the next year or two where as you gain comfort but are not quite there yet, we could actually see an improvement in margins until you're ready to react on pricing.
Many things go into improvement in margins. I wouldn't want to speculate that. Again, we have a very specific goal to get at least $0.04 of underwriting profit on every dollar that comes in. We're working on great product, great people, having a great return for our shareholders, balancing all those things, and being as efficient as we can while not having negative second order effects on other parts of the company. Again, would need to say something that I don't know about the future, because again, anything can change. We've seen that in past with mother nature. It could be competition, could be positive, could be negative. How we react is really the key.
Fair enough. Then my second question is around Snapshot. With regards to the upfront permission from the Snapshot customer before using the data for claims, can you maybe help us understand what happens if the customer says no? Maybe I'm just really suspicious in my nature, but I would think that if I got that answer, I'd probably react in a certain way.
We have very rarely had to ask a customer. Usually, it's driven by the authorities, and I'm talking about a handful of times. We're not even there yet. We continue to learn about driving behavior and learn about claims behavior. At that time, in the future, I will be very clear on how we're going to do that, both externally and to our customers. It is so rare, literally a couple of times that I know about, where we've been asked by authorities to ask the customer for that data. It's on a very irregular basis and a very extraordinary circumstance. If it happens on a regular basis because we believe it's important for claims handling, whether it's efficiency or accuracy, that's a whole different message. That'll be something that we alert any consumer to when and if we make that change.
Just to point out, there are plenty of instances where that data helps confirm the facts of the accident and accelerates the conclusion of the claim. We shared an example that would ferret out what appears to be perhaps a fraudulent situation, but I have seen multiple instances where we've used that information actually to confirm the facts and accelerate paying the claim.
Right. That's kind of my point. I'd be more than happy to share my data with you if I knew that it would support the claim.
Yeah.
Okay. Thank you.
Yeah. Thank you. Victor, can we take another
Our next question comes from the line of Ian Gutterman from Valeo Smid. You may begin.
Hi. Thanks. I have one on claims, then a quick one on market competition. I just wanted to understand, obviously, as you're adding Robinson and if you have more preferred customers. That's a higher limit profile than you've had in the past, I assume, right? I assume there's also just a higher service expectation, right? I mean, as far as what's an acceptable claim and just sort of what does that mean for the claims organization, just handling more complex or higher limit claims that maybe you're not used to? Does that have any implications for severity?
The great thing about Progressive claims is that we literally treat every single customer the same in terms of not only the segments within Progressive, but whether you're a claimant or insured. That's sort of been always our philosophy. Our claims reps don't necessarily know if you're a Robinson or much about you other than you've been in this incident, and we want to make sure we take care of it for you. It's more complicated, as an example, if you have maybe multiple claims, say a hailstorm hits and it affects your vehicle and the roof of your home. That becomes more complicated. Mike Sieger's working with Pat McCrank at Progressive Home to make sure in those circumstances, we figure out a way to have somebody quarterback those claims and understand how important it is to have Progressive folks on the street for those claims.
Historically, we've used a lot of independent appraisers on the ASI/Progressive Home side. We are testing this year to see how important it is to really nurture and take care of those customers, we're going to know a lot more. We have a small test going on this year, more to come on that. Yeah, I think that anytime you have multiple products with a company, your expectations are higher, and your needs are higher, and we're figuring that out along the way. I think that's really important. Overall, we try to treat every single customer with respect and dignity and understanding their circumstances, because it's never a good circumstance to be in a claim.
No, of course. I just didn't know if there was a difference between someone who has 100,000 BI limit and 500,000 BI limit. Those tend to be, I think, more complex claims. I don't know if they require new skills on your part.
It requires segmentation for sure.
Okay
In probably 2009, 2010, 2011, we really started segmenting claims, right rep, right claim, right time with the right leader. Any sort of those high limit claims goes to a centralized unit. I'm sorry if I didn't understand your question.
Got it
very surgically our claims. The big claims that might have big injuries are not going to be necessarily handled in the field by an adjuster with two years experience.
Got it. Okay, just on the market, and I'm going to try to do a hypothetical and hopefully you'll get where I'm going. It's not going to be, I'm sure, exactly how things work. Let's say what we call a normal environment. Maybe a customer that you acquire from a competitor is, say, a 102 there, and they end up being a 94 for you because of your expenses and underwriting advantages and all that sort of stuff. Again, I don't know if those numbers are exactly right, but work with me, right? Where I'm trying to go is if we're in a tougher environment where I would assume people who are shopping aren't 102s anymore, maybe they're 110s or 115s just because industry results deteriorated. Have you been surprised that, I guess I would have thought in that environment it's harder to find 94s, right?
I could say you could find more guys who would be shopping that you could win, but I would think it would also be harder to get them under the bar, if you will. Have you found that at all? Have conversion rates changed or just maybe shopping's up or conversion's down or anything like that? Are more new wins sort of closer to the bar than they would've been in the past, or has that not really been an issue?
From a shopping perspective.
Am I asking the wrong question? I guess I'm obviously doing a very hypothetical way of thinking about it.
Yeah, I get where you're coming from. I may not have the answer that you perfectly want, but I think we have been in such a good position, like you said, we have what we believe is industry-leading segmentation, so we understand that. We have and continue to evolve our upfront underwriting. That's a big key to understand who's coming in and what we think their loss experience will be. That's one thing. Having the product, having the ability to spend like we are with advertising and having the new prospects coming in and conversion going up. Although you wouldn't say it hasn't been difficult, we're just in a really good position. We're getting a lot of customers that have long-term 100% prior from other companies, so we know that there's a lot of shopping going on.
Where they fall exactly in the combined ratio, again, I'll go back to we want to have both calendar year and a long-term 96 for those customers, and that's what we are targeting in the aggregate, and we've been pleased with results in being able to acquire those customers. I think it has to do with brand. I think it has to do with competitive pricing. I think it has to do with accurate claim settlements, and all those things that go under those four pillars of the strategy portion of the four cornerstones have been really all in sync. We've been excited about that. Don't you want to add anything?
I think you got it.
I won't cut you off, Ian, so ask as many questions as you want.
No, that's good. Thank you so much.
Well, it is 3:00, so thanks. Does that answer it? I think he's off.
I think he's gone.
I think he cut me off.
Yeah. Yeah, he cut you off this time. That actually brings us to the end of our call. Victor, I'm going to hand it back over to you for the closing scripts.
Yes, ma'am. That concludes The Progressive Corporation first quarter investor event. Information about a replay of the event will be available on the investor relations section of Progressive's website for the next year. You may now disconnect. Everyone have a great day.