Hi, welcome back. I'm Brett Feldman, Goldman's U.S. Telecom, Cable and Media analyst. I am pleased to welcome back to Communacopia, Dave Schaeffer. Dave Schaeffer, the Chairman and CEO of Cogent Communications. Dave, thanks for being with us this afternoon.
Brett, thanks for hosting me. Thanks for Goldman Sachs for a great venue, and most importantly, thanks for all the investors taking time out of their busy day to hear a little bit about Cogent.
All right. Well, let's jump into this. We're well into this pandemic. Hopefully, we are approaching the point at which we are out of it. Your business has been affected both positively and negatively over the course of these lockdowns. Just to get started, can you maybe just give us an update on what the demand backdrop for the services that you offer looks like, and maybe how you see things shaping up as we begin to head into the back half and into next year?
Yeah, sure, Brett. The pandemic has been both positive and negative for Cogent, the COVID pandemic for Cogent. With regard to our corporate business, which is 62% of our revenues, we have seen significant headwinds. Our corporate business is focused on selling internet access in the central business districts of major North American cities. That business has averaged an 11% organic top line growth rate for the past 16 years. Only once previously did we have a negative period of growth. That was in the great financial crisis in Q4 of 2008 and Q1 of 2009. Moving to the pandemic, we've now seen five sequential quarters of negative growth in our corporate business. Our corporate business negative growth rate did improve sequentially from Q1 - Q2, and we do expect to see our corporate business return to its normal growth rate of about 11% over the next several quarters.
I think Delta variant, however, has slowed down the return-to-office momentum and created some uncertainty in the market. There's clearly geographic differences. We had seen in the West and South, first, a significant reopening, and then with the emergence of Delta, a slowdown. I think the Northeast and Upper Great Plains have been slower to reopen, and Canada remains primarily closed at this point in time, with only moderate reopenings. Pivoting over now to our NetCentric business, which is 38% of our revenues, we've actually been a net beneficiary of the pandemic. We actually had the very best growth in the company's history in Q2 at 30.5% organically year-over-year, while about 55% of that revenue was outside of the U.S. Even if we adjust for foreign exchange, the business grew at 23.5% year-over-year, much better than the nine percent average growth rate.
We do expect to see our NetCentric business continuing to grow at an outpaced rate, although we do expect to see a reversion back to that more normalized nine percent growth in our NetCentric business as people return to work and streaming growth returns to a more normalized rate.
Going back to corporate, and you mentioned how Delta is creating some variability. Could you give us a little more insight into what that means? For example, when you're talking to corporate customers or potential corporate customers, what are they saying about their return to office? Ultimately, where is your sales funnel right now relative to what you might have expected based on the trajectory we had thought we were on?
Yeah. We saw a peak in corporate turnover churn in November of 2020. Subsequent to that, we've seen sequential improvement in turnover from existing customers each month, and that trend continues. Our customer base tends to be economically isolated from the downdraft of the pandemic. We serve only Class A skyscrapers in the central business districts of major cities, and as such, the landlords do a pretty good job of vetting those tenants. We've actually seen our DSOs improve throughout the pandemic and our bad debt expenses percentage of revenue decline to the lowest in the company's history last quarter. In terms of new sales, our sale starts with a on-net internet connection dedicated service in that skyscraper. We have seen that business remain relatively consistent, and we have seen a uptick in companies taking one gigabit as opposed to 100 megabit connections, actually increasing ARPUs.
From that initial sale, the corporate relationship expands in two dimensions. One, we sell internet access to that same company at other locations. Those locations could be in other major metropolitan markets, or they could be branch offices in the same market. What we saw was a significant reduction in those branch offices in the same market as the primary service. We also sell the customer a VPN alternative to their MPLS private network. That technology is based on either VPLS or SD-WAN. Approximately half of our customers would historically take that service. At the beginning of the pandemic, we saw the VPN sales almost go to 0 as companies were uncertain about their secondary locations and their decisions to replace MPLS. Over the past six months, we've seen a re-acceleration in both secondary office DIA proposals and proposals to replace MPLS with a IP-based VPN. That trend is continuing.
However, I think in August, we did see a lot of companies that had planned to return to office in September, push those plans out over the next several months. As a result of that, our sales funnels remain very full. We continue to have lots of discussions. Our pace of new order creation is increasing, but we are seeing customers push out their ultimate return to office. I think the recovery in the corporate business is going to take several more quarters.
When you say recovery, does that mean a return to sequential growth as an inflection, or a return to annual growth that's aligned with your long-term targets?
I think it means sequential growth of about 2.5% sequentially. If you look at our corporate business, it had been amazingly consistent over the past 15 years, other than that brief downturn in the financial crisis. In our corporate footprint, we are only about 25% penetrated by number of tenants in the building. What we are seeing is an increased need for bandwidth from the remaining tenants, either to support work from home, support SaaS applications, or support any kind of cloud-based computing. As a result of that, we think that we will continue to see about 2.5% sequential growth in our corporate business as businesses return to normal.
Any prediction on what that timeframe might be, where you can get back to that trajectory?
If you had asked me that question six months ago, I would have said after Labor Day. I think now, based on the uptick in case volumes and the uncertainty around Delta variant, I would say probably still a couple of quarters out, so probably sometime in 2022.
You mentioned that you're only about 25% penetrated in these buildings that you serve, therefore 75% of the tenants aren't your customer. I believe in the past, your best opportunity to win over into that under-penetrated market was when there was turnover in the building. Someone's moving in, they need to make a decision. If you can position yourself at that point of the decision, you had a pretty good win rate. What are you seeing right now in terms of the turnover in those buildings? Is it still stalled or is it starting to look like a more normal environment?
In normal times, approximately seven percent of Class A office space turns over per year. Ironically, in the pandemic, we've seen that turnover rate go down. Companies are uncertain about their ultimate office configuration. Because these buildings tend to be the most desirable in a given market, the occupancy rates remain high, in the mid to low 90 percentile. What we have seen is landlords increase concessions to win over new tenants. That's either more free rent or lower rents. We also see two different impacts on corporate customers planning for their ultimate office configuration. Increasing the amount of square footage they need would be the need to move from open floor plans to doored offices as a result of the pandemic.
Offsetting that is the desire to have some employees continue to work in a hybrid manner, and therefore not be at 100% occupancy five days a week, allowing for some degree of hoteling. I think it's too early to tell, but it does appear that the average tenant's total square footage may be declining slightly. That is a positive for Cogent because the building size remains fixed, and if each tenant is taking slightly less space. That means there's a potential for an increase in the number of tenants per building. The average building we serve has 51 tenants, roughly 550,000 sq ft, and is 41 stories tall. We have just under 1 billion sq ft on net in North America, comprising about 1,800 large multi-tenant buildings.
You mentioned that one of the reasons why some of your tenants need greater bandwidth is because of the increase in working from home, which is someone who's not that familiar with this business might be surprised by. I think what you've explained is that they need to essentially remote into their corporate headquarters in order to gain access to their work applications. That's the way, and maybe I'm not explaining it as elegantly as I should be. A question we've gotten is that if we are in a protracted period where maybe it's more permanent, that you will have a distributed workforce, not distributed to branch offices, but distributed to a work from home environment, could that architecture change such that they don't actually need the bandwidth coming into their headquarters but perhaps a cloud facility somewhere else?
A very large company like Goldman Sachs will typically have its work from home VPN concentration and security placed in a data center. For most mid-sized businesses, they perform that function through their company's firewall. That firewall functions both as a VPN concentrator and a security device. If I am an employee working from home on my residential broadband connection, I launch an ad hoc VPN through my client software, establish a session with that firewall, and all of my traffic is going from my residence to my corporate firewall, going through the firewall, being authenticated, either grabbing an application on site or being returned out of the firewall back to a cloud application. For most mid-size companies, it's more cost effective to do that at your primary location, particularly with the affordability of a one gigabit connection.
One of the things that gives Cogent a distinct advantage in this environment is the fact that our bandwidth is non-oversubscribed and non-blocked, and our bandwidth is fully symmetric. Very different than most other broadband providers who have asymmetric services and oftentimes oversubscribe those services and may even charge an additional premium if the customer actually uses more of the bandwidth at any given point in time. Our service is completely unmetered for our corporate customers. I think as businesses work, evolve to a more hybrid model and certain applications are cloud-based, that actually bodes well as an additional catalyst for those companies that have not migrated to Cogent and our footprint to choose to migrate to us because of the superiority of our product.
One of the key differentiators in our corporate market is the fact that we install faster, we are three times more reliable once installed, and deliver somewhere between 30 and 60 x the throughput for price parity with our competitors. That is very different than the dynamic we have in our NetCentric business, where our subscriber acquisition is aided by lower price. In our corporate business, we price a parity to the market but offer a vastly superior product. I think the hybrid work-at-home environment has become yet another catalyst for those primary corporate locations to switch to Cogent. As a result of that, I think there is a fairly deep pool of branch offices that have not yet switched from companies that have migrated to Cogent.
As these businesses return to the office, they will then look to switch those branch offices to us as well.
Speaking of competition, there's been a significant uplift in the investment in fiber. There's been a lot more private capital looking to support investments in fiber. Are you seeing that the buildings that you serve are having more fiber options coming in with companies that have adopted a similar model?
We build our network by purchasing fiber from others. We have 59,000 inner city route miles of fiber, roughly 17,000 route miles and 39,000 fiber miles of metropolitan fiber purchased from 276 different suppliers globally. In the typical building we serve, there is usually two competitors, the incumbent and one alternate provider. Some buildings may have three competitors, some may have none, but on average, we have two competitors. What differentiates Cogent in those buildings is three factors. First, we have gone in and pre-wired the building. We have built a riser infrastructure that allows us to turn up a customer at a low cost and in 9 business days, much quicker than our competitors. We have a breakout box every third floor, and we pulled fiber all the way up the riser.
Typically, it would cost about $1,000 a floor to home run fiber to the basement, which makes the installation for our competitors prohibitively expensive. The second advantage we have is that we only sell internet access and VPNs. We're not trying to sell a bundle or suite of services, which makes the decision to switch to Cogent easier and allows us to specialize in giving the customer the services that they want, rather than trying to force them to buy services they don't want. The key trend in business communications has been the delamination of the application layer from the network, and the customer's desire to purchase bandwidth independent of the application.
The third advantage that we have, which is truly unique, and our competitors do not architect their networks this way, is every location on our network is part of a ring, giving us physical diversity, two paths from the customer premise all the way back to our central office. Very different than the hub-and-spoke architecture of a telco architected network, or the branch and tree architecture a hybrid fiber coax network would deploy. That results in us having about 3 x higher reliability than our competitors. In our footprint, virtually all of our customers would acknowledge that our service is superior to any of their alternatives. In fact, we win between 40% and 50% of the proposals that we issue in our footprint, because of this superiority.
During the onset of the pandemic, we saw that your sales force productivity dipped, which is not surprising. Can you give us an update on where you are in terms of being at the productivity levels you target? Bigger picture, what is your outlook for growth in the sales force from here?
Yeah. First of all, let's talk about the marketplace. On the corporate side, we saw a reduction in demand that definitely had a negative impact on sales force productivity. Offsetting that was the fact that roughly about 72% of our reps are corporate, 28% of our reps are NetCentric, and those NetCentric reps saw an uplift in demand. The second thing is, when the pandemic hit in March of 2020, we quickly pivoted to a work from home environment. Cogent has 70 sales teams, approximately 600 quota-bearing reps that sit in 35 offices around the world. All of those individuals immediately went to work from home. We also have a model that is heavily dependent on direct sales, with 99% of our revenues coming from our direct effort and only one percent from agents and channel partners. We rely heavily on outbound telesales to drive revenue growth.
Anyone who's ever done telesales knows how difficult that job is. As a result, in normal times, we churn about five percent of our sales force. The second challenge we had was to be able to recruit and onboard and train reps, never setting foot in an office. In fact, almost 400 employees of our 1,100 employees at Cogent prior to September, had never been in a Cogent office. We were onboarding people remotely. We revamped our training tools, but still were not as effective as we could have been in the office. The third thing was maybe the most challenging of all, which is how do you terminate an unproductive worker remotely? We struggled with that, measuring the market conditions and our inability to train as effectively, and we began to put in place a program last fall to manage out those underproductive employees. That continues.
We have seen our sales force turnover elevate to over seven percent a month, up from the long-term average of about five percent per month. We have seen that number start to come down. We went to a mandatory return to office, we did see some reps who had never been in an office not adapt well to that office environment. We continue to go out and recruit and add reps. We plan to grow our aggregate sales force on average between 7%-10% per year. We have grown at about 7.5% per year in number of quota-bearing reps over the past five years. We think we'll probably be slightly below that growth rate this year, we would continue to grow the sales force. We also think that as customers on corporate network designs are re-engaging with us, we're going to see a re-acceleration in corporate rep productivity.
NetCentric rep productivity has remained high or even accelerated through the pandemic. As a result of that, we have seen a couple of quarters of sequential improvement in aggregate sales force productivity, and we expect that to continue throughout the year and return to our historic averages of a little better than five installed orders per rep per month.
Let's talk about the NetCentric business. As you noted earlier, revenue growth has been unusually strong, growing over 30% in the most recent quarter, over 20% on an FX adjusted basis. Can you just remind us what has been the principal tailwind that you experienced over the course of the pandemic? What's a better way of thinking about a normalized rate of growth from here? I'll just add a third part to the question, has Delta changed that at all? Are you starting to see any of the early lockdown tailwinds actually re-emerge in the business?
At the highest level, our growth rate in NetCentric is driven by traffic growth and price declines. If we look at the past 20 years of Cogent Communications' history, prices have declined at about 23% per year. Volumes for us have grown significantly faster than the market at about 35%, and market average growth is about 22%, 23%. Our revenue growth is actually a little more subtle than that, in that it includes some additional factors. one, what is the size of the customer? If all the growth is coming from the biggest customers, price per megabit goes down more rapidly. If our growth is coming from a broader base of customers, the effective rate of price decline is more moderate. The second impact is on where these customers are.
The internet was primarily a U.S. phenomenon 20 years ago, with 85% of global traffic being in the U.S. Today, only about 33% of traffic is in the U.S., and the U.S. as a percentage of global traffic is declining. More of our NetCentric growth is coming internationally. We serve 49 countries, and we have a footprint in over 1,300 carrier-neutral data centers. That's more than any other provider. That has bode well for Cogent's growth as the market has become more internationalized. The final factor that's impacted our growth is how we get paid. The internet is not a singular network, but rather a mesh of networks that exchange traffic with one another. While there are 65,000 autonomous system numbers or unique network identifiers that have been issued, about 9,000 of those really account for the entire traffic volume of the internet.
Cogent is directly connected to more autonomous system numbers than any other provider, with over 7,500 networks directly connected to us. When we connect to another network, there are two mechanisms of connection. One is through the sale of transit. Roughly 70% of our traffic today is two-sided, meaning we're getting paid by one transit customer sending traffic to another transit customer. In the other 30% of the instances, we're only getting paid on one side. That is, a customer, a transit customer, sending traffic to a settlement-free peer where we generate no revenue. We do not buy any transit, so these are the only two mechanisms of exchange. That percentage of two-sided traffic has increased over the past two years from roughly 50% of traffic to 70% of traffic, in large part because of the pandemic and the increase in streaming.
Streaming video is the key application that is driving unit volume growth. Cogent is the leading provider of transit to the streaming industry. As such, we have over 7,500 access networks also buying their upstream from us. When we put these multiple factors together, we've actually seen our revenue growth outpace our normalized traffic growth with price declines at trend line. Now to the final part of your question, Brett, which is: What has the pandemic done to traffic and revenue growth? The first thing we saw was a broadening of the peak window of traffic. With most traffic coming from streaming video, pre-pandemic, we saw peak usage periods between 7:00 P.M. and 10:00 P.M. anywhere we operated globally. That peak viewership window has actually widened to 3:00 P.M. till 12:00 A.M.
With people starting to return to work, we've seen that window begin to narrow somewhat, and then with Delta, in those regions where we have seen an uptick in cases and a return to people staying at home, we've seen that viewership window remain wide. I think as businesses reopen and employees return to the office, we'll revert back to a more normal viewership pattern. I think the percentage of traffic that is streamed versus linear will continue to increase. Going into the pandemic, about 18% of traffic was streamed and 82% was linear. Today, over 40% of all traffic or video consumption is streaming, and less than 60% of it's linear, and I think the pandemic permanently shifted those curves. Streaming, as it exists today, probably can drive internet growth rate in the mid-20s for the next 8 - 10 years.
I think it will be enhanced by the addition of higher resolution and augmented in virtual realities and more interactive video as opposed to more one-way video.
How much risk is there to your business from media consolidation? It's not just been that there's been an uplift in traffic because people are streaming more. They're streaming more different services. There's been a sampling that's been going on, so you've been getting a broadening of your base, which is very favorable to your pricing dynamic. There's a big debate among investors, well, what happens as we emerge from a pandemic and everyone realizes they probably oversubscribed to media, and they either start to get rid of services they haven't consumed as heavily as they initially did, or some of the services combine into companies. Is that a material headwind to your pricing?
Cable is a victim of its own success. Paid television inflation rates have been about double the underlying inflation rate for the past 30 years. As a result of that, many people got priced out of the market. What they did is they reverted to their broadband connection and a single streaming application. As we have more streaming applications becoming available, we now see many consumers taking two or three different streaming packages. Even with two or three concurrent packages, the consumer has a much lower total cost than the linear television alternative. While there will be some consolidation, there's also fragmentation in the streaming market as we continue to see more specialized content being produced for long-tail applications. I think it's premature to believe people will get down to one, two, or even three streaming providers on top of their broadband connections.
The pricing for these different services has remained low, in part because of the competition. I think the proliferation of choice, the ability to pick content that fits your particular usage patterns and be able to time and location shift it, say that we'll continue to see streaming increase, and I don't think that consumers want to be tied to any one provider. As a result of multiple providers, we are a beneficiary because with a more balkanized market, we do not offer the same level of price discounting as we would to, say, one very large provider. That's been a tailwind to our NetCentric revenue growth, and we don't see a consolidation down to one or two players. At least that's not the way traffic has been trending for the past year and a half.
I want to talk a little bit about your balance sheet and capital allocation in just the few minutes that we have left. Your net leverage at the end of the most recent quarter was 3.4 x net debt EBITDA. That is at the high end of your target range of 2.5-3.5 x. You said the board could potentially be flexible with that target range in order to ensure you can sustain your capital returns. Can you give us an update here? What is the outlook for where your leverage is likely to trend? If you were going to revisit what level of leverage you're comfortable with, how do you think about what the puts and takes are of that decision?
We're operating in a low interest rate environment. We have further lowered our interest rate by swapping our fixed rate debt to floating rate. Both through the notional nominal rate being lower on our new debt that we refinanced last year and the swap, we have significant cash savings. Independent of those cash savings, we've been growing our cash flow at about 15% per year for the past decade. With that growth, we've been able to grow our dividend sequentially for 37 consecutive quarters, and the dividend growth rate has pretty much mirrored the growth in free cash flow, roughly about 15%. We also have natural de-levering. Different than most wireline telecom companies, we have top-line growth and margin expansion. We've averaged about 10% organic growth and about 200 basis points of margin expansion.
It is true during the pandemic, our growth rate has been about half of its normalized growth rate, but our margin expansion has still been about 130 basis points per year, allowing us to grow free cash flow. In terms of leverage, we have excess cash on the balance sheet. We have a plan to methodically return that capital to our equity holders through a consistent growth in the dividend, coupled with episodic buybacks in periods of market volatility. We have a range of net leverage of between two and a half to three and a half times EBITDA. We are at the higher end of that range, but I think with the excess cash that we have from our refinancing, we'll see that leverage rate begin to drift down even with the sequential growth in our dividend continuing for the foreseeable future.
It is true the board evaluates this each quarter. Allocation of capital is probably the most important question our board focuses on each and every quarter.
Hi, Dave. Well, I'd love to go on, but we are out of time. Thank you so much for being here with us, and we look forward to seeing you in person at Communacopia next year.
Sounds great, buddy. Thanks a lot. Stay safe, everyone.
Thank you.
Bye-bye.