A real delight to close out the conference with ZoomInfo, and thank you for your time and attention, day four of this, absolutely enjoyable, to me at least, I hope it has been for you guys as well, the Goldman Sachs Communacopia and Technology 2024 conference. That, I think, will be the last time we'll be saying that. Let me say that again. Welcome again to the Goldman Sachs Communacopia and Technology 2024 conference, the last session. Bring it on, Henry, and your new friend, Graham. Welcome, welcome to the conference.
Thank you. Nice to be here.
Thank you for coming. So, Henry, we talked about it last year, the company's been through a lot of change. Where do you see the company now, and where do you see the company four to five years from now?
Yeah, I think, you know, we've been working really hard to transform our growth algorithm, really, and over the last two years, three years, our end market has been under a tremendous amount of pressure. Particularly, you've seen layoffs in on go-to-market teams, sales teams. You've seen a lot of scrutiny over contracts that you hadn't seen in years before. And we've been really focused on building a great product and moving our business upmarket. And so, what you saw in the last quarter, from an operational perspective, was our ability to start showing that we moved. In new business, we had our best mid-market in the enterprise quarter on record, and so we're bringing in better customers upfront. Our 100K cohort grew again for the first time since Q4 of 2022.
We stabilized and improved our net retention numbers, and we've started to move the business more upmarket into mid-market and enterprise, where our net retention is better, where our growth rates are stronger, where our customers are buying more of our DaaS and OperationsOS products, and then we released Copilot, our AI-driven platform, and we've begun migrating our customers over to that platform. We now have, at the end of the quarter, we had $18 million of ACV on Copilot, and we think that the future of go-to-market is built upon using AI to rank, prioritize the accounts that you're focused on, uses signals that exist in third-party data, but marries that to first-party data, and then chooses and writes the content for the right channel for you to go after.
Every customer I talk to is telling us, "We wanna bring our first-party data together with third-party data, and we wanna use that to orchestrate how we go to market," and we're far ahead of anybody in the space in giving them the ability to do that.
Great. So if these things play out the way you're anticipating, what could the company look like, four to five? I know it's a hard thing to visualize, but you, as a founder and entrepreneur, have goals and ambitions for the company. So where is this journey taking you? If all these initiatives in the short term, which look like they're gonna be successful, they do play out at scale.
Yeah, look, I think the first thing that we're focused on, at least in the short term, is really getting our arms around net retention.
Mm-hmm.
Continuing to drive the improvement that we saw in net retention, and then keeping our new business engine and new business motion going. Then, I think you can really see your way to high single digit, low double-digit growth.
Mm.
In the near term.
Mm.
Then we will grow from there. We'll continue to be, you know, a very efficient business. We'll continue to buy back shares of our stock. At the end of the quarter, we had $400 million left in our authorization to buy back shares of the company. I think you'll see us continue to do that, and we committed to $1 of leverage free cash flow in 2024.
Mm.
and that we'll continue to grow that $1 of leverage-free cash flow in 2025.
Mm. $1 per share?
$1 per share.
Okay.
Yeah.
Just making sure.
Yeah, not just $1.
Yeah, but I appreciate-
Just trying to set expectations.
No, I understand. You know, but I appreciate the fact that, you know, I was trying to be like an analyst talking to a long-term vision, and you're very grounded in what you wanted to get done in the short term, so I appreciate that.
I mean, I can pontificate for you-
Yeah, I know.
about the future. I'm happy to do that.
No, you're being grounded.
But I think right now we're being really focused on-
That's good, yeah.
Getting the company back to growth [crosstalk].
Yeah, yeah.
doing that efficiently, buying back shares of the stock.
Yeah
and moving the business upmarket.
Yeah, and one of the, I wouldn't say a challenge, but one of the things that are delicious. I have a list of these questions, but then, depending upon how the answer goes, I have to kind of change the tone of the question, right? So, yesterday we had fortunately, I did not do this interview, but David Solomon, our CEO, interviewed Marc, and he asked him. He had a lot of questions written up, but we got only time for one question. But one question was: What are customers' priorities, Marc, for calendar 2025? And it was a generative AI discussion, was three case studies that took 35 minutes, and it was over.
Yeah.
So, right? I mean, it was just one question.
I mean, I can tell you what our customers' priorities are [crosstalk].
Yes
For twenty-five or what I'm hearing [crosstalk]
Yes
From our customers.
Yes, yes.
It's very interesting because I think when-
I hope it's not 25 minutes again.
It won't be 25 minutes, it'll be three.
Okay.
What we heard.
It's your show, by the way. You can talk for 25 minutes, too. It makes my job easier, incredibly easy, right?
Our customers. I think what happened inside of corporations is the C-suite came down across the organization, and they said: Hey, we wanna drive generative AI initiatives throughout the company to drive productivity, efficiency, and innovation in our businesses. And the first place that made the most sense was customer support, and customer support is. The reason why customer support is such a good place for generative AI to make an impact is that all you need is first-party data.
Mm-hmm.
You need your knowledge-based data, your product documentation data, your support ticket data. It's all data that you own, that's first party, that goes in and then can start answering questions right away.
Mm-hmm.
When that worked, what people said was great. Now do it in go-to-market.
Mm-hmm.
Do it in sales, do it with our account executives, our account managers. We want to see that same level of productivity driven there. And the problem with going from support, which is all first-party data related, to building AI for go-to-market, is AI for go-to-market necessarily requires third-party data.
Mm-hmm.
You can't just build it off of the data that exists in your CRM system.
Mm.
Largely, your customer data, your prospect data, in your CRM system is not complete, it's not enriched, it's not usable.
Mm.
for AI that prioritizes accounts
Mm.
AI that identifies when accounts are in market, AI that can write the messaging for those accounts because it understands what the company is, how big it is, what industries it's in.
Mm.
what jobs it's hiring for, what the recent press releases said.
Mm.
None of that data exists organically in my first-party data.
Mm-hmm.
So you have to take your first-party data, you have to master it so that it's clean and complete and enriched, and you have to marry it to third-party data that gives you a broader view of the world-
Mm.
In order for you to drive AI in the go-to-market.
Mm.
And so what our customers are telling us is, "Yes, I want to use AI in go-to-market. I need to fix my data in CRM in order to be able to do that, and then I need your third-party data to marry to that, to help me go-to-market, using AI to be more productive and more efficient." And that's what we've built with Copilot and where we're continuing to invest dollars.
Mm-hmm. Gilly's looking at me saying, "Dude, enough already." Let me jump in here.
No, I think that this is just a great transition point around data as a service, which is now 13% of your ACV. Maybe talking a little bit about that, what the service is, what position ZoomInfo to win, and kind of where you're seeing that traction, across different customer cohorts, with data as a service.
Yep. So data as a service is our solution that can plug into a customer's CRM or data warehouse or anywhere data on their customers and prospects live. It uses advanced matching and merging technology to take that data, cleanse it, master it, enrich it wherever it sits, and then make it ready for use for use by the sales team, by a business or analytics BI and analytics team, by the sales force, or to plug into an AI engine. And so our data as a service. We cover 100 million companies, 300 million business professionals across the globe. Almost all of our enterprise customers use some version of our DaaS and OperationsOS product. It's the fastest-growing piece of our business.
There's still continues to be a tremendous opportunity there for DaaS.
Yeah, and, it's super helpful color, and maybe we'll come back to that, but I want to bring Graham into the conversation a bit. So thanks so much for joining us, and, I want congrats on the new role, first of all. You know, what are you seeing? Like, what are you utilizing, I guess, in the transition to bring you into the, like, an interim CFO role, and what are you seeing from that, where you can maybe have the biggest impact today?
Sure, uh-
First of all, Graham, this is your first public investor conference that you're presenting at?
Done one other one.
Oh, one before already?
Yeah.
I'm sorry to hear that.
Uh-
I thought we'd be the first ones. So welcome to your first.
We can say we are.
This is his favorite one.
Yeah, this is my favorite one.
Okay, we love that. Okay, welcome.
This is his favorite.
Thank you. Thank you for the question. Yeah, you know, first and foremost, we're focused on continuing to improve our forecasting and reporting execution, and then just making sure that we are positioning the finance org within ZoomInfo to continue to be a strategic partner to our go-to-market and our R&D teams, to continue that positive operational momentum we had coming out of Q2. So we saw enterprise retention improve sequentially for the first time since Q4 of 2021. We saw mid-market retention, which you know, where we've seen kind of the greatest decline, stabilize for the first time since Q2 or Q4 of 2021. And I want our finance team to continue supporting those go-to-market and R&D teams to continue that momentum into the back half of the year.
No, that's super helpful. So it's good to know, like, where your focus really is on, and when you're thinking about the different components of growth, right, and specifically NAR, how should we think about the mix of that and how that looks going forward, either in the back half of the year, but also where you want the business to go over the medium term?
Sure, yeah. So we'll start with NRR. We saw NRR stabilize at 85% in Q2. That sequential stabilization was the first quarter of not seeing decline since 2021. And as we start to think about the path back to mid-single-digit growth, we start with line of sight of taking that 85% retention up back into the lower 90s. So if we do that, we continue to invest in new sales where we are, we get to that mid-single-digit ARR growth, which will then translate, you know, a quarter or two down the road into to revenue growth of that level. And then beyond that, as we cross low to mid-digit 90s retention, get up closer to 100%, that's when we start to see that path back to double-digit growth.
When you think about the components to drive that 90%, is it mostly new business, which is an area that you've highlighted more recently? Is it expansion? How are you thinking about maybe even pricing? What are those components?
With retention, we think about it in two components. It's the renewal rate of our existing customers and then our ability to upsell them. And what's great is that we now have Copilot to help on both fronts. So the stabilization we saw in retention also, we also saw in renewal rates. So some of our, you know, mid-market, our software and tech-heavy customers that did feel down or that we did see downsell pressure from over the past couple of years, we're starting to see that abate. And then we now have Copilot, where we have this compelling upsell offering that we didn't really have for the past year and a half. So we have two drivers to, you know, that we've seen begin to stabilize retention, and that we expect to improve retention going forward.
So if I could bring Henry back in on the Copilot topic. Henry, what is the differentiation of the ZoomInfo Copilot? 'Cause everybody has a Copilot. There are more Copilots than pilots these days. So what is your
I think-
And more agents than real people.
Yeah, I think it goes back to what we were talking about around third-party data being a necessary component into a go-to-market Copilot. And so what you might get from a Microsoft Copilot or a Salesforce Copilot is gonna be based on the first-party data that exists in your system. And that data, for go-to-market purposes, it hasn't been good enough for the last decade, and just for your non-AI activities, and it can't be good enough for your AI-related activities going forward. That. And I'll give you an example. Today, we take your first-party data, we marry it to ZoomInfo, third-party company and contact data. We help you master that, and then what we brought in around that data is intent data that tells you what companies are researching, what sort of products and services on the web, review site data.
So we can tell you if one of your potential customers or existing customers is researching a competitor, and looking up reviews for competitors or looking up your solution. We bring in earnings call transcripts. And so one of the interesting things about earnings call transcripts is, if on an earnings call, I tell you that we're hiring aggressively in sales for international, in our international markets, because we're seeing really great traction there, that might be important to a ZoomInfo-like company or a sales engagement company or a CRM company. It might not be important at all to a cybersecurity company.
Yes.
And so what we've built is a customer contact database for every one of our customers, so that when we see that earnings call transcript come in, we're able to parse out that this is important to you, but not important to you, and it's important to you, but not important to you, and so we can aggregate signal on a customer-by-customer level. We brought in podcast transcripts. We brought in, expert call reviews, job postings, just a plethora of data that we're able to take now and drive signal-based motions for our customers.
And so when we take a look at what our customers have closed won over the last six months, we're able to look at those closed-won opportunities and say, "These were the signals that were happening within this customer set before you closed them, and then here are the next thousand customers that are displaying those same signals.
Mm-hmm.
Those signals, they don't exist inside of your CRM or your first-party data. They are. It's necessary for you to go into third-party data to have a Copilot for go-to-market.
That's got to be a huge boost of productivity of a salesperson, 'cause we've been hearing this as a theme-
Yep
W ith a few other companies. I mean, we've had the good fortune of interviewing, I don't know, twenty-five plus companies in the last four days or so, mostly CEOs, and we hear this as a theme, so I just wonder if sales productivity is gonna get a big boost because they're able to do their research much more efficiently-
Yep
Q uickly, more knowledgeable in front of the client, right?
I mean, it's effectively like giving every account executive-
Yeah
T heir own SDR-
Yeah
W ho goes through the market, identifies which companies are in market-
Yeah
I dentifies who to contact them, prioritizes those accounts, writes the copy for the account executive to send, and sets that up for them every single day when they come into work.
We didn't see this a year ago. I mean, this is all happening.
Yep, that's right.
I n the past few quarters.
Yeah.
It's truly amazing, right? And $18 million in net new ACV just for that product that we didn't even imagine two years ago.
Yeah, and we launched that in the last week of May [crosstalk].
Yeah
I n the quarter, and we don't think that's an anomalous
Yeah
T rend.
Yeah. Yes, David, please jump in, yeah.
Sure, so on that point, should the $18 million ACV gains over the other SaaS for that product going forward, or was that something about the mix for the customers that are pretty good, and [crosstalk].
I'll just paraphrase the question: Was it a one-time anomalous boost in new business because you just launched a product, or do you think this could be representative of continued momentum in the size of deals you booked?
Yeah, it's sustainable. We expect it to continue to grow, and that eighteen million comes from a mix of, you know, off-cycle upsells to our existing customer base, migration upon renewal, and actually selling to new customers. So as we. You know, part of it is a function in time. As we move through more renewals, as we add more new business there, we expect that number to continue to grow.
i.e., in a quarter?
Yeah, I mean, partially it's we released it, you know, near the end of the quarter, and then partially just we, as more customers come into the phasing to migrate there, and as we move more of our new business onto it, yes.
It's a good sign that it's the last day of the conference and the last presentation that our clients are starting to engage with us, finally, after three days of shyness. So that's great, but absolutely, well, let's make it interactive. Anybody else has a question? You don't need to take the mic. You just shout out the question. I will say, "It's Kash Rangan. This is me. I have that question," so you won't be recognized on.
Graham, maybe just to follow up on that, you mentioned off-cycle upsells. I think that that's, like, a really encouraging sign. Is that sort of representative of the go-to-market changes that you are doing? Or is that from the pull from customers and the interest that you're seeing from them?
It's both. You know, we've arranged our go-to-market teams behind Copilot in a way that, you know, optimizes our ability to sell that into our customer base, and it's the demand from our customer base once they see how compelling Copilot is.
When you think about the new customers that are coming on, I think you've mentioned 90% of those are coming on with Copilot in the quarter, last quarter. What is the customer profile of those, and how has that changed versus your existing cohort or prior quarters that we've seen?
Yeah, you know, I think that across the board for Copilot, the early adopters are somewhere kind of in that mid-market tech segment and vertical, and those are, you know, usually our best fit for Copilot, where it is right now. As we add functionality and features on our roadmap, that will expand across our whole customer base.
Are you seeing a change in the tech vertical? I know that that's been, you know, an area of pressure for, presumably over the last year, so wondering what you're seeing there.
Yeah. So software technology was about 40% of our total business at its peak. It was down to 33% at the end of last year, so that was the largest driver from a vertical perspective of our deceleration of growth. And in Q2, similar to mid-market, which is very tech software heavy, we actually saw that software vertical, the retention there improved sequentially for the first time since, again, Q4 of 2021. So we're seeing those positive stabilization and improvement trends across software and then mid-market and enterprise.
What are the investments that you're seeing that are required or on an ongoing basis around Copilot? Are you expecting that to have any implications to the growth profile? How are you thinking about that balance, both within, obviously, Copilot, but also the broader initiatives that you have for the company?
Yeah, I'd say the investments behind Copilot, which we've been, you know, very focused on, and the majority of our investments have been behind Copilot over the past year plus, those investments are already in our run rate, and, you know, there's some level of infrastructure, AI consumption costs that will continue, but we've already accounted for those, kind of, in our go-forward plan, and when we did align to start building Copilot, much of that investment came from reallocating resources from across the business that were already on board.
Andrew, he's doing well. I mean, this is, like, your job interview. I mean, being evaluated.
We're very lucky to have him.
That's, that's great. I wanna bring us back to GenAI. So very good perspectives on, on Copilot. How do you see this, I mean, if you were to visualize or envision how it all plays out, what does the, the structure of the CRM industry look like? Or is everybody gonna have to have a Copilot, a salesperson is gonna have a Copilot? Does the salesperson have multiple Copilots because one thing does the ZoomInfo, value add, and then something else? Are we gonna be flooded with so many Copilots that... and agents, where do you see this all going?
Yeah, look, I think, of course, every salesperson is gonna have access to their own Copilot, and that Copilot should automate a lot of the tasks that a sales rep is currently doing and then really drive their productivity. I think it would be not ideal if they had to have five different Copilots to automate the tasks of a sales rep. And I think what we're doing with Copilot today is designed to take, like, the heaviest, most difficult part of a sales rep's day-to-day off of their rote manual task list, and then continue to move through the other tasks that they would be doing.
Mm-hmm.
I'll give you an example, because one of the really great things about Copilot is, not only is it helping our core audience of BDRs, SDRs, and some account executives, but it's also broadened the aperture of solutions that we can provide the go-to-market organization. And so when we released Copilot to our teams, one of the really interesting things we saw internally was all of a sudden, account managers were the number one users of the ZoomInfo platform.
Mm-hmm.
Where historically, they came way behind the new business folks. They used it here and there. Now they became the number one users of the platform, and that's because one of the things that Copilot is able to do is it pulls in all of the calls an account manager has had with a client. It pulls in all of the emails, all of the calendar invites, and marries that to what's happening in an account. And so for me, right now, before I go sit down with a customer, all I have to do is go into Copilot and say, "I'm meeting with this customer this afternoon.
What do I need to know about it, about them?" And it'll tell me the things that they're happy about, the things they're upset about, what new opportunities are in play, when the last time we talked to them was, what's happening, and did they have a good last quarter, a bad last quarter? Did they talk about sales or marketing on their last earnings call? And a project that would've taken an account manager four, five hours, maybe more, to prep me for a meeting, I'm getting through one prompt in Copilot, and thirty seconds later, I have all of the answers.
That's great.
That's a meaningful productivity uplift.
Oh, it's not meaningful, but it's crazy meaningful.
Yeah, and it really opens up-
Yeah.
the use cases for us.
Yeah.
So instead of selling just to account executives and BDRs and SDRs, we now have an opportunity to expand to account managers, customer success managers, and really bring a broader assortment of go-to-market professionals into the ZoomInfo platform.
Two things that are running through my mind as you walk through that. Thanks for that. One is, how long did it take for you to develop this, and what technology did you build this on? Is it OpenAI or,
Yeah.
- or, Gemini, or?
Yeah, so Graham mentioned, I think we've touched on this, we're using a number of different models underneath Copilot for different use cases, and some of that is based on which LLM produces the best results, and then some of that is if we have two LLMs that produce similar results, and one is a third of the cost of another LLM, we're making cost decisions, too.
I see.
Meanwhile, from a cost perspective, all of the models are coming, the prices are coming down exponentially.
Mm-hmm.
And so it's built on, and there's OpenAI.
You mean the price per token?
The price per token.
Yeah.
So we have OpenAI in there, we have Anthropic in there, we have Gemini in there, we have our own LLM that we've built in there as well. And then we've been in development now on this for a little bit over 18 months.
Mm-hmm. 18 months, you got $18 million in VC. Was that a goal? Like, do you spend 18 months-
No.
On 18. So, your perspectives, you, you're deep in the front office domain. There's a concern that this technology is gonna make salespeople, marketing people, service people, so much more productive, we're gonna be not needing as many. So certainly, nobody's calling for this workforce to be trimmed or laid off, but do you think this might mean that we add headcount at a slower pace in the future? Not you guys-
Yeah, yeah, yeah.
B ut your customers-
Yeah.
the domain of front office, will this diminish the need for adding to the workforce in the future?
I think that I think that what all of these tools have always done, not just the AI-driven tools, but the software tools before it, are that they drive productivity in your workforce, and you can decide what you wanna do with that improved productivity. You can invest it back into your product. You can invest it in acquiring more customers through your marketing channels. You can invest it in hiring more salespeople to take on the demand that the marketing channel is generating because you have AI running sort of who to go out to and when. But I think there will be decisions to be made around the increase on what you do around the increase of productivity, and businesses are gonna make those decisions differently, one business to the next.
I love that because the defensive view to look at, oh, yeah, it's gonna lead to people cutting costs and getting more out of there. But we're so capitalist and so competitive as a country, as an economy, as a business world, that the moment you start to see these savings, how can we get the best-
Right.
out of that savings?
Why wouldn't it mean hiring more salespeople?
Yeah. Exactly. Exactly, right. And so people that don't do it will lose to people that-
Exactly. Take the savings from Generative AI and plow that back into growing the business. That's the cycle of tech that we've always seen.
Yeah.
That it's like the top 1% are getting massive productivity, and the others are not gonna sit behind and watch this thing go by.
There are very few companies who are like: Yeah, once I get to 15% growth.
Yeah.
I'm gonna stop.
Yeah. Yeah. It's like you're gonna get there, and you're gonna wanna continue to grow, and you're gonna need to invest back into [crosstalk].
See, you're making me even more bullish on gen AI.
You should be very bullish on gen AI.
I know. Yeah. So this is a great way to end this conference, right? I mean, who knew that we'd be picking up these threads, and maybe we're thinking about it the same way because we're software people, we're optimists. But I cannot help but wonder if, given your purview that you sell to the front office, your customers are not saying that, "Well, Henry, you got this great Copilot thing. By the way, we're gonna cut down BDR.
Yeah.
We're gonna cut down the [crosstalk].
I mean, actually, a great example of this is look at what GitHub Copilot was able to do. When I think of GitHub Copilot, I immediately think 20% more engineering output with the same people.
Mm-hmm.
But at no point was I like: Great, cut a fifth of our engineering staff.
Mm-hmm.
I thought: Okay, there's a great opportunity to build more for our clients, get more done by giving our teams GitHub Copilot. Let's build more and innovate more and do it with the folks that we have.
Mm-hmm.
And then, if we're building better product for our customers, we're gonna be selling more. I'm gonna hire more engineers to do that.
Mm-hmm. Anything from your mind as we wrap it up?
Yeah, Henry, just one more for me. You know, you talked about the different usage with account reps on the ZoomInfo platform, and when I think about marketing and talent and operations, how are you seeing the usage of those, attachment of those, and just how do you think that that's gonna change over time?
I think right now, the two areas that we're most focused on from an attach perspective are Copilot and the customer base, where our goal is that over the next three years, we migrate all of our customers over to the Copilot platform, and we monetize that. Today, we're seeing solid double-digit uplift when we move customers over to Copilot. And then our DaaS and OperationsOS product, which serves primarily in the enterprise, where we're driving attach there, as well, as customers need that solution to do their own internal AI and go-to-market efforts. But those are the two areas we're most focused on from an attach perspective.
So Graham, is that the right way to think about it, double-digit uplift based on penetration of Copilot and us in the user base?
Absolutely. Yeah, we're really excited to see that.
Okay, great. I'm sorry. I think there's one more.
Can you be any more precise, double digit?
That was the idea. Not 99, and it's more than 10.
Mm-hmm.
That was more precise.
As we wrap up, any final questions from our clients? Yes, please. I know your name, but I'm not gonna call out your name because I don't want your name to appear in a transcript. I just realized that after three and a half days, I prided myself on saying, knowing every single client who asked a question, I'd know their name. But then I realized at the very end, it's not a cool thing.
Maybe they don't want [crosstalk].
They don't want it. Sorry.
Yeah. Yes, absolutely. It's a great question, because right now we're focused on two things from a development perspective. One is go-to-market friction, and so what are our reps telling us is keeping them from being able to sell Copilot into our customer base, and we're prioritizing for those. And then the second is conversion rate of signals to action within Copilot users. And so when we deliver you a signal, how are you taking action on it? And so what we're seeing from our Copilot users is that they're using the platform materially more than non-Copilot users. Now, it's early, and we anticipate that what we're... It's like you get a new car, you're gonna use all of the stuff, and then over time, you'll use less of the stuff.
It's so much more than what we're seeing with our core base of users. We anticipate there will be remaining a large uplift of engagement for Copilot users. On engagement, we're seeing double-digit improvement of engagement.
Any more questions? It took a long time for me to learn after three and a half years. Don't call the client out by their name, although it's. Any more? Yeah. I will not repeat your name this time.
Just on the SMB side, I mean, clearly, you've had better growth performance at enterprise and market now, but SMB seems like it's so [crosstalk].
Yeah, look, I think SMB continues to be an area that's under pressure, obviously, much more so than the rest of the customer base. What we're doing to really offset that is, number one, in new business, we talk. I don't know if Graham touched on this, but one of our issues in the SMB base this last quarter is we had an escalating number of SMB customers write off, and they weren't collectible. And so we rolled out a risk model for the customer base, and we're pushing any low-end SMB customer through an upfront prepayment before they gain access to ZoomInfo. We did $11 million in the second quarter that got pushed through that. We're seeing that trend continue.
We think we can continue to improve on that, and so we're getting that upfront prepayment from the SMB customers. And in the quarter, we had our best mid-market and enterprise new business quarter. And so we're bringing in healthier customers into the customer base than we have historically. And then within the existing SMB customer base, they continue to be under pressure.
But this is a predictable factor at this point?
It's more predictable than it's been.
Predictable pressure. On that note, I'm feeling more bullish on generative as a result of this discussion. Any closing thoughts, Henry? What would you like the audience to walk away with?
Look, I think we feel like we're making a lot of all of the right decisions to continue to improve the business, to get it back to growth. I'll tell you, you know, we're committed to driving growth of this business, driving that growth efficiently. We've committed to this floor of $1 of free cash flow per share in 2024 and growing that in 2025, so you can anticipate that we'll continue to be a meaningful buyer of our shares at the company. I was also a meaningful buyer of our shares as an individual at the end of the quarter. And so we are bullish internally about getting back to growth at ZoomInfo and improving the trend lines in net retention and continuing our success in new business.
Got it. On that note, you had the last word at this conference.
Good.
Thank you so much once again, guys, for coming.
Thank you.
Let's give a round of applause for Henry and Graham.