BLOG

Kristi Faltorusso:
Unchurned is presented by UpdateAI.

Brent Grimes:
A single health score isn’t that useful. Right? I think a a lot of companies, the industry’s kind of accepted that a general health score is the way to go, and it’s it’s better than nothing in terms of prioritizing customers, but it’s really really limited. Right? Because if you try to do everything with one score, it’s not gonna be very good at any of the the specific things you’re trying to do. So we’ve introduced this approach that we call outcome based scoring. And the idea is that you should have a you should have many scores in your business. And every score that you develop should be optimized for a specific revenue outcome.

Josh Schachter:
It’s super interesting. And it is difficult in customer success to to show quantifiable measurable results because it is so multi correlated. Right? Regression is such a a difficult thing in our space. So how are you showing, the impact? Hello, everybody.

Brent Grimes:
Oh, hey, Josh.

Josh Schachter:
Hey. Hey, John. Thanks for your feedback. Yeah. Thanks for cutting off my intro. I appreciate that very much. Welcome to Unchurned with myself, Josh Schacter, Christy Falter Russo, and the indomitable John Johnson. We’re here today with Brittney Griner.

Josh Schachter:
I’m sorry. Brent Grimes. I get this confused a lot. Yeah. Brent, I’m sorry. I’m sorry, Josh. Has anybody else ever said that to you, like, that that like, for whatever reason, when I hear your name, I can’t hear anybody. Has.

Josh Schachter:
Well, I’m a single person.

Brent Grimes:
You’re a trailblazer, Josh. You’re the

Jon Johnson:
1st Okay. I’m a trailblazer. Yes. He is. Such

Jon Johnson:
a positive. Yeah. You just

Kristi Faltorusso:
really really play blazing a trail here.

Josh Schachter:
Yeah. Yeah. Speaking of blazing, so, John, how was your weekend?

Kristi Faltorusso:
My weekend, it was

Jon Johnson:
it was good, Josh. Thank you for asking. Do you know something that I that I that the listeners don’t?

Josh Schachter:
No. No. No. No. No. No.

Jon Johnson:
I just No. No. No. No.

Brent Grimes:
There’s no

Kristi Faltorusso:
I did a good amount of mushrooms this weekend.

Brent Grimes:
Just a few.

Josh Schachter:
Yeah. Yeah. Yeah. Yeah. The the the pharmaceutical kind, the therapeutic kind. Yeah. Yeah.

Brent Grimes:
How about you, Josh?

Josh Schachter:
You know, I worked 8 hours. So we’re filming this on Monday, and yesterday was Sunday. Right? It comes out on Wednesday, but I worked 8 hours yesterday. It’s the cross that we bear.

Brent Grimes:
We’re not

Jon Johnson:
proud of this, Josh. We’ve we’ve talked about this.

Kristi Faltorusso:
Yeah. I thought we weren’t celebrating this.

Jon Johnson:
Yeah. Yeah. This is a boo. Is there a boo?

Josh Schachter:
I met with, I met with with one of our largest prospects in person a couple weeks ago with the COO, and he’s been running the company with his 2 other cofounders for 16 years. And they have, like, I don’t know what’s called, like, 50,000,000 plus ARR company. And I said to him then, I was like, you know, does it ever stop? Like, what hours do you work now? You know? He said, well, now it’s, like, 12 hours a day. Before it was you know, and then he’s in his fifties. Before, it was 15 hours a day. I was like, alright. 12 hours. I can deal with that in a few years.

Josh Schachter:
He’s like, yeah, man. But I didn’t see my kids grow up. So I was like, oh, well, that’s interesting.

Jon Johnson:
Yeah. We got a pivot.

Jon Johnson:
Well, you know, this is the this is

Brent Grimes:
the prison of your own making, you know. So it’s just, like, you have to be a little a little sick in the head to jump into this, but it’s, I wouldn’t trade it for anything.

Josh Schachter:
Well, let’s I I would I would trade it for a $1,000,000,000 acquisition. So, right, let’s let’s, and hopefully, you would too. So we know your

Brent Grimes:
price. Alright.

Josh Schachter:
Yeah. Yeah. Yeah. Yeah. Exactly. Yeah. It’s out there. Brent, I don’t think I did a proper introduction of you unless you count my my mistaken, association.

Josh Schachter:
Let’s let you introduce yourself. I know we’ve got a a a wealth of history in the space in customer success, and now you’ve got a platform that’s not just CS, but, you know, is very active in the CS market. So tell us a little bit about yourself and and Reef. And I’m interested in Reef because frankly, like, we’re both, you know, up and coming platforms that are CS centric. I think we share an investor or 2, and, let’s go from there.

Brent Grimes:
Yeah. Super excited to be here. So I’m Brent Grimes, founder CEO of Reef AI. I am based in Honolulu, Hawaii. And so Reef is split between Hawaii and Pacific Northwest. So one of our goals is to prove that you can build and scale a company, outside of in in the US, but outside of the continental US, which I don’t think has been done a lot. So excited to be on that journey.

Josh Schachter:
Is that I’m gonna I’m gonna stop you. I’m gonna cut you off now immediately just for the sake of it. Is that, like, when you’re pitching to investors, when you have pitched to investors, does that actually come up? Is that actually, like, do you have to objection handle that, and do you have to prove yourself against that?

Brent Grimes:
Yeah. So it’s, I don’t think this could have could have been done 5 years ago. Like, pre COVID, I think it would have been too big of a a barrier. But post COVID, I think, one, people are used to distributed work. There’s startups popping up in a lot in a lot more places. And I think there’s a little bit of a, like, okay, you live in Hawaii. Do you actually work that hard? But I talk about my MuleSoft experience to clear that. So when I I moved here in the middle of my MuleSoft run, which I was at for 7 years prior to starting Reef.

Brent Grimes:
And my first 3 years, I was on a red eye every Sunday night. I’d leave Sunday night, land Monday morning, kinda work either at San Francisco or somewhere else in the world and come back Thursday or Friday. So 3 years of that, at least kind of build a little bit of credit around your ability to dig in and and grind when you need to. So it hasn’t been, you know, I didn’t know going in whether it’d be a big objection, but it hasn’t been hasn’t been an issue so far.

Josh Schachter:
That’s good to hear. Good to hear.

Brent Grimes:
Cool. So Little bit. More then on on my background as I mentioned. So joined MuleSoft in 2014, kind of get launched the CS org and, scaled it through IPO and Salesforce acquisition. Prior to that, I was at a company called ServiceSource. That’s where I really got my religion around recurring revenue, management. So So what we did is we outsourced recurring revenue for a lot of big technology companies at the time. Right? And this was just when SaaS was emerging, so it was a lot of recurring maintenance revenue, there’s a little bit of SaaS revenue.

Brent Grimes:
Right? But, most of these organizations did a really poor job of managing those revenue streams, So we’d come in and just take over the take over the management, and, we had a a gain share model where we’d only take a percentage of the incremental revenue that we drove, and it was a really successful model, you know, we took the company public, and I learned 2 big things from that from that experience. 1 is recurring revenue stream can be really, really valuable even more so as SaaS was becoming the predominant business model. 2, it’s very nuanced, different, and in some ways harder than kind of new customer acquisition. Right? It’s just a very different game, I think in a way that was underappreciated at the time. So that’s where that’s what really sparked the passion for this emerging customer success space and and what I learned there. So I I really took that those learnings into the the ride at MuleSoft, where, we built and led the organization and really transitioned our business model from a very much open source inbound model to a very heavy, kind of, top down, outbound go to market, where we played around with customer scoring to the to an extent that really sparked the idea for release.

Josh Schachter:
So tell us about it. Yeah. Tell us your origin story.

Brent Grimes:
Yeah. So, when we were at MuleSoft, as we were the company was really nice. So me, the my colleagues in customer success. So I have, someone that I that I wore down over a few years, got her to join, Panel Lombardo, Oh, that’s nice. Is now, leading growth at Reef, and I’m working on a few few mules to bring in as we as we continue to scale. But, we were scaling the business quickly, and at the time, we were doing what a lot of companies do, which is try to do a little bit with most of your customers and and kinda keep things going. And we realized as we were preparing to go public, I was getting a lot of pressure to reduce cost. So do more with less, we need margins to to look better.

Brent Grimes:
So we realized pretty quickly that we can’t just keep scaling the way we were scaling before, and we have to get smarter about making selective investments into customers and not trying to do a little bit with every customer. And so we had some of the traditional kind of off the shelf CS solutions. We had Gainsight at the time. We had, Salesforce and some other tools we were playing around with, but we really needed to get beyond kind of traditional health scoring. We needed something that was more data science based, more probabilistic, and so we just couldn’t find anything that would meet our needs. So we decided to do this as an internal project. So I begged, borrowed, and stole internal resources from data science, ops, analytics, and we essentially bootstrapped this thing. So we pulled a bunch of customer data, we aggregated the data, we started this with a manual regression model, we eventually started playing around with machine learning, and we were able to score our customers with a level of confidence that we were willing to make these trade off bets and move from a traditional customer coverage model to what we call outlier engagement, which is, you identify outliers both on the growth side and the risk side, and channel your resources in an outsized way into those outliers.

Brent Grimes:
And really, you spend a lot of less a lot less time with customers in the middle just keeping the lights on. Right? Do you trust your scoring? So and it wasn’t just customer success. We got sales on board, we got services on board, we got presales on board to really invest in these customers in a in a different way. And my goal going into this was, maintain performance at a lower cost footprint. Because we were we were our net retention was good, we were probably a 120% plus at the time. So we we didn’t have a performance problem, but I was I was worried that we couldn’t keep it up if we had to hit that cost. But what happened was 2 to 3 quarters later, we saw our growth rates start to reaccelerate, we saw our net retention start to to improve, and we got the best of both worlds. We got lower cost and higher performance.

Brent Grimes:
And that’s when the light bulb went off for me that there gotta be something here. There’s a there’s an area of customer success that’s been severely under invested in, and it’s this whole idea of advanced scoring. We’ve got a ton of customer data, but we’re still relying on kind of assembling fields and weighting them and using our best guess kind of gut instinct to score customers, where there’s a whole world of of data science to apply where you can really start to get more, you know, probabilistic about the way you prioritize customers and then you kind of drive investments in order to, you know, grow faster and churn less and just perform form better in terms of in terms of net retention. So let me pause there. Any any questions on that? And we can

Jon Johnson:
Yeah. I I I’m like, I’m sure we all have 100. Can I go first? I’m gonna go just with one of mine. I’m like jotting it down. Oh, John, you’re so kind. Queen. You’re so kind. Brent, so what are some of the data points that you’re pulling in that aren’t part of, like, the standard health score models that we hear people talk about all the time.

Kristi Faltorusso:
What are the other data points that you’re looking at that you found more more impactful?

Brent Grimes:
Yeah. So part of it’s about the data points. Part of it’s about what you do with the data points. Right? So, I’ll answer I’ll answer your question, but I’m gonna start with the stuff that is pretty common. Right? So we’ll connect to your CRM, we’ll pull in the stuff that you’d expect. Right? All of the sales history, whether you won or lost, you know, contract information, product information, at the line item level, start dates and end dates, you know, firmographic information about your customers. We’ll pull in support data, we’ll pull in marketing engagement data, We’ll pull in product telemetry. Right? But I think there is there is something really important in our approach, which is it’s not a black box scoring approach where we have, like, a common algorithm across all our customers, and we plug in your data points, and we kind of spit out a, you know, your customer is a 78.

Brent Grimes:
And so, you know, what do I do with the 78? I’m not exactly sure. I know it’s better than a 77, but I don’t know exactly why. Right? So we really took a different approach, which is, one of the things that’s really failed in scoring is just building trust around the scores and what they mean and why should people should trust them. So scores really need to come with a a PR campaign. Right? And there is a level of trust that you can kind of, instill in those scores. So what we do is we say, look, yeah, there’s common there’s common things across our customers that we bring in to reach because we have unique data inputs for every customer, but it all flows into a common data framework so that we can really, deliver scores quickly and we can operationalize those scores in a way that’s really easy for your teams to act on, either in Reef or in your current customer success platform. But the thing that’s really different about Reef is it really is like having a data scientist specifically focused on what you care most about. Right? Because you have a say in kind of what we bring into the model and test for correlation.

Brent Grimes:
Right? If you’re like, hey, there’s something I think is really important that I’ve just never been able to prove that actually correlates the churn of growth, we can include that in the model. Right? So product telemetry, every customer has a different product that they’re selling, and we’ll make sure that we try to understand what are those key product events or key product experiences that are a good reflection of value realization. I mean, it’s interesting that they’re logging in and they’re interesting that they’re clicking on certain things. That doesn’t necessarily mean they’re getting value right. So we really lean on your own understanding of your business to identify what are those things that actually are a reflection of value utilization and bring those into the model. And then as we as we develop models, we allow you to get full transparency into the performance and accuracy of the model, so how well does it perform? And 2, what is the importance of the different things we’re testing? What are the features that have the highest correlation to churn or to growth, and really help you have that transparency to know you can trust the score and to communicate to your team. Here’s why we’re betting on this score. Here’s what it means.

Brent Grimes:
And when you get a recommendation, you can believe it because of x y z.

Josh Schachter:
So you’re, you’re basically benchmarking you’re doing a regression of what different types of factors are correlating to churn or renewal or expansion. And then you’re basically benchmarking a book of business against itself to show effectively, like, the matrix. I think of, like, a conditional formatting almost in Excel of here’s the distribution. Here’s the, you know, here here’s the the the folks that are top 20%, bottom 20%, etcetera, based on this regression analysis of all these factors.

Brent Grimes:
Generally, that’s correct. We regression is one of the approaches that we use, but I think that, one of the things we learned is that a single health score isn’t that useful. Right? I think a lot of companies, the industry’s kind of accepted that a general health score is the way to go, and it’s it’s better than nothing in terms of prioritizing customers, but it’s really really limited. Right? Because if you try to do everything with one score, it’s not gonna be very good at any of the the specific things you’re trying to do. So we’ve introduced this approach that we call outcome based scoring. And the idea is that you should have a you should have many scores in your business. And every score that you develop should be optimized for a specific revenue outcome. Right? And if you can optimize a model for that outcome, and then drive recommendations based on that, you’ll be able to 1, have an effective score for that outcome, but 2, really understand how you performed before driving these recommendations and then how you performed after in the in the impact.

Brent Grimes:
Right? So you can really start to understand ROI at a granular level that you can understand per score, per person engaged in the score, and it really helps to close this gap on, well, how do we quantify the impact of customer success? Alright? And so yeah. Go ahead.

Josh Schachter:
Well, no. No. I I it’s really it’s super interesting. And and as you know and and as Christy knows, we all know, like, it is difficult in customer success to to show quantifiable measurable results because it is so multi correlated. Right? Repression is such a a difficult thing in our space. It sounds like you’ve got a a a good angle for that. So how are you showing, the impact of the platform? Like how are you producing those case studies with the quantification?

Brent Grimes:
Yeah. Great question. So we do it, first we you have to understand the business at a very granular level, and we really focus on on revenue. Right? Because revenue is the thing that you need to start to prove the before and after and and the impact on on what you’re doing. Alright? So as I mentioned, we take this outcome based scoring approach, and just to explain more about what an outcome based score is, it could be something very general, like a churn prediction score. Right? Hey, we have a problem with logo churn. We wanna model that is optimized to predict for logo churn, you know, 6 to 9 months in advance of renewal, and we wanna drive recommendations out for the team, while there’s still plenty of time to engage and change the outcome before you get to renewal. Right? That’s a very common use case for a roof.

Brent Grimes:
But you also might have something much more specific and granular, like, hey, we’re not having a problem with full churn, we’re having a problem with downgrades of a specific product. So can we have a model that’s optimized to predict for downgrades of this product, 6 to 9 months in advance of renewal, so that we can send a signal to the team that they need to engage this product specifically to start to drive use cases and put that product in a better position before we get to time of renewal. And then, or it could be something on the growth side. Right? So one of the common use cases we have is, like, hey, we need to drive more large deal expansion pipeline. So we have a model called the big deal finder that will go in and we’ll look at what are your past wins, let’s say, over the last 12 months where you had large outsize expansions. And we use a little bit of a different model approach. Right? For for our churn prediction, we use what’s called a survival model, it’s a it’s a tried and true model for churn prediction, but for growth, it’s common for us to use what we call a, clustering similarity model, which basically profiles your large expansions, your customers that expanded in a large way over the last 12 months, and we find the common attributes amongst those customers. So it might be, what products did they own and how do they engage those products? It might be firmographic information, were they in a certain industry or were they large small, and so we’ll find the common patterns amongst these customers and then overlay that pattern across your existing customer base and say, who are the customers in your base today that most closely match that pattern that are statistically more likely to produce, more more pipeline, convert at a higher rate, close at a higher rate, at a higher ASP.

Brent Grimes:
So we really have these this library of these optimized scoring models that we can apply to specific revenue outcomes to start to drive improvement. And then, so back to your question around, well, how do we calculate ROI? A great example, we just did a QBR with a customer recently, but we built a score to predict for downgrades of specific products, or churn of specific products at time of renewal, because again, they weren’t having a full churn problem. So what we were able to do is we were able to optimize a model that predicted for this, downgrade risk. And then in our model, we were able to prove that if a customer is red in this scoring model, they’re 43% likely to churn. If they’re yellow, they’re 8% likely to churn. If they’re green, they’re less than 1% likely to churn. Right? So as a strategy, it’s like, okay, 8% for yellows, it doesn’t make sense to really spend a lot of time triaging your yellow customer, because 8% is pretty low. There’s a big pool of them.

Brent Grimes:
You’re gonna waste a lot of resource. But 43% is a meaningful signal. We can’t tell you exactly which 43% are going to churn, but the pool is small enough that if you go triage every one of those customers that are red, well in advance of renewal, you’re gonna move the needle on renewal risk, and you’re gonna drive more more revenue, less churn. And that’s exactly what happened. Right? So in 1 quarter, we were able to show 500 k almost in ROI. They had about $2,000,000 in red account that we were able to identify, and we we were able to send 83 recommendations on those expiring contracts that were at risk. And what happened was, for 51 out of the 83, they were able to move the customer from red to yellow or red to green. And again, these are all objective signals.

Brent Grimes:
This isn’t what the the opinion of the CSM, these are objective signals that we know have this correlation to to churn in this case. Right? So

Josh Schachter:
That’s right.

Brent Grimes:
By moving by moving customers from red to yellow, red to green, we were able to take what would have been traditionally because of our benchmarking on that 2,000,000 they would have churned traditionally 881 k. And we were able to cut that 88181 k down to about 400 k, kind of after we ran this exercise. Right? So within one quarter, we were able to show 500,000,000 in ROI.

Josh Schachter:
How do how do you know for instance that, like, the correlative power of, let’s say, a billing concern is has this amount of a potential impact and probabilistically relates to churn in this way? You know, 10% increase or 10% chance or 90% chance, etcetera.

Brent Grimes:
Yeah. Because we essentially, like, prove it out in the data science, and then we’ll show you a report that says, hey, here are, you know, we tested, let’s say, you know, a 100 different conditions for correlation to churn in this case. We’re gonna show you the bar chart of top to bottom, or the top 20, or over the top 30 of the most, the highest correlating features on the top to bottom, so you can understand what the things are, what what the data tells you in terms of what’s most important. Out of that list, there will be certain things that you can’t control. Right? Maybe there’s just an industry that’s having a a downturn right now and they’re just cutting back on everything. Right? And so, you know, a little harder to action but not impossible, at least in terms of forecasting, in terms of, like, planning the business. You’ll know, like, hey, we’ve gotta either put special attention into this industry or we’ve gotta write off that we’re just gonna have kind of lower retention in this industry while it’s struggling. But there are a lot of things that you learn that you can control.

Brent Grimes:
These are the things that we really focus on in terms of putting context around recommendations, in terms of what can we actually do to control our own destiny, drive the right action into these customers when we when we find risk, and, work to change the outcome. Awesome.

Jon Johnson:
So Brent, are the recommendations coming from the are the are the teams designing the recommendations or kinda like, I don’t know, effectively, the playbooks? Or is your technology guiding them to say based on what we’re observing, here’s what you should be doing, and here’s how to run out of that.

Josh Schachter:
Christy, that encapsulates that encapsulates old school CS versus new school CS platforms right there. I’m sorry. I love this so much.

Brent Grimes:
Josh? No.

Josh Schachter:
I love this so much. But this but this is this is why Brent created company. No. Brent’s

Kristi Faltorusso:
And that’s why I want to specify this, Josh, is because everyone who’s listening is still doing things the old way. Right? And so we need to break this out and explain to them if it’s doing it differently, how and why.

Josh Schachter:
You’re right. Yeah.

Jon Johnson:
Brett, half

Kristi Faltorusso:
assed talking to Josh. Josh. And everyone’s as smart as you, Josh. I love that. Has dotai in their URL.

Josh Schachter:
We’ve had it for 4 years just by the way.

Kristi Faltorusso:
Okay. For the rest of us. Okay. Brent, go ahead.

Brent Grimes:
Alright. Great question, Christy. I think,

Jon Johnson:
it’s a combination of

Brent Grimes:
question. The combination of both, to be to be perfectly honest. Right? So we learn, we we uncover new information with the scoring that we do. Right? There will be some things that are very intuitive that we learn, There’ll be some surprises. Every inevitably, there’s some things that we’re just we’re just blind to before that come up in the scoring. So making sure that, those learnings are considered when we drive recommendations is really important, and we take the context of, okay, why is this customer at risk? What is the specific revenue outcome that we’re scoring for? And we’d make sure the recommendation is really closely tied to the situation and and the specific conditions. That’s not to say though that the experience of our customers, they know their business better than we do going in. Right? And they’ve got a lot of tribal knowledge and kind of an instinct around what’s important in their business.

Brent Grimes:
So it’s not just, hey, trust trust the black box and and hope for the best. It’s really incorporating, what have you learned and what have you known to be effective, and we’ll test that in the model. And if it actually does show that there’s correlation, you probably have existing plays or playbooks that you’ve developed that you just need to insert properly, and not just kind of put them out there for the world to consume on their own, where you can actually say, okay, here is a time and place to go run this playbook or run this play, and incorporate that into the the model. And there’s really two levels of of kind of value that the reach can provide in this. 1 is what we call smart targeting, which is using scoring to find the right customers to engage at the right time. And then there’s smart recommendation, which is learning from the why of the risk or the why of the growth and the models themselves. And then at the point of attack, providing the best recommendations possible for a given customer. I’ll say that we’re really far along with smart targeting, and the smart recommendations is really still work in progress for us.

Brent Grimes:
Right? We’re still doing a lot of experimentation to say like, hey, if we send 20 recommendations for 20 different customers, can we actually have each of the recommendations for those customers optimized for this for that customer specifically in that scenario. Right? So that is still road map for us that we’re developing, but making a lot of quick, kind of, r and d progress on. But the smart targeting plus the, hey, what do we know about this customer and, the model, and what is a good, kind of, common sense way to engage this customer now that we know the best time and and place to engage them, is all coming together. So that’s the way that we view kind of this day and tomorrow at what we propose.

Kristi Faltorusso:
Brent, just on on the recommendations, if the team sorry, John. And then I won’t speak anymore. I’m sorry. I know you’re biting your lip. I know. You’re so good to me.

Jon Johnson:
I appreciate it. To look pretty.

Kristi Faltorusso:
No. It’s just It’s just hard

Brent Grimes:
to tell this time. About to open his mouth.

Kristi Faltorusso:
I do. I do. And then I see him, like, trying to do something. I just cut him off.

Josh Schachter:
Hey, everybody. It’s Josh. I’m taking a quick break from the podcast to tell you a bit about UpdateAI. I started UpdateAI to solve 2 major challenges for CS teams. The first is that we save CSMs 4 to 5 hours per week with our productivity through AI. Secondly, we give leaders a window into all the conversations across each account and the entire portfolio. So we help knowledge transfer, we help increase the coverage model of your CS teams, and we help you detect emerging patterns and what your customers are telling your CSMs across all the risks, product feedback, advocacy moments, and expansion opportunities. So come check us out at www.updateai.

Josh Schachter:
It’s completely free to sign up and trial.

Jon Johnson:
So as your your technology is, like, learning all of these things, how are you tracking what the team is doing to then be able to say, like, these are the right things to do based on what we’ve learned. Right? Like, our our CSMs effectively having to, like, log, like, here is what I did and here was the outcome for you to validate that these are the best next steps?

Brent Grimes:
Oh, you you you struck a nerve here. So I do have a very strong point of view around the amount of time that CSMs are forced to waste in their jobs day to day is is really kind of, inexcusable. Right? I think that CSMs are kind of I think the busiest people in in their companies and they’re but they’re mired with a lot of kind of low value, low impact busy work that really isn’t helpful at the end of the day. Everything from keeping CSPs up to date and entering a lot of data manually, to engaging customers that don’t really need engagement just to say that they did a did a check-in. Right? There’s just a lot of and so I have a very per strong personal philosophy around a less is more approach with customer success. And I think that you’ll see that, you know, over the last year and a half, like, with the cutbacks in customer success and kind of, you know, correction that, like, CS teams don’t have the luxury of just throwing a lot of bodies out there and hoping for the best. They have to actually really be smart about the way they allocate resources. They have to really, kind of, prove the impact of what they do and start to calculate, kind of, yield on on CS investment.

Brent Grimes:
So, we’re very big fans of look, you need to free you know, CS should be doing less busy work. They should be have time freed up to spend more time engaging customers, doing the high high value work that they were hired to do. And so where that plays in terms of recommendations is that we wanna make it as easy as possible to action the brief recommendation. What’s important for us is that we get a feedback loop because, you know, we want a thumbs up, thumbs down. Think like the, you know, the Netflix algorithm. We wanna know if it was a good recommendation, if the CSM thought it was valuable, kinda based on their own knowledge of the customer and context, and they can they can reject a recommendation. We don’t auto assign actions, because we think that it just leads to a bunch of, you know, the what I call graveyards of guilt, a bunch of tasks that just never get done that people feel bad about, but we can’t possibly possibly pursue.

Jon Johnson:
I just I’m gonna interrupt you to say, like, thank you from the bottom of my CSM heart for not auto assigning suggestions.

Brent Grimes:
You’re welcome. No. It’s it’s a it’s a it’s a foundational part of the model. This idea that, look, these are smart people, they know their business, they know their book of business. We’re not gonna presume that, kind of, we know more about their customer than them, but we’re going to feed them with information that’s useful, when we find it. Right? And so they make the call on whether the recommendation makes sense in context, and then they can accept it, they can reject it, and then they can go engage the customer, because they’re already running a play, or they can say, hey, what you know, give me give me the play to run, because I’m not exactly sure what to do, and then Reef will generate what we call a sprint. Okay? Here are a few actions you can take over the next week or 2 to go mitigate this risk or drive this growth opportunity, and then we’ll track all of that. We’ll track the quality of the recommendation, we’ll track whether it’s being actioned, and then all of that is feeding back into our data model, So that we can, 1, it makes our scoring model smarter, because the quality of recommendations, will continue to tune the model and make them perform better.

Brent Grimes:
But it also allows us to then show, okay, what are the recommendations that were actions, and then what’s actually happening to the revenue outcome we’re trying to improve, so that we can start to give you that data to correlate, hey, here’s where, you know, this CSM spent their time over the last month, and here’s what they would have done before getting these recommendations. Here’s what they actually did, they they saved over 3 k and predicted churn. Right? So they’re actually making a difference based on where they’re spending their time.

Jon Johnson:
I love that.

Josh Schachter:
You guys had a a pretty sizable booth at Pulse Gainsight’s Pulse. How?

Jon Johnson:
How are you doing? Question there. No.

Josh Schachter:
No. No. No.

Jon Johnson:
No. We did.

Josh Schachter:
Yes. They did.

Kristi Faltorusso:
Josh is like, and so where exactly did you take the funding for that?

Josh Schachter:
No. That’s not exactly what I’m asking. What I’m asking is, it it feels like you wanna take them out.

Kristi Faltorusso:
He’s saying they don’t feel very complimentary.

Brent Grimes:
No. Actually, I mean, no. To be honest, like, there’s a there’s a place for Gainsight. There’s a place for and we we operate in in very different problem spaces. Right? Like Mhmm. I would say that we, make Gainsight more useful and allow you to get more ROI on your Gainsight investment. Right? Because there are things that a lot of things that we don’t do that a CSP does. Like, we’re not we don’t do kind of heavy workflow, we aren’t about kind of general customer data capture and management.

Brent Grimes:
Right? We’re really about being the best scoring engine out there to help your team prioritize their time. Right? So I I think one of the biggest drawbacks of Gainsight is is that it’s really hard for people to know, like, what’s a high value activity, what’s a low value activity. You know, I I have a lot of stuff that gets assigned to me, but where should I really be spending my time? If we can help people cut through the noise and really say, okay, here’s recommendations coming through, because people can, you know, reef is self contained and it allows you to operate independently. You don’t have a CSP, but if you do, we can play nice with their CSP as well. Right? So it might be your ops and your leadership team that engage Reef directly. CSMs may not even know Reef exists. Right? The recommendations are just showing up as CTAs in Gainsight, or or activities in your activity management system. But we try to make sure that they’re flagged with a different level of priority and import, because we can prove that there’s revenue on the line for this recommendation, and it should be treated as a as a high priority.

Brent Grimes:
Right? So we do believe there’s a world where we can just help people get more more investment more return on their gains at investment, and really prove the ROI of the combined systems in a way that they haven’t been able to before. So, again, I think it’s, you know, a nice provocative, prod there. And, again, I think that we’re we’re on a little bit of a different mission. Right? We’re not out to be a CSP, and we do use cases for CS, for sales, for support, for services. Right? So I think we can because of the the broad nature of scoring and where it can be applied, I think that, you know, our vision is, you know, our heart is in CS. It’s waiting that’s where I spent a lot of my career, and I think it’s really where ARIEF was born. But your vision is to really provide deep scoring horizontally versus just, just being a a kind of end to end CS solution.

Jon Johnson:
Well, I think one of the things that stands out here too is you you said that CSMs may not even know that Reef exists. They’re just gonna get a CTA in Gainsight. Right? So Yeah. That’s something that can be played across every CSP. Right? If there is a design engine for this health scoring.

Brent Grimes:
Yep.

Jon Johnson:
And I will say, like, some of the the hardest parts of any CSP is actually trying to mold how their framework for health scores intersects with the way that our companies are trying to design. And one of the long held beliefs that I think we’ve talked about many times on this podcast is that a lot of the processes that CSMs go through every day are not predicated on customer need. They’re actually predicated on the way that a CSP is built. And we’re just kinda, like, shoving things into boxes in an unelegant way when in reality, you know, there’s a lot of hoops that customers have to jump through. And our job as CSMs is to make those hoops as easily achievable as possible. And if we’re locked behind the restrictions of a CSP and a health score that is like, oh, it’s just these four things and then maybe your gut. I don’t know. Yeah.

Jon Johnson:
There’s a lot of intelligence behind that. And I’m gonna stop blowing your skirt up for a second and ask you a question. I swear. So when you’re introducing the idea of, like, AI and intelligence behind a health score or, large datasets. Right? I mean, you’re talking about looking at customer customer data. You’re talking about looking at behaviors, telemetrics, inputs, outputs, exports, imports, all of these things. What is the market been telling you kind of on the bleeding edge of this to say, hey. That’s my data.

Jon Johnson:
Don’t touch it. Or, how are folks feeling, like, in the space when you present this as the win, but then you kinda get down to the nuts and bolts of and we have to look at your data, or our tool looks at the data.

Brent Grimes:
Yeah. Yeah. That is the that is the the question of the day, and it’s quickly evolving too. Right? Because I’m sure you guys see firsthand on the the generative side. There’s a lot of a lot of big questions and and frankly concerns around the way the way that, you know, models are trained and and the way that, the data is used. There’s a few things that we’ve done. One is just we invested very early in data security, so fully soft 2, type 2 compliant, you know, GDPR, like, making early investments in data security. We knew it was table stakes based on kinda, the the size of companies we’re working with, the datasets we’re working with.

Brent Grimes:
There’s another thing too that’s really nuanced that’s important is that our models intentionally avoid, PII, personally identifiable information. So, all the data that we ingest, we can aggregate at a company level and we don’t need to know what’s happening at an individual user level for our models to perform really well. Right? And so, the way that these infosec teams work is this, if you do have to deal with personally identifiable information, it puts you in a wholly a whole different category of scrutiny and red tape and, you know, friction and kind of getting started. So by being able to really focus at the company level and stay away from PII, it’s really made our ability to really cut the red tape in our in our sales cycles, build trust with these companies much much faster than if we did rely on that PII.

Jon Johnson:
That’s awesome. Man, this has been so great, Brent. Like, I feel like we we have like, this the process that you’re going through is so customer focused, and I I really do appreciate that. You’ve hit a you’ve hit a lot of nails in, that we’ve talked about over the years on this podcast. So thank you so much for for talking through this, and thank you so much for introducing us to, Reef. Josh, do you have any other other questions?

Josh Schachter:
No. I I love this conversation. I always love talking to other founders of startups. It’s, we kinda we share that blood. And I

Brent Grimes:
have a question for you

Josh Schachter:
guys. Yeah.

Jon Johnson:
Okay. Deal.

Brent Grimes:
So, you know, we very much approach this from the, because we’re, you know, we’re playing we’re both playing in this AI space. Right? But I think coming at it from very, very different angles. Right? So we’ve we’ve approached this with kind of machine learning, with structured data. Right? So dealing with kinda data in your in your customer databases, you know, very kind of disjointed and disparate, but structured customer data. You guys have really attacked the market from kind of unstructured data and really kind of incorporating, like, all of this unstructured data through conversations, through videos, through, through lots of different means to really try to kind of find insights and and do a lot of similar things in terms of, well, where where should you be prioritizing your time, and where should you be kind of finding risk and opportunity? I think there’s a cool kind of potential future where these worlds come together. Right? Because I think that I think they’re both really valuable and they’re both kind of exposing part of the problem or part of the opportunity in terms of the in terms of the signal that you find and kind of how to action that. I think they’re really interesting kind of future for customer success that’s, like, much less about, again, entering information into your your CRM or CSP. I mean, I think, you know, it’s in in 10 years, I could see a world where CRMs are just fully automated.

Brent Grimes:
There’s no need for manual manual data capture. Right? But how do you guys think about these worlds of of structured versus unstructured and kinda what are you seeing as as kind of when we think about the the future of customer success and what a what a CSM may may do in interacting with their data and kind of how these worlds come together?

Josh Schachter:
Yeah. Well, that’s a that’s a a big question. It’s a great one and it’s timely because I actually just wrote an article on on my thoughts on AI and then I actually just rewrote our vision statement just over the past week. So it’s perfect timing on that. I don’t think that there’s any data advantage anymore when I think of of of of startups and, you know, different services that are out there. I think that it’s contextual advantage. Yeah. That’s my that’s how I’m pointing it.

Josh Schachter:
Right? But when you look at the power of AI, the power of generative AI, particularly, it’s it’s so much more powerful if you have structured the data. And structured the data, but but then presented it in a way where it’s organized and contextual. You know, you can feed your LLMs and those sorts of things in in a way that makes sense. And so that’s that’s kinda how we approach it is that we take these conversations and by nature, a conversation is very unstructured. And we’re structuring those conversations and creating context from the conversations, but all the also from other things. Right? Like like you said, like pulling it from the CRM and, a little bit of UX as well on the experience and the platform level application layer, creating that context and then pushing it into our data science, right, our LLMs that we’re working on top of and all that kind of stuff. And so I agree. I mean, you and I are gonna catch up after this call and and, like, as you know, for for startups, like, our our roadmaps are also crazy.

Josh Schachter:
So what could be, like, an amazing partnership is, like, it just doesn’t happen because it’s just like we both have no time to, like, actually swim it into the like, you know, like, merge it into the swim lanes. Right? But, but it’s true because, like, what we’re doing is we’re detecting the insights. Like, some of the ones that you mentioned as an example. There’s a billing concern, you know, that’s surfacing, or there’s this, you know, a theme around downgrade re you know, requests and things like that. We’re detecting that from the conversation, could easily pass to you guys, and then you do your magic with correlating the the impact on the the bottom line of that. I I totally see that. You know, our we feel when we first started the business, we felt like I mean, obviously, we know all the data sources, product telemetry, and like that. We felt like conversations with customers are both the highest ROI and the highest cost center.

Josh Schachter:
And so and the the less harder to decipher, because they’re unstructured by nature. But, but yeah. No. I I I totally see something there. And then at the end of the day, like, I I share a vision with you of, it’s criminal how much laborious work goes into the day to day grind of customer success. And I actually do think that there’s been well, I’m gonna shut my mouth.

Jon Johnson:
But We almost got him.

Josh Schachter:
Yeah. You almost got him. But, but, anyways, like, at the end of the day, I think not a year from now, not 2 years from now, but 3 years from now, I think there’s AI agents to fulfill all these things, to fulfill the recommendations that you’re that you’re starting to to surface that are gonna be so much stronger 3 years from now. And then there’ll be AI agents to actually fulfill these requests in very interesting use cases. And that’s not to say that there’s gonna be, you know, the entire job market for CS is going to erode. There will be job loss to be clear, but there will also be a shift to more strategic work, higher value work. It’s more interesting, quite frankly. And so we’re building the future of structuring the conversations, creating a knowledge base for customer for for teams around their customers, and then ultimately, not today, but ultimately creating the same types of use cases and workflows that I think you’re starting to do as well to be ready for a time when in 2, 3 years, there really is the opportunity to have AI agent workflows that are automated.

Brent Grimes:
Really cool. I’m excited I’m excited about the future. It’s gonna come, fast, it seems slow at the same time. So it’s, it’s the journey we’re on, like Yeah.

Josh Schachter:
Yeah. I mean, listen, I just presented to to, the Chicago group. You know what? I I I I Slacked everybody. I chatted everybody a few minutes ago. Like, let’s guys, let’s wrap this up in 5 minutes. And now that I’m talking, I’m happy to keep

Brent Grimes:
going here. He’s getting bored. Yeah.

Josh Schachter:
No. No. No. But but, anyway, I’ll I’ll leave it at this. Like, I I spoke to the Chicago, the CUSP, CS community last week, and one of the the the charts that I shared was, like, how quickly AGI is coming. And and just only a couple years ago, only like 4 years ago, we thought we were, like, 30 years away from it or something like that. Yeah. And now if you talk to folks, they they think AGI is about 5 years away.

Josh Schachter:
AGI, by the way, like, the standard definition is, like, effectively, it’s it’s as an AI that’s as intelligent as the 50% median of human intelligence. So, like, the scholars out there think that we’re only 5 years away from that. It’s crazy.

Brent Grimes:
Yeah. Yeah. For sure. Yes. We are in in, in the in the, tornado right now, so it’s, it’s yeah. I think there are, you know, 10 years ago, it’s 10 years from now is gonna be unrecognizable. Hopefully hopefully in a good way, and hopefully, we’ll not be, like, you know, trying to appease our machine overlords, but, we’ll see how it plays out.

Josh Schachter:
Skynet is coming.

Jon Johnson:
Yep. Awesome. Well, thank you so much, Brett. Where can we find where can people find you? Find out more information about, what you’re doing.

Brent Grimes:
Yeah. Just come to reef.ai. Super easy to, remember and and type in.

Jon Johnson:
Spelled correctly, thank you. Not like r y y f f e with the No.

Brent Grimes:
That’s my biggest Very easy. Here’s the logo here, Reef AI. I love it. Video.

Kristi Faltorusso:
For those looking, he was pointing to logo on his shirt?

Brent Grimes:
Yeah. Yes. Ef.ai. Learn more about us, set up time. We’d love to chat, nerd out on all things customer success, customer data, and, yeah, really excited about the the momentum. Excited to chat with you guys. It’s been a blast.

Josh Schachter:
Awesome. Thanks, everybody.

Kristi Faltorusso:
Awesome. Thanks, guys.