The Proactive Playbook: Building a Risk Intervention Engine That Actually Scales

So I'm Jessica Foley. I'm the senior customer success and strategy operations manager at Coursera. And what Coursera does is help individuals and companies through online learning develop their skills and progress their knowledge. So kind of in that spirit today, I want to discuss building risk intervention engines that scale for CSMs to catch risk.

So we've got six stops on our expedition map today. We'll cover why old school playbooks break its scale, how AI detection works inside Gainsight, the automation engine that ties it all together, the actual workflow, how to provide ROI to your leadership. And then I'm going to turn over time to you all to build your own intervention at this workshop. So quick raise of hands.

If you have ever found out that you have a customer that's churning and you realize that you saw the warning signs a couple weeks early, anybody ever have that happen? I am super impressed. I don't know if I'm going to teach you anything today if you're able to catch those weeks in advance. But we can all learn from everybody today.

So early in my time at Coursera, we had a mid-market customer. They had been with us for about eight months. And everybody thought that they were engaged and they would easily have a renewal. And then all of a sudden, they let their CSM know that they were going to churn.

And everybody was caught by surprise. And then when I went back into the data, I started to uncover and saw that we had six weeks of declining usage. We had our decision maker that went dark through email. And we didn't have anybody that had done any courses online in the last month.

And when you pull all those data points together, you realize quickly that's easily an identified churn. But we didn't have anything automatic in our system that was flagging that to the CSM. So that was loss ARR for our CSM, but also a decrease in their confidence. And so making sure that we are pulling all of those data points together and providing them in real time so that the CSMs who are developing those relationships, building in front of the customer, that they're able to pull those together and really take action and decrease that churn.

And I know that story isn't unique to myself or to Coursera because 70% of churned accounts actually show early warning signs that a CSM missed it. And it's not to say that the CSM is bad at their job. It just means that our old approach has four real problems. So one, we've got manual and time intensive that we expect from our CSMs.

So they're spending hours reviewing dashboards and trying to figure out in the health score what are those triggers that is actually going to make a real impact for the customer. It's also signal dependent. And so making sure that the CSMs aren't missing flags, because if you miss a flag, there's a good chance that the customer is going to slip through the cracks. Also making sure that you're not being reactive.

A lot of us have probably built playbooks, but maybe they trigger too late. And it comes after you've had a hard conversation when you really could have used that playbook beforehand. And lastly, people treating it as a one size fits all. So you've got a customer that's dropping engagement in month three, and you've got a customer that's dropping engagement in month 12.

Those are two totally different strategies. But if we're assigning the same playbook and treating it as a one size fits all, it's not going to work for everybody. So here's how Gainsight flips this. We've got a four step process.

We've got the signal that feeds into the health score. The copilot surfaces the pattern. And then the CTA fires automatically. And you've got no human touchpoint that needs to happen to make this go through.

So if we use the example, 40% logins drop, and you've got a 90 day average turnaround for health score, it's going to drop to red. Your copilot is going to flag it. You've got a risk CTA that fires it. And the whole chain happens without anyone manually reviewing a dashboard and trying to figure out where those signals are.

Also, if you don't realize this, you can use your scorecard group and then set different thresholds for segments. And so then that way, on your month 12 contract login and your month three contract login, you're treating them differently. And you've got the context to have a strategic conversation based on where your customer is in their lifecycle journey. So I want to get tactical now.

So building a rule is going to be in three steps. And if you are not an admin, go ahead and take a picture of this. You can then connect with your admin next week. But really, the idea is you're going to set your condition.

You want to have your health score like we were talking about. That's under 40. And then you have your renewal that's coming up in the next 90 days. You want to make sure that you set your action.

So you're going to create your risk CTA. You're going to assign the CSM. You want to make sure that you have an SLA. We have the SLA within 48 hours that we want people to respond to it.

You're going to attach your risk intervention playbook so that everybody knows what actions need to take. And then also, you want to have this run daily and make sure that you're suppressing any duplicates. Step three is you're going to fire that CTA and make sure that it's automatically showing up in the CSM's cockpit. And then having co-pilot, having that pre-confirmed action to take.

All of that, you don't want to have any human have to trigger that. Now, I want to be super upfront because there's been things that I have fallen through when setting this up. So one is making sure that your risk isn't your rule isn't too broad. Because then if you have your accounts that are firing on all risk, basically that means nothing to a CSM.

They're going to ignore it. And that almost is worse than having no risk CTA because the problem is that you think you have guardrails in place for catching these. And then your CSM is just ignoring it because it's flooding their inbox and causing noise. The other thing is making sure that you have a closed success reason and having specific outcomes.

Because that's going to drive down for your ROI dashboard and showing that you actually have metrics that are tied to this if you have a risk CTA and then it's going through closed. So you want to make sure that you've got your Bionic rule. You can also feed in your Salesforce data, your product data, making sure that you've got Zendesk and pulling those through. So if you've got multiple criteria that are coming into your rule, you can make sure that you don't have any details that are being missed or giving false positives.

I just want to give her a chance to take a picture because I get that. You come in late and you're like, I missed it. (laughing) So zooming out, we've got a whole engine and we've got four pillars. So you're going to detect with AI, trigger via the rule engine, intervene with the playbook, measure and report.

And each of these pillars is automatic. So the CSM's job shifts from having to dig through and find this risk to actually actioning the risk. Now, from this piece, we want to make sure that we've got full tight conditions and require the closed reason. So the two pieces that we were talking about through.

I do want to pause here and I would love to give feedback from the audience. You guys can just shout it out using your loudest voice. What are some things that you all use as triggers that would show risk in your customers? (audience member speaking off mic) Support interactions, that's a great one.

(audience member speaking off mic) Absolutely, so for those of you who couldn't hear, she said a jump in support actions or a complete drop off. Either of those could be triggers that there's a churn. Say that a little louder. Sorry.

Product usage, yes, that's a great one. So again, if you start to see a decrease, if they were kind of active and then all of a sudden you see a drop, that would actually be one. One more. (audience member speaking off mic) Yeah, that's a really good one.

So what he said is exporting out of the product and having data removals, that could trigger that maybe they're gonna be moving to another customer. They're trying to gather everything together. So that's actually a really great one. Okay, so looking at it end to end, you've got the auto detect which fires your CTA.

You've got the AI that notifies the CSM and you've got that co-pilot in there. That's super important because it's gonna give you the account contact, you've got the suggested outreach, all before the CSM even opens their account segment. Yeah. (audience member speaking off mic) Yeah, so I would put it in the suggested CTA, then that way it's really simple for them to open up and say here is exactly the actions that I need to take.

Then that way you're not having to have your CSMs go to a different place to get that action. Good question. And then also making sure that you've got the CSM reaching out within 48 hours. Making sure that you set that SLA so it's super clear across your team of when they need to take action.

Again, having that extra layer in place, if the action isn't taken within 48 hours, then you're having the automatic escalation moving up to the manager just as a catch-all. No one wants to have to be reminded and having a follow-up, so you're getting this push through the system automatically. And then that way the system's doing all the work for you. And then the last piece that I had talked about, the CTA that closes, making sure that you're logging that reason, that outcome, and then that makes sure that it feeds into your dashboard.

And you can have an ROI dashboard that then feeds into the metrics for your leadership. Okay, so I wanna give you just kind of a simple test for your CTAs when you're logging them to make sure that you've got the right intervention engine. So again, making sure that it automatically fires, you're assigning an owner, you've got specific SLAs, you've got your playbook attached, so again, somebody's not having to go find the action that they need to, and that you're again logging that outcome reason. So automatic, assign, your SLAs, playbooks, and then requiring that closed reason.

So if all five of those are true, then you're going to have a successful engine. If any of those are not true, there is a good chance that you could have a gap that slips through, and then you've got a customer that potentially churns, and we don't have the automation that people are seeing and flagging early. So now, all that doesn't work if you're not able to return the ROI to leadership. So let's talk about that.

You've got a couple metrics that you can lean on. One is the time to first response. So we talked about having that SLA. So if we're targeting under 48 hours, making sure that you're tracking when the CTA was created, and then the action that your CSM is taking within the timeline.

The other thing could be an intervention success rate. So you're looking at the accounts that were saved over all the intervention accounts that were in place, and then your saves could either be renewals or expansions when your CTA is closed in a specific timeframe. You could look at 90 days, and then whatever is important to your business, whether it's renewals or expansions. The last is the ARR retain.

So now if you're having the metrics that you can tie back to the opportunity, you can be super clear on how much dollars you're actually able to save, and then tracking the percentage, and being able to close it. So you can then report on a monthly or quarterly basis up to your leadership, depending on what that cadence is. But then it shifts the conversation of people saying, "Hey, are the CSMs being proactive enough? "Are they doing the right outreach?" But then you can come to say, "I've had 47 risk CTAs launched in the last quarter, "and we were able to solve those, have 85% SLA, "and then we were able to save $100,000 in ARR." Those are metrics that are gonna stick with your leadership, and then they stop questioning, are the CSMs being proactive?

It's like, here's a clear example, and how it's tying directly back into the business. All right, so before we go into the workshop, there's four things that I would love for you to walk away and do next week. One, audit your last 10 trends and gain site. Take a look at your timelines.

Look at the health score. Figure out where those signals are coming from first, and then were any CTAs created? From that, look at your highest volume and your pattern, and that's gonna dictate which rule you're gonna create first. The other thing too, is if you don't have gain site co-pilot turned on, at least launch that for your risk CTAs so that that is going through the flow.

And then also create an ROI dashboard, like what we were talking about for the metrics. You've got your CTAs that are being completed, the SLA that they're completed in, and then the ARR that's going through. So now I'm gonna turn it over to you guys. You have worksheets in the center of your table.

You all can work together. You can work individually, and then just kind of share what you're learning. But the idea is that you're gonna pick a customer segment, identify your top three risks, map out your rule engine, and if you're not an admin, you can kind of skip that piece or take a chance and see how it goes. And then the most important piece is you also wanna define what your success metric is gonna be.

If you're reporting back out to your boss, or just even individually how it's going, you wanna make sure that you're tying that together. So this is what the worksheet's gonna look like. You've got each of your sections to kind of go through and fill out, and then this is something that you can take away. So you can go back next week and you're sitting at your desk, and all of a sudden you start looking at all the emails and the slacks and all the things that happen while we were here that you need to get to, and then you say, "Oh yeah, by the way, "I wanna make sure that I go back "and look at my churned accounts "and see what risk actually needs to happen." Now I also understand that I am the one that's standing before you at lunch, and so you may take time to fill this out, and then you're ready to leave, or you're having great discussion, and it takes it all back.

We're gonna bring everybody back together, but if some reason you finish and you need to leave, I honestly just wanna say thank you, one, for coming to this session. I really appreciate it, and I'll walk around and answer any questions or kind of help you get through, but I don't know if we had anything come through Slido that we need to address right now. Perfect, okay, so I'll turn it over to you guys. You can work through your worksheets individually or as a group, but use the resources that we have in the room.

We have a lot of intelligent people, and I'm sure that their struggles with churn are the same, and then you can see if somebody's already set up things. This back table, almost everybody raised their hand that they had their churn, so maybe they spread out, and you can utilize them as well. (audience member speaking off mic) Oh, perfect! All right, so it sounds like you're integrating co-pilot into automation.

Is this a feature that I missed in Rule Engine? So this was something that came out of Pulse last year, and I can kind of pull up the details and share. I'm not sure who had this, so I can come visit you, and we can kind of walk through, but it does go through the automation piece. Perfect, I'll come and see you.

And then after 48 hours, you said the CTA should move to the CSM manager if not completed. Does the full CTA shift to the CSM manager? So for that, honestly, you could do either. What I prefer is having a duplicate CTA because in that way, the CSM doesn't lose it, but the other piece that you can be aware of is, somebody maybe is out on PTO, or they're at a customer visit or something, and so realizing within your business, like it's 48 hours, the right piece, when those risks come in, you definitely want to take action, but you want to make sure that you're not just flagging everything up to the manager, and then again, they get overwhelmed, and they're like, I'm not gonna pay attention to this.

Anything else? Cool. (audience member speaking off mic) No, I think, so what she asked is, is it better to alert the CTAs that were escalated into a dashboard? I think a dashboard is absolutely a perfect way, and you could tie it into the ROI dashboard, and so it just sits as a report underneath.

For my CS leadership, they also like to get pinged in their email, and so then having that CTA and it launches through, that becomes just a preference within the company and how your leadership likes to work to figure that out. (audience member speaking off mic) All right, go ahead and start working on your worksheets then. Is it on? Okay.

So we have about five minutes left and right after I got off stage somebody submitted one more question so I wanted to make sure that we addressed that. But I did have a clarification. So the question is what is the relationship between the leadership dashboard and they said a CSA KPIs and I was wondering if maybe they meant CSM or maybe I don't understand what a CSA is and making sure that you have the focus of the customer behavior. Is anybody, did somebody ask that that's still here?

Is it a CSM or a CSA? Okay, cool. Perfect. So the leadership dashboard and the like CSM KPIs.

So the ROI dashboard that I was talking about, you can tie that into a leadership dashboard. I know for our company we have pretty lengthy dashboards and so for me if it's more than two scrolls that's too much data and we start a new dashboard. So it's really your call depending on what your leadership looks at. For the ROI dashboard we only look at that like once a month once a quarter and so we make sure that we've got that piece separate.

The leadership dashboard that they use every day to kind of run their regular cadence really focuses on our KPIs. So we look at utilization, we look at decision maker engagement, CSM engagement, and then we also, now I'm drawing a blank on the extra ones, but we separate those two between the ROI dashboard and then the actual leadership dashboard. Any other questions? Cool.

Thank you again very much. I hope everybody's having a great second day and if you want to connect on LinkedIn feel free. Thank you.