Scale without Sacrifice: Scaling Digital CS for Long-Tail Retention

[Speaker 1]
Hello everyone, I hope you're having a great day too at Pulse today. Welcome to our session, Scale Without Sacrifice, as it says on the slide here. We'll be speaking on the topic of customer success for the long tail.

My name is Irvin Akhipov, I lead the SAP Digital CS PMO. I'm joined on stage or I guess on the floor by Daniel Cartagena, who leads our CS Platform Strategy and Configuration. Today we'll be sharing some insights and strategies, ideas, related to delivering a durable and scalable CS motion.

So a little bit of background about us first. We're both based in Vancouver, Canada. Weather today makes us a little homesick, reminiscent, so we've got that in common.

I've been at SAP for about 11 years across three different roles, starting in sales, helping to design the digital channels, leading to our store transaction program, using a lot of direct response channels. In the last seven years, I've been leading the digital CS programs globally. This is my fourth Pulse and my first in Europe.

Yeah, perfect. Hello, everyone. I'm Daniel.

I've been with SAP for six years, two different chapters. First one in 2016, working in sales operations for Latin America for the small and medium business segment. So it's my first interaction with the long tail.

And after two years, I joined SAS Institute, leading the sales strategy and planning for Latin America and a couple of market units from the US. This is my third Pulse, first one in Europe, first one joining as a speaker. So I'm very excited to be with you today.

[Speaker 2]
That's awesome.

[Speaker 1]
OK, so enough about us. We'd like to know a little bit about who's in the crowd today. Please raise your hand if you're a customer facing or a CSM.

What about any Ops people, CS Ops people? And CS leadership, you consider yourself a CS leadership? Oh, nice.

It's amazing. OK, anybody here have a podcast? Nobody?

OK, let's do that. OK, a little bit about what we're going to cover today. There's quite a few topics.

So we're going to start with a little bit of background and some of the challenges that are inherent with an enterprise company like SAP. Daniel is going to talk a little bit about the designing, the baseline, using our JO, how we started, where we are today. We'll talk a little bit about pooled CSM models, our feedback loop and our methodology around that.

And then we'll talk very briefly about a self-service as a channel and all the different digital properties that that includes. And then we'll wrap up with some key takeaways. And along the way, we'll also sprinkle in some maybe some thought leadership around A.I. and what we can expect to have A.I. contribute to our CS engagement. Each of these topics behind me probably could be its own TED talk. They're pretty they're pretty meaty subjects. We'll try to leave as much time as possible at the end for questions.

So anyone that wants to dig in after. OK, a little bit about SAP. You probably either recognize the logo or the name.

Founded in 1972, 55 years of history in ERP related software, most recently pivoting to the autonomous enterprise. The market cap is around 214 billion euro publicly traded company on both the New York Stock Exchange and the Frankfurt Exchange. In fiscal year 2025, we did about 36.8 billion in revenue. OK, 110,000 employees, more than 1200 products across 137 countries. And 440,000 customers. And that includes direct and indirect channels on prem and a lot of in the cloud, obviously.

OK, so these are pretty impressive data points, but they also sort of expose some inherent and incumbent challenges that we have with doing a global CS program. So first, founded in 1972, obviously a rich history, but also includes a lot of legacy processes along the way that have been carried forward with SAP, right? A lot of legacy customer data to sift through, right?

1,200 products, incredible breadth of offering, but also a lot of portfolio complexity, right? And that complexity changes with go to market updates that happen on a, you know, two or three times, you know, every couple, every decade or so. 157 countries, so very rich in terms of geographic reach.

But there's regional nuances and localizations that we have to deal with in terms of our program as well. OK, and 440,000 customers, so an incredible install base, both on prem and in the cloud, but significant number of long tail customers are represented in that 440,000, specifically 20,000 today. OK, so six years ago, we were asked to blueprint a motion to safeguard the segment of long tail customers to drive scale.

And that's kind of where our digital CS story begins. Yeah. OK, so this leads us to our first poll or Slido poll.

Approximately how many long tail customers does your organization support? So I think you have to pull out your phone and get a little bit interactive. I'll give you some time for that.

Yeah. OK, well, OK, so it's 500 to 1,000.

[Speaker 2]
It's a pretty decent number.

[Speaker 1]
Yeah, pretty decent number. It's great. Thank you.

OK, so, I mean, obviously you can define your long tail customer, however your organization wants to define it at SAP. We pretty much define it as a customer with a smaller contract value that doesn't have a dedicated CSM. OK.

OK, so why is the long tail important? I'm not going to go into detail about this. I think we saw an incredible keynote the first day from Chuck about why it's important.

It's easy to overlook one small account with a small contract value. But in the aggregate, it represents a material revenue, right, potential. It can shape market reputation and also drive future growth.

So that's why it's super important. What makes them special? Well, the smaller contract value means that you need to figure out a very efficient cost or cost to serve to maintain profitability on all those customers.

So that's really important. Also, just because they're small customers doesn't mean they're not the squeakiest customers. Sometimes we find that they're the squeakiest customers and they need the most help and are always engaging in the most way.

And they're also looking for instant gratification and improved time to value. Right. They've got smaller budgets.

They need to show some productivity and value out of those budgets. OK. So our challenge was to support this long tail segment without dedicated high touch resources.

So we needed to deploy a digital first data driven motion, one that scales, but also remains flexible to a wide variety of customer risks. So that's what we're set out to do. Perfect.

So the long tail is a treasure, right? Thousand of golden coins spread around the world. The best way to collect them with an scalable approach, right?

In our case, digital program that supports thousands of accounts while improving, of course, customer experience. Oh, sorry. This is a little waggy.

So how do we start? We knew that we had an amazing opportunity in front of us. Right.

But first, we have to identify the long tail. In our case, as I've been just mentioned, customers without a dedicated CSM below on a specific contract value. Right.

This group was very diverse. So the question was, what does every customer need regardless of industry, region, segment or solution? And this became our guiding principle.

So we identified a handful of universal moments that matters so we can create a consistent baseline for every customer. Right. And we have some examples here.

We have a welcome emails and go like celebration and post-renewal touchpoint. And after testing and learning from the golden path, the next question was, do all customers follow the golden path? And the answer was no.

Some customers show risk signals and other shows like growth potential. So we included two new paths, risk and opportunity path. So these paths leverage dynamic signal from our system.

So we deliver content based on those signals. For the risk path, we have implementation delays or low consumption. For the opportunity path, we have rapid adoption or high consumption.

So some examples are implementation check, touchpoints, overconsumption, touchpoint and customer advocacy. So with this approach, we were able to respond to customer needs earlier. So instead of like waiting for customers to tell us that there was a problem, we were able to use a real time risk indicator to support our customers.

So with these three paths established, we move from the concept to the execution, right? So as we added more and customer scenarios, more solutions and languages, we covered nine languages with our touchpoints. We ended up with twenty six thousand templates.

[Speaker 2]
Yeah. Sorry. Twenty six hundred.

[Speaker 1]
Twenty six hundred. A lot, right? So it means one person configuring ten templates per day during an entire year.

Right. So it created a lot of operational challenges. So the three main are time consuming, of course, for configuration, but also for updates.

Imagine that you need to change a broken link. You need to go and replace like template by template. Also, greater risk of human errors because you are doing repetitive tasks.

And of course, very difficult to scale to more solution and products. So we step back to understand how can we leverage on a scalable approach for journey configuration? And the answer was tokenization and data designers.

So we knew that we found out that tokens were more powerful than just for basic personalization, such as adding maybe a customer name or end user name. Right. And we extended the concept much further.

By tokenizing product context. So we have names and also the context in general. We also tokenize content blocks.

So we use those those tokens to for like intro paragraphs, for instance. We also tokenize resources and links. Very important because we track the consumption of the link in our platform so we can see the bottlenecks and the entire journey of our customers.

And of course, for a delivery strategy. So we use different sender names and different fallback options. Also, we have signatures and so on.

So that was the first step. So we were able to create customized templates for like every scenario. Right.

Second step, we need to identify the right people, the right audience. Right. Because not all solutions have the same consumption or adoption patterns or implementation timeless.

Right. So we leverage data designer to group customers. So, for instance, 50 percent consumption for product A maybe is good in the first year, but maybe before product B is is not good.

Right. So we have different group of customers. But even with the content, but even with the right triggers, we face the challenge of customer qualifying for more than one campaign.

So we have a low usage campaign. Our customer owns maybe five, six products with low consumption. We don't want to send five emails at the same time.

Right. Because it's not a good experience. Even with you have like different content, like the right content.

It's not a good experience. So we also use data designers to stagger touchpoint delivery so that customers have time to consume and understand our communications. And with these three components, so triggers, content, and the right delivery sequence, we were able to dynamically assemble multiple experiences without having to create like separate versions for each one.

Right. So the impact was magnificent. So from 26 to 100, we optimized to 189.

It means 93 percent in the number of templates. But yeah, it's a lot. But we also tackle the three main challenges.

Right. So time consumption, much better. We reduce human error.

So better customer experience and content. And we also were able to expand to more solutions. Right.

So with this new journey management, we were able to explore more new signals, new risk scenarios. We were able now to work at the product level. That is very important for us.

And we also explore like two different business needs. We two new post sales and categories. So the Evergreen journey is the golden risk and opportunity path.

It's always on in perpetuity for our customers. On demand touchpoints, very popular. So it's one time engagement to a broad audience.

Right. And the objective here is to increase and promote the adoption and consumption of our solutions. So some examples are like adoption webinars or new capabilities or feature activation and so on.

Right. And we also have gains that assist. This one is for initiatives that don't qualify for an on demand or an evergreen.

So we use the reusable template and the CSMs can trigger those campaigns and we have the interaction inside. At the end, the main objective is that we want to centralize all the customer engagement inside so we can gain customer intelligence and we can use this information to improve like our program, the customer customer experience. And with these three components and with the with the digital program established, we achieved some results.

So first, we expand customer reach by almost four times compared to 2022. This is very important. We also covered 90 percent of the solution portfolio.

This is even more important because we are now covering at the product level. We started with solution. Now we are covering products.

And last but not least, we are we have engaged with 87 percent of the entire population of long tail customers. So we are sending a consistent like message to the long tail base. But good results.

Right. But we know that we can achieve even more with AI. We have two main opportunities.

The first one, of course, is the program creation. So we want if we have, for instance, a campaign for a customer with high consumption in with a goal like they did in the last year. We want the AI to generate the audience, the success metrics, the email templates, call to actions and contacts.

We want basically to reduce operational effort. On the other hand, we have adaptive journey personalization. So currently we use dynamic signals, but it's not the same as having a dynamic journey.

So the idea here is that the AI can adjust the next course of action based on customer needs and industry context. So we have same product to different industries. 50 percent usage for for customer A.

Maybe it could be a feature activation for a specific industry, but for customer B could be something very different. So we want the customers to move through a nonlinear journey. And the main idea is to move from a push approach like us sending information to a pull approach.

So we inspire our customers and to share information with us. Right. Things like success plans or business goals.

This like information is really hard to collect, especially from the long tail. So with this, like with a detailed program established, we have also some scenarios where the customers reach out to us. And this is very important because the human component takes more like relevance.

Awesome, Daniel. Very good. OK, next slide, Opal.

Who handles inbound queries from long tail customers at your organization? OK, I should see a support, whoever is available. Other would be interesting to see.

Multitasking. Multitasking. 15 seconds.

I'm just trying to balance out a little bit. OK, I've received some sales or support. Whoever is available.

That's pretty awesome. OK, 6% on an inbound team. Thank you, operator.

OK, so as Daniel mentioned, we had a very effective J.O. program, right? It sent a lot of post-sale communications. But emails are one directional.

And they don't necessarily inspire immediate action, especially for risk. So we needed a fallback for customers that needed additional help or had questions that they wanted to reach out to for some support from a person. And that fallback channel needed to handle 20,000 customers potentially.

So we introduced a pooled CSM model. Put your hand up if you've heard of a pooled CSM model or if you've used a pooled CSM model. OK, so just to level set, a pooled CSM model is a pooled collection of individuals with similar roles or functionality.

For this example, we used CSMs. When there's an initiated task, it gets assigned to a pool. It gets picked up by an individual in that pool. And when that individual completes the task, they return to the pool for the next task.

So essentially, the pool operates as a shared service. No one person is responsible for any one specific customer. And it operates as a discrete resource, essentially.

You can think about it as like a fleet of taxis or Ubers. If you call one, you know you need to get from A to B. And they'll provide that service.

And when the car shows up, that's the service you'll get. But you won't necessarily get the same driver every single time. And maybe you don't want always the same driver.

So done properly, this offers a great deal of flexibility and value. If you need more capacity, add a resource to the pool. If you need less capacity, take it out of that pool and put it in a different pool.

It's a very flexible model. OK, so we use this to help address the inbound that we could potentially get from 20,000 customers. We deployed a simple pooled team model to respond to those inbound queries that were initiated from our email touchpoints.

And we assembled those pools by region. So we put one pool in one of the four main regions you see here. It was North America, LAC, EMEA, and APAC.

And that allowed us to respond to customer queries in the preferred language of the customer. And we had that data already, so that was one of our criteria. Each region had its own specific mailbox.

And the inbound queries were routed to that corresponding mailbox, leveraging the token strategy that Daniel had spoken about earlier in the JO design. When the pooled CSM would pick up that query, they'd resolve it, and they would reply using that same mailbox or an alias to maintain that discrete relationship with the customer. And that was super important to make sure that we could maintain the integrity of our pooled CSM model.

And once resolved, that CSM would then record that event to the customer timeline in Gainsight and pick up the next inbound query essentially. So that really represented the reactive approach that we had to inbound queries from our customers. It was straightforward.

The workflow was we sent out a touchpoint. Customer would use a touchpoint to get in touch with our pooled CSM. The CSM would address the query and then make an entry in Gainsight.

So after optimizing this, we started to ask the question, why don't we use the same model in a proactive way maybe to address risk potentially for those customers? So that's kind of what we did. We already had the risk signals identified in the JO design that Daniel had spoken about earlier.

We created simplified mitigation playbooks that we embedded within CTAs to address those risks. And in parallel, we used existing Gainsight functionality to create pools that mirrored the pools that we had previously for our reactive model. So when a risk CTA was triggered, it was assigned to the pool.

That task would be assigned in a round-robin fashion to an individual within that pool. And then that pooled CSM would be alerted to the CTA, complete the task and the playbook. And then once it was closed, it was automatically recorded to the timeline of that customer.

And that all happened in-app in Gainsight, which was pretty awesome. That was one of the huge benefits. The other benefit that we realized was because everything was being done and initiated by the app itself within Gainsight, we gained a lot of real-time visibility into the data.

So we could tell in real time how many CTAs were initiated, what region were they allocated to, what was the mean time to close those CTAs, how many were still open versus closed. So these were all really great benefits that we were able to harvest from this model. So after working with both of the reactive and proactive models, we became very affluent and started to really ideate on what we can do with this pooled configuration.

And what are all the additional possibilities that we can do with this pooled model. So it didn't have to just be regional. We could create pools based on product expertise and specialization potentially.

We could potentially create pools based on customer stage, whether it be onboarding, renewal, consumption, or commercial conversations. We could even do pools assigned to industry or use case. So it became very flexible.

And the interesting thing is that we started thinking, you know, we could even do something like a nested pool or an embedded pool. So you could have a regional pool. And within that regional pool, you could have an industry pool or a customer stage pool.

So it became very flexible. And the imagination was the only thing limiting us to being able to deploy this. The other thing too is that the pools didn't have to have exclusive resources.

You could share resources across multiple pools. And that was also a key benefit, especially when you wanted to adjust capacity across the entire model. Okay.

So I'm talking about some AI opportunities with pooled models. Typically, what would happen is we get an inbound request from the field to the CSM in that pool. They'd have to do a lot of legwork and administrative tasks to figure out what the background of that customer was.

I'll get out of the way for anybody who's taking pictures. What the task was, what the footprint is, what was the history, some more context, all that stuff. It was quite a bit of an administrative drag.

But with AI, you can do all that in the background and then create a recommended course of action that's served to the CSM to deliver to the customer. It cuts down on a lot of that manual process. So that's a huge opportunity.

Okay. The other thing too is a little bit more simple. It's a simplified asset generation.

So think about our long-tail customers and how many there are. A lot of them could use a QBR light in some format that we just can't do because it's really hard to scale. But with AI, you can create these really simple QBRs, right?

Pulling in all the relevant context across that customer footprint or history or whatever. And either deliver it to a CSM to deliver to the customer or publish it to a common workspace. Maybe you have gain site spaces and you can publish it there and then the customer can follow up with additional questions related to that QBR.

We'll do questions at the end. Go ahead. Go ahead.

I just quickly want to check about the creating. So when you say creating, you don't think you're touring? We're talking about value conversations, capabilities, solution, product features, all that kind of stuff.

Okay. Cool. So to keep getting better, we need to understand if we are engaging with the right people at the right time with the right content, right?

So that's why we have our engagement, our feedback loops, so we can improve like our processes and and of course like the customer experience. Perfect. Okay so we have a pooled CSM model and a digital program established.

We were able to capture feedback from multiple perspectives. So from a internal feedback we have of course the pool team. They manage all the queries and also we have observation from other stakeholders.

From external feedback from our customers we use the gain site service in our touch points. So we ask like simple questions. Is the information relevant for you?

Was it easy to access? Was at the right time? Is information aligned with your business goals?

Disclaimer we don't ask about like product feedback is just about a touchpoint value. So with this information we turn all this data into insights in a centralized dashboard. So we can track product consumption, journey progression, customer interactions and campaign performance.

In three main areas. First one is a touchpoint effectiveness. Second one is operational performance.

And the third one is a business outcomes. So it's basically an attribution model. So with this information has helped us understand what interactions are working well, what interactions are not working well.

So we can improve our triggers, our content, basically like improve our program in general, right? So and then processes start again, right? So we engage, evaluate and enhance.

This is our feedback process. Yeah, that was great. And this is part of the DNA of our program since day one.

It was always in perpetuity to refine, evaluate and iterate. Okay, last but not least, Slido. Thank you for being around.

Thank you. For routine postal questions, how do your long-tail customers prefer to engage? Email, shared inbox.

We have other up there but then we thought maybe Teams or Slack could also fit in there. So if it's Teams or Slack, you can put it in the other. One-on-one.

Okay, it's getting close. It's like a horse race. Let's see.

Okay, let's leave it there. So one-on-one and an email shared box. That's interesting.

Okay, so we're going to quickly touch on self-service here. So we built a robust JO program. We talked about that.

We talked about a scalable pooled CSM model. We then asked another question. What if some customers were motivated to do things on their own and we could give them more autonomy over their experience?

So case in point, here's some crowd engagement for you guys. Who here does their own banking? Come on.

I know everyone kind of does their own banking. Who here likes to engage with the teller or the CSR at the bank when they have to do a routine transaction? Not me, not me.

Yeah, me neither. Like nobody has their hand up. So that's pretty indicative, right?

It doesn't make sense if you can do it yourself in a fraction of the time. It makes sense, right? So we learned that many long-tail customers also felt the same way.

They want easy options to help themselves whenever that's possible. Okay. So just quickly going to touch on a couple of these self-serve properties, digital properties.

Oops. Is this the right one? This isn't the right one.

I think our slides are in the wrong order. Okay. Maybe we'll go backwards then.

Digital communities. So actually, I'll just go off the slide here. Customer portal.

It's a very important self-serve property for customers. This facilitates some really, really important autonomy over their own information. The biggest one and the best example that we can provide is customer contact data.

Customer emails change, project teams change. Having to maintain that and upkeep it for enterprise company is very, very difficult, right? And it takes a lot of time.

Imagine 20,000 long-tail customers that are having attrition in their project teams and having to maintain that sort of data. And if we don't maintain that data, they don't get the right information to the right people, or we're not able to deliver a property. Okay.

The other things that this customer portal can do, and at SAP, we call this SAP for me. It's a digital companion. That's the way it's branded.

But there's other things that you can do in there, like you can track project milestones. The customer can have visibility into their own consumption data. Customers can also update their goals that are maybe evolving or push back implementation dates.

And they're by themselves without needing a CSM to go in and manually update them. Okay. Second one is learning platform.

So I think we all kind of know what a learning platform is. It's, you know, at SAP, it's called a learning hub. It could be SkillJar in Gainsight as a customer can go and access learning journeys or get enabled on certain things.

They can access registration for webinars or consumed existing archived webinars, right? Stuff like that. They can even get accreditation or complete official knowledge, get those official knowledge badges that we always see as part of the program.

And also, we promote all, like, all those platforms in our touchpoints. Absolutely. Yeah.

Especially the customer portal. That messaging is included in a lot of our touchpoints as it goes out. Okay.

And last but not least, my notes are not changing, so I'll just mention them. The digital community. That's where customers can really go and extract an incredible amount of data and knowledge sharing from peers.

Peers that are either in the same region or have the same use case or in the same industry, right? They can create those relationships that are really important from other people that have gone through similar circumstances or similar implementations instead of relying on a CSM, and they can self-help on all of that information, right? So all three of these things are foundational as properties, right?

They all do things very well, enable, engage, empower the customers, okay? The problem is they're all very siloed, right? And those isolated signals are not being aggregated, okay?

And customer success is evolving into a more connected, holistic experience. These properties are becoming something much more valuable, a rich source of customer intelligence. So think about the signals that these are hiding in plain sight.

A customer relearning or revisiting enablement content, a customer repeatedly asking the same question in the digital community, or a customer updating their value drivers, or delaying a key implementation date in their customer portal. So again, individually, these may look like isolated interactions, but collectively, they tell us a much bigger story about emerging risk. And this could fundamentally change how we approach customer success from reacting to risk to anticipating it, right?

But who has the time to connect all those dots, all those data points across all those platforms? Well, huge opportunity for AI, right? With AI, the potential exists to aggregate signal patterns across all the digital pattern, all the digital properties, to distinguish meaningful signals from the noise, to attribute the most important signals to the most significant risks, and recommend CTAs or risk levers personalized to the specific customer or cohort of customers.

And even further, apply those same tactics to similar customer profiles or lookalike models to get ahead of risk in the future, okay? We could effectively identify and address risk before it even materializes. That's really powerful.

Okay, so let's just wrap with a couple of key takeaways. Build a golden path around the moments that matter most. First point, start simple.

Add risk and opportunity journeys based on customer signals. Leverage the data you have today. Scale through automation and personalization.

Make it resonate. Make the content resonate with the customer. Complement digital engagement with pooled human resistance.

Offer a fallback channel when the customer needs it. Create feedback loops, as Daniel had mentioned, to continually improve the experience, to set it. There is no set it and forget it, you know, plan here.

It's always reiterating. You always have to adapt and try to make things better. Invest in self-service capabilities for customers who prefer to help themselves.

Don't be the bottleneck for your customer's success, okay? And be open to AI to enhance the experience for all customers, right? AI will bring an incredible amount of value to CS and allow us to do things that we can never thought possible.

And on a final note, if you can do these things effectively to retain your long-tail customers, you can probably do most of them to help all your customers. Something to think about moving forward. So that concludes our canned portion of the content.

We'd be happy to take some questions with the moments we have left.

[Speaker 2]
All right. Thank you, gentlemen. Oh, I like this one.

Okay, top question with the most likes. Do you impersonate real CSMs or approach digital in a neutral, human-less way? In either case, or if you're doing a mix of both, why did you choose that approach?

[Speaker 1]
It's a great question. We try to make our interactions and our engagements as human as possible, okay? A lot of our touch points that we send to our customers are sent with the tone of a human.

We don't try to mask the fact that it's digital. So we try to strike a balance between this is an automated engagement, but at the same time, it's also got a human tone to it. Yeah, I know.

We also use tokens to personalize with the signatures and...

[Speaker 2]
So you do have CSM names sometimes?

[Speaker 1]
Yeah, emails. If you have questions, you can reach out to... Yeah.

Got it.

[Speaker 2]
Perfect. So, second question. Are CSMs support both long tail and mid-high touch?

I feel for them, whoever that is. So we don't have the head count to recruit and not in a position to automate. Thoughts on how named CSMs don't get sucked into too much time on needy customers?

I'm guessing needy customers in the long tail.

[Speaker 1]
Yeah, we're specific to long tail, so this is probably more relevant to named CSMs, but I think it all comes down to standardized playbooks and setting expectations ahead of time. Yeah.

[Speaker 2]
They have to know how to engage with each customer and then which customer falls into which bucket, just as somebody who's in that role. Which signal most reliably triggers human intervention?

[Speaker 1]
I think we track links. So if a customer is interested in a specific resource, so we trigger also some post communication. So this is one of them.

Yeah. I would say one of the most reliable triggers where we start to engage in a human-led way is when a customer is out of compliance. So it's an overusage situation and we need to get them back in compliance.

We can't just let them be out of compliance in perpetuity. It's also an opportunity for a potential upsell. So a human is more inclined to get involved in that type of process.

And also, when we see a lack of engagement as well from an under-consumption scenario, we also want to make sure that we support them and reach out and figure out what are the reasons why they might be in that scenario as well. So there's a few. I would say those are probably the two biggest ones.

[Speaker 2]
All right. How large is the team that is building and maintaining these journeys and what are the various roles you have? So tell us about the team.

[Speaker 1]
So we started out with a very, very small team. I'm pleased to announce that we've been growing, you know, in a consistent way. We still feel the bottleneck of all of the activities that have to happen.

But the size of the team, I don't know, what do we have? Eight or nine? Yeah, yeah.

Content. Across content, configuration, yeah, some PMO duties, intelligence, and data scraping, yeah. So I'd say eight, nine, or 10 at most, yeah.

We hope to scale that up a little bit to get more productivity.

[Speaker 2]
All right. I think you answered the next one, that you spend, for the most part, ARR as the criteria for long tail?

[Speaker 1]
Right, yes. Okay.

[Speaker 2]
How can you transition or just advice on transitioning customers to a new self-service or digital segment without making them feel forgotten or left behind?

[Speaker 1]
This is a great question, actually. We've had to do this, right, in the past. And it's not about losing a CSM, it's about introducing them to a more real-time, adaptive, hyper-personalized experience with a CSM potentially as a fallback option.

So we don't position it as, sorry, you've lost your CSM, you're not going to be part of our automated journey. We've introduced this automated journey for all our customers. This is what you can count on in the future, right?

So it's a bit of a soft skill, you've got to kind of wordsmith it, but that's the concept.

[Speaker 2]
Yeah, that's a perpetual challenge.

[Speaker 1]
It is. Yeah, yeah.

[Speaker 2]
I love this next question about contact data because...

[Speaker 1]
Mm-hmm. Right?

[Speaker 2]
And we say if anyone can solve that problem, they're well on their way to buying a desert island, or tropical island, rather. Are customers abiding by the expectation that they should be the ones maintaining their contact data, or have you faced challenges there?

[Speaker 1]
Like I said, this is one of our biggest challenges for long-tail customers and for high-touch customers. And for high-touch customers, it's a learning process, right, to have them go and self-serve, but also it's a learning process for our CSMs to maintain in perpetuity on a regular cycle their contact data because they know who they talk to on a daily basis, but they also need to respect the fact that the system needs to be updated with the right contacts so that we can provide a digital experience accurately as well. Also, for the long-tail, we have CTAs, so we use bounces. When we see that the customer, like, it's a soft bounce, so we have a CTA to...

That's a great point. Yeah. We contact the customer and they update the...

Or we update the... One of the really cool things we do is when we get a bounce from an email that we've delivered, it comes back with a code. We've assigned that code to a certain CTA.

Depending on what the code reveals, and then that's the follow-up that our inbound team will take to it.

[Speaker 2]
That's smart.

[Speaker 1]
Yeah.

[Speaker 2]
I like that. Okay, last question, just due to time. I'm guessing that your percentage of churn in this segment has a different expectation than in your higher-touch segments, so how do you define that?

Is there a certain number that you're... Or how do you agree on what the number is for the percentage of churn in this long-tail segment?

[Speaker 1]
Well, typically, we look at historical data, and we try to improve on that every single year, right? We're finding we're getting better numbers on churn, so the churn is actually... I don't want to say officially slowing down, but based on the way that our customers are segmented today, we're finding that our long-tail is actually performing quite well.

Okay? What is our... What do we consider a certain percentage of churn?

We would like to see 0% churn. I mean, yeah, of course.

[Speaker 2]
Is it feasible?

[Speaker 1]
Probably not, but I think we're doing a pretty good job. Yeah, yeah.

[Speaker 2]
Well, and just kind of as a follow-up, because I know your CSM would want me to ask you this, are you able to tie your digital strategies and what you're doing, anything specific, do correlation analysis to what is helping reduce that churn?

[Speaker 1]
Yeah, this is why we mean by creating an attribution model. So we want to see the influence of our touchpoints in the increase of consumption or maybe revenue. But this is in the roadmap for sure.

[Speaker 2]
Awesome. Love that. Thank you so much.

I appreciate your session. That was great. I'm sure I will be sending this to some of my customers that are not here.

Everyone's always interested in this topic. So thank you, everyone.