I’ve been reading too many blogs and listening to too many podcasts. You know how I can tell? I just wrote a provocative, click-baity headline for this one. 🙂 Sorry about that.

Because my headline implies that Customer Success needs saving, let’s start there. Despite well-known (and respected) people saying that Customer Success is failing or, at the very least, needs a reset, I strongly disagree. There is absolutely no indication that customers are willing or able to fully adopt complex B2B software without significant assistance. And that’s why Customer Success began in the first place and is still as necessary as ever. Sure, the retention numbers haven’t looked great for the past couple of years but that’s mostly about the economy, not the usefulness of Customer Success. In fact, if we’re going to call for the elimination of organizations based on the last two years, we’d be firing every single Sales Rep at every tech company in the world. Let’s not do that. And let’s not count Customer Success out when it remains just as important as it’s ever been.

To answer my own question in the headline, “NO!” Customer Success does not need saving and, even if it did, AI would not be the savior. However, if we change the question to “Can AI Help Customer Success?” the answer is a resounding “YES!” In fact, as most of you know, it is already doing so. More on that later.

Once Customer Success found its footing in the mid 2010s, one question started to bubble up to the top of the industry’s mind. “How do we scale this thing?” The answer to the “scale” question has always been technology and Customer Success is no exception. We’ve been trying to figure out how to best use the existing set of tools to accomplish this goal – webinars, email, in-product assistance, etc. – with varying degrees of success. The reality is, AI included, there are no magic wands. Even the best use of digital tools provides only incremental improvements despite our sincere hopes for more. But, to be fair, even the very best companies have just scratched the surface of the gains available through better use of all the one-to-many vehicles available to them. I firmly believe that we will have significant advances in this arena over the next few years and then we’ll add AI to the mix. I’m not great at predicting the future but I think AI’s role for now, is not to replace all of those electronic methodologies, but to improve their effectiveness and continue to strengthen the associated productivity gains.

In the meantime, the hype around AI continues to be both astronomical and growing. But any doubts about whether it will provide real productivity gains have been quickly dismissed. It’s already having an impact in even the simplest use cases. Because so much of the early hype was inspired by ChatGPT, generative AI was the primary focus. And that focus is already paying off in one simple way – meeting summaries. Every customer-facing role includes lots of time in meetings with customers. The advent and improvement of both CRM solutions and Customer Success solutions has created the repository for all customer-related information. That includes notes about any, and every, meeting. And because the vast majority of those meetings are now done remotely and recorded, some pretty standard AI techniques can be applied as a solution to this challenge. Although it’s a pretty straightforward application of AI (which is NOT just useful for Customer Success), the value is significant. Just imagine the time spent, even on a relatively small CSM team, documenting every one of their customer meetings. Let’s say there are 10 CSMs on my team and they average six customer meetings a day. That’s 300 meetings a week. First off, many of those meeting summaries, without AI, would either not be done at all or be pretty sloppy. But let’s say every CSM takes this task seriously. If so, writing up a good summary of each meeting probably takes an average of 20 minutes. If my math is correct and we remove all that effort using AI, that’s a savings of 100 hours! In one week! Even if you think my assumptions are crazy, you’re still going to come up with a pretty big number when you do your own calculations.

Let’s look at another example of what AI can do today. This one falls under the heading “Sentiment Analysis”. Again, this is pretty straightforward work for an AI engine. Think about the volume of conversational data you have as a result of customer/CSM interactions. For most companies this will include:

  • Lots of Gong/Zoom recordings
  • Thousands of email exchanges
  • The text portion of Customer Support cases
  • Any open text response to your surveys

That’s a LOT of data which makes it perfect for AI. The way most of us currently analyze sentiment is through NPS surveys and/or CSM opinion. The recommended frequency of NPS surveys is no more than twice a year and the average response rate on email surveys is 6-8%. Not only that but the responses are often heavily biased based on recent events. Ever sent out an NPS survey the day after a new release went out with a couple of key bugs in it? Bet your scores weren’t quite what you hoped for. Anyway, the point is that robust sentiment analysis on the terabytes of customer conversations happening every single day is unquestionably going to be far superior than whatever you get out of NPS. And we all know that all the objective data in the world doesn’t remove the need to hear exactly what the customer is saying. And that can help you save a couple of unexpected churns a year. I’ll let you do your own calculation on what that’s worth to you.

But the real question being asked by many is not whether AI can help our CSMs but whether AI can BE a CSM. This is obviously a much more complex challenge and the simple answer is “not yet”. In my opinion, probably not ever. Think about all the technology we’ve seen come along over the past 20 years and whether that has CREATED or REPLACED more jobs? Clearly, the answer is “created”. We’ve made some pretty solid progress in using technology to scale our efforts but, what they’ve mostly done is to remove some of the mundane tasks our CSMs are responsible for, freeing them to do more strategic activities.  Our two scenarios above (meeting summaries and sentiment analysis) are good examples of this. No CSMs are being replaced by this functionality but time is being saved and customer insights are being improved thus increasing both effectiveness and efficiency.

Having said that, I have recently seen AI playing the role of a CSM in one particular use case – completing a Success Plan. That’s an AI (with a personality by the way) interacting directly with a customer to accomplish a relatively complex task. And, in my experience, a task that most CSMs do not find very fulfilling (which means they don’t do it as well as we’d wish). This is clearly just the tip of an iceberg but live interaction with customers is a big leap from even the most intelligent email campaign. Not to mention that, in some ways, the AI is better than a CSM because it can literally “know” everything there is to know about your product, your company, and the customer’s company, too (at least what’s available publicly). If you’re like me, it’s not hard to imagine a number of other tasks around which we can draw some concrete completion criteria and also relatively easily train AI to complete. The future application of this technology is virtually limitless. Oops, there I go jumping on the hype train. But it’s hard not to get excited about this given what is already available and knowing that we’re just getting started.

I’m an advisor and member of the Board of Directors at Intelligage (formerly CompleteCSM). All of the above scenarios I’ve described are available in their current product which is being used by both Customer Success and Sales teams at a variety of companies. Check out our website for more details and to get a pilot started, even have your first discussion with a Digital CSM!

Meet with Katie our first Digital Customer Success Manager – https://intelligage.io/meetwithkatie/