Back to Blog Home Page
AI adoption is dominating conversations across every industry—and customer success (CX) is no exception. CX leaders are feeling real pressure from boards, investors, and executives to “do more with AI,” driven by promises of efficiency, productivity, and cost savings. But in a function built around people, AI for AI’s sake isn’t the answer.
Customer success is a people business. While AI can surface patterns and signals faster than humans ever could, it doesn’t understand motivation, context, or what success actually looks like for a customer. Humans do. And that distinction matters more than ever as CS teams navigate how—and where—to apply AI.
That’s why I recently joined Erin Mills and Ken Roden on the FutureCraft GTM podcast to talk about AI in customer success. We explored the question most CX leaders are wrestling with right now: what can AI realistically take on, and where do humans remain irreplaceable? Along the way, we got candid about where CX fits into the revenue organization, what happens when you rely on lagging metrics, and the risks of tiering your customer experience.
In this post, I’m pulling out a few of the most important takeaways—and sharing how I think CX leaders can use AI to drive real outcomes without losing the human element that makes customer success, well, successful.
You can also listen to the full conversation on the FutureCraft GTM podcast.
Customer Success Starts Earlier Than You Think
I’m a strong believer that customer success shouldn’t begin after a deal is signed—it should start while the deal is being shaped. Especially with larger, more complex customers, I like to bring someone from CX into the sales cycle at the technical win stage, or when a deal reaches roughly 70% probability, and we’re the vendor of choice.
There’s two big reasons for this.
First, accountability starts with alignment. If customer success is going to be responsible for delivering outcomes, we should be part of defining what we’re actually committing to deliver. Too often, starts before anyone even gets out of the gate—when implementation goes sideways, expectations aren’t aligned, or success was never clearly scoped in the first place. In those cases, by the time CX inherits the relationship, the damage is already done.
Second, human partnership is a differentiator. In a market where technology is increasingly similar, the experience around the technology matters more than ever. What differentiates one vendor from another is human interaction: how teams show up, collaborate, and build trust. Bringing CX into the conversation early gives customers a real sense of that partnership—and often strengthens the deal rather than slowing it down.
I’ll admit, sometimes this takes convincing. Sales teams worry about losing velocity or overcomplicating the process. But once they see the impact—clearer expectations, stronger relationships, smoother handoffs—they don’t want to go back. Technology absolutely plays a role in making customers successful. But humans do too. And customer success sits right at that intersection.
Optimize for Leading Signals, Not Lagging Signals
One of the most common mistakes I see in customer success is over-optimizing for lagging metrics like NRR, churn, and NPS. They matter—but they all tell you the same thing: what already happened. By the time a customer tells you they’re going to churn, you’re already months behind being able to make a real difference. At that point, the only option left is offering a discount. And if discounts are your retention strategy, you’ve already missed the opportunity to deliver value when it mattered.
The real leverage in customer success comes from leading indicators—the signals that tell you whether a customer is on track before outcomes are at risk. The challenge is that those signals aren’t universal. At some companies, heavy engagement with support means customers are invested and trying to make the product work. At others, it signals frustration and broken workflows. The metric alone isn’t the insight. Context is.
And that’s exactly why customer success can’t rely on automation alone. AI can surface patterns—like a spike in support interactions—but it can’t interpret the context behind them or decide what to do next. Only humans can connect those signals to real customer goals, understand intent, and take the right action. In customer success, that human judgment is what turns data into decisions—and signals into success.
AI Still Isn’t Personal—And That’s Exactly Why Humans Matter
Back in 2019, I remember being genuinely excited about where AI might take us—especially the idea of true personalization. Not just smarter segmentation, but experiences that actually reflect individual needs, context, and intent. Fast forward to today, with generative AI everywhere, and I still don’t think we’re quite there yet.
A recent example made that clear for me. I asked a friend whether I should buy a jumpsuit, and she suggested asking ChatGPT—sending a photo, measurements, and details about the outfit. For context, I’m 5’2″. The response? Buy it, but plan to get it tailored. Helpful advice, but also a perfect illustration of the gap between segmentation and personalization. The model essentially concluded: you’re short, here’s what short people should do.
That’s the pattern I keep seeing. AI is incredibly good at bucketing—grouping people by traits, behaviors, or attributes—and serving up a reasonable response. But even now, it struggles with the things that actually make decisions personal, like:
- Context: what’s happening around the decision, including constraints, timing, and competing priorities that don’t show up in data alone.
- Motivation: why someone is choosing one path over another, and what success actually means to them.
- Tradeoffs: what someone is willing to give up—or accept—in order to move forward.
In customer success, that distinction matters deeply. Relationships aren’t built on categories—they’re built on understanding. Knowing what a customer is doing is different from knowing why they’re doing it. AI can help scale insights and surface patterns, but true personalization—the kind that builds trust and drives long-term success—still depends on humans who can interpret nuance and act with judgment.
Design Customer Success Around Choice, Not Tiers
We can’t talk about customer success without talking about how customers actually want to interact with you—and that can change depending on the moment, the task, and what’s at stake. Too often, though, we default to a familiar model: enterprise customers get a high-touch, white-glove experience, while self-service is reserved for smaller customers.
That’s not how I think about it. Self-service shouldn’t be treated as a lower tier of support—it’s just as critical as a white-glove experience. Most of us would rather find a quick answer in a doc or video than schedule a meeting. Not because we don’t value human interaction, but because we don’t always need it. The digital experiences you build should be top-tier and work just as well for any customer, regardless of size.
Of course, there’s a line. The question isn’t whether customer success should be digital or human—it’s when each makes sense. Here’s how I encourage CX teams to think about that balance:
- Where should digital, self-service be the default? If a document, video, or in-app workflow can solve the problem quickly, that’s often the best experience.
- Where does a human need to step in—and why? The moment digital interactions stop being helpful, customers want a person. Recognizing that shift in real time is critical.
- What signals should trigger human intervention? Missed milestones, stalled progress, or changes in engagement are often early signs that value is at risk.
The teams that get this right design customer success around choice. They use digital experiences to remove friction—and humans to handle complexity—meeting customers where they are, instead of forcing them into a predefined tier or experience.
Use AI for the Work You Dread, Protect the Work That Makes You Valuable
When it comes to AI adoption, one of the hardest parts is simply figuring out where it actually belongs. Sure, AI promises efficiency, productivity, cost savings, and better decision-making—but if you apply it in the wrong places, you won’t see the impact you’re hoping for.
So, I start with a simple question: What parts of my day do I not enjoy—or just aren’t my strength? I don’t love formatting slides, polishing bullets, or rewriting the same emails over and over. AI is great at those things. What it’s not great at is the work that actually defines customer success—interpreting what’s happening with a customer, deciding what it means, and guiding them toward the right outcome.
That’s why I encourage teams to be intentional about what they delegate to AI—and what they protect as human work:
- Delegate the tedious and repetitive: drafting routine emails, summarizing notes, formatting decks, and pulling together reports.
- Protect the work that requires judgment: understanding customer goals, building trust, interpreting analytics, and deciding how to act on what the data shows.
When teams get this balance right, AI becomes a force multiplier instead of a distraction. It takes the work people dread off their plates—without touching the work that actually drives customer success.
Customer Success Can’t Be Automated—and That’s the Point
AI is changing how customer success teams work, but it doesn’t change what customer success is for. At its core, CX has always been about helping customers achieve outcomes that matter. Technology can surface signals, remove friction, and create efficiency. But understanding context, navigating tradeoffs, and guiding customers toward real success still requires humans.
The opportunity for CX leaders isn’t to adopt AI everywhere—it’s to use it intentionally. To design experiences around choice, not tiers. To focus on the signals that matter before outcomes are at risk. And to protect the human work that makes customer success meaningful. When we get that balance right, AI doesn’t replace customer success—it makes it stronger.
