You're closing new logos, and the wins channel lights up every time, but your net number isn't moving the way the bookings say it should. The customers you fought to win last year are slipping out the back while you pay full price to replace them, and nobody in the building seems to feel the loss because losing a customer doesn't make a sound the way landing one does. If that's the gap you keep staring at, this episode is for you.
I sat down with Lukas Alexander, VP of Customer Success at ChurnZero, who's spent more than a decade building global CS teams and raising net revenue retention year over year. You'll walk away knowing why your incentives are quietly pointed at the wrong thing, which retention metric actually tells you whether your team is winning, and how to put AI between your human and digital touchpoints so it helps the relationship instead of eroding it.
This post is based on Episode 66 of Revenue Rewired | You're Spending to Replace Customers You Could Have Kept (w/ Lukas Alexander).
If you'd rather listen than read, find the full episode on Apple Podcasts, YouTube, Spotify, or Amazon. It's worth your time.
Why Does New Business Get the Budget While Retention Gets Ignored?
Lukas had the clearest explanation for this I've heard. A new logo is a clean event. It wasn't there, and now it is. It shows up in the CRM report, it goes on a slide, somebody drops it in the new-customer-wins channel, and the team reacts. Retention is the opposite of that. It's the absence of a bad thing, and as Lukas put it, nobody throws a party for the churn that didn't happen. The value is real, it just never announces itself, so the attention drifts to the thing that does.
Underneath the visibility problem sits an incentive problem that's older than any of us wants to admit. AE comp has always been wired to closing a deal, and for years, CS comp wasn't tied to a dollar at all. So growth gets to feel like progress and retention gets to feel like maintenance, like keeping the lights on. There were also real business reasons to chase logos, Lukas pointed out, like a funding round where your investors want to see new enterprise names. None of that is irrational. It's just incomplete.
The reframe he offered is the part I'd write on a whiteboard for any CEO. We treat acquisition as an investment and retention as a cost center, when the math runs the other way. If your bucket is leaking, you're acquiring just to stand still, and you're paying full customer acquisition cost to do it. You can sell new logos all day, but if you're losing the same amount out the bottom, you're spending hard money to stay in one place. As a marketer, I think about this through lifetime value, which is what an average customer is worth across the whole time they stay with you. Every churned account shortens that tenure, which means you're paying the same to keep a logo for less time, and the number you thought you were building never compounds.
What Metric Should You Actually Use to Grade Retention?
If you want one number to start with, Lukas pointed straight at net revenue retention, and the reason is that NRR carries both halves of the story. Gross retention rate tells you whether you kept the logo. NRR tells you whether you kept it and grew it, because it folds in expansion and upsell at renewal. Your operating costs climb year over year as you build more product, so customers paying you the same amount they paid two years ago are quietly getting cheaper for you to keep and more expensive to serve. NRR is the number that exposes that drift.
He was firm on one dependency, though. There's no good NRR without good gross retention underneath it. You can expand a single account by a hundred thousand dollars in a quarter and feel great about it, but if you lost a hundred and twenty thousand worth of small accounts in the same window, you haven't gained anything. Retention is the foundation, and expansion is what you build on top once the floor holds. The work of the first year is moving a customer up the maturity curve so they're actually ready to buy the enterprise tier or the professional services when the renewal comes around.
Where this gets useful is when you stop looking at NRR as one company-wide figure. Pull the snapshot across the whole book, then break it down to the segment level, because the truth lives there. Enterprise accounts tend to grow at the renewal event, while SMB and mid-market customers grow mid-cycle through quicker, more transactional buys with shorter procurement. Then take it one level further and look at NRR by CSM. You'll find reps who are excellent at saving the logo, rolling up their sleeves to get a configuration right, but whose expansion number is flat. That's not a mystery to solve, it's a coaching conversation you can finally see clearly.
Where Does AI Belong in Your Retention Strategy?
Lukas gave me a mental model I keep coming back to. Picture two tracks running through your customer journey, a human track and a digital track, with AI sitting in the middle reading the engagement signals. The customer moves along the digital track until AI catches something, relationship risk, an escalation building, a usage drop, and decides it's time to pull a human in. The old way to split those tracks was crude; enterprise gets a human, and SMB gets automation, full stop. That still has its place, but if retention is the game, AI is what lets you move a customer between tracks at the right moment, rather than locking them into one tier for life.
The trap he sees teams fall into is automating the wrong things. A CS leader says the priority this year is launching a digital motion, so everything gets automated, including the customer-facing touchpoints that needed a person. At the same time, the administrative drag a CSM actually wants off their plate- building the proposal, prepping the deck, and updating contact records- stays manual. They automated the relationship and kept the busywork. It should be the reverse. I brought up the consumer version of getting this wrong: a black hole I'm sitting in right now with a major ad platform where I've been waiting six weeks for a reply, and the maddening part is that a well-placed AI handoff could have said let me get a human on the line and saved the whole relationship.
The consumer world has quietly figured out a version of this that works. Lukas described the good experience: you reach out for support, you get an honest AI-first reply that says this is AI, help me answer a few questions, and based on your answers, it either solves the problem or hands you to a person at the right moment. The win underneath it is consistency. AI follows the same signal-based logic every time, so if you map where customers hit peak frustration and cut the escalation off before that point, you're not just saving churned revenue, you're cutting the overhead of repeat callbacks too. The older systems triggered off a calendar date, a renewal email at 250 days out, where AI can finally trigger off what the customer is actually doing.
How Do You Start Fixing Retention Without Boiling the Ocean?
The advice Lukas closed with is the part I'd tape to a wall. Before you redesign anything, run a churn analysis and find out where the churn is actually happening, because the fix for a failure-to-launch problem looks nothing like the fix for a product gap or a data-integration problem. Bucketize it and attribute real dollars to each bucket. If you're aiming for 105% NRR and you're sitting at 100, and five of those points are coming from failure to launch, then you're having a conversation with your onboarding leader on Monday. Diagnose first, then go solve, instead of trying to integrate your human and digital tracks before you even know where the leak is.
He made the same point about hiring, drawn from a blog he wrote for ChurnZero on interviewing for a VP of CS or CCO role. The riskiest candidate is the one who walks in with a 90-day plan full of I'm going to do this and this and that, without spending any real time diagnosing. The strong ones sit with the data and attribute dollars to buckets before they promise a single action. I'd add the other side of that coin, which I've watched go wrong too many times. When companies finally hire that person, they need to give them room. You can't turn the Titanic in 30 days, and organizations that churn their own leaders for not moving fast enough are just deepening the hole they hired someone to climb out of.
The speed question came up, too, whether to start now or wait for the tooling to settle, and Lukas didn't hedge. Start now, because if you wait and lose, you're in a worse spot than if you'd experimented and stumbled. We were laughing in a quarterly planning meeting this week about how six months ago, AI couldn't build a usable slide to save its life, and now you can hand it a template, and it rebuilds the whole deck in your model. It only gets faster and sharper from here, which is why he frames AI fluency the way we used to think about having Microsoft Word on a resume. Table stakes. The CS leaders getting hired right now are the ones who've already rolled out an AI program, because plenty of companies haven't, and that experience is exactly what they're scanning for.
FAQ
Q: What's the difference between GRR and NRR, and which one matters more?
A: Gross retention rate measures whether you held onto the revenue you had, and net revenue retention measures whether you held it and grew it through expansion and upsell. Lukas tracks NRR above everything because it captures both, but he's clear there's no strong NRR without solid GRR underneath it, so think of gross retention as the floor and NRR as the ceiling you build on top.
Q: Is it cheaper to retain a customer than to acquire a new one?
A: That's the whole reframe of the episode. If your customer base is leaking, you're paying full customer acquisition cost just to replace revenue you already had, which means you're spending hard money to stand still. Retention isn't a cost center; it's the thing that lets your acquisition spend actually compound instead of backfill.
Q: Where should AI sit in a customer success or support workflow?
A: Picture a human track and a digital track with AI in the middle, reading engagement signals and deciding when to bring a person in. Let it run the routine, signal-based work, and trigger a human the moment it detects relationship risk or an escalation, rather than automating the relationship itself and leaving your CSMs stuck doing the admin.
Q: How do I figure out where my churn is coming from?
A: Run a churn analysis and attribute real dollars to each cause: failure to launch, a product gap, a data problem, a single-threaded relationship. Once you know that the five points of lost NRR trace back to onboarding versus product, you know exactly which leader to talk to and which fixes will move the number fastest.
Q: Should CSMs be expected to sell?
A: Increasingly, yes, and that's the skill gap Lukas flags. As AI absorbs the administrative load, the exposed question becomes whether your CSMs can multi-thread, build the executive relationship, and carry out expansion, which means sales-methodology training for post-sales teams stops being optional.
Ready to Find the Revenue Leaking Out of Your Customer Base?
At StringCan Interactive, this is the work we live in, helping mid-market sales and marketing teams find the revenue leaks they've stopped noticing and close them with a plan that fits how they actually operate. Retention is one of the loudest leaks precisely because it's so quiet, and a clear-eyed look at your NRR by segment and rep usually surfaces a number that changes the next quarter's priorities. Jay and I wrote the Revenue Rewired book on exactly this: the leaks costing you growth and how to find them.
If your team is spending to replace customers you could have kept, listen to the full conversation with Lukas and then come talk to us about what your own retention numbers are telling you. The diagnosis is usually faster than people expect, and it's almost always cheaper than the churn you're absorbing without realizing it.
