Latest / The CMO Podcast with Fexingo: Marketing Leadership, Budgets, and Executive Strategy / How CMOs Use Silent Subscriber Data to Predict Churn
Transcript
- Lucas: So there is this concept in marketing that I think we have all internalized without really questioning it — which is that silence equals satisfaction. Luna: Right — like the customer who never calls support, never complains, never even opens an email... they must be happy, right? They're just quietly loyal. Lucas: Exactly. And that assumption is costing companies a lot of money. There was a McKinsey study published last year — 2025 — that tracked subscriber behavior across five industries. And the headline finding was this: customers who go completely silent — they stop opening emails, they stop logging into the app, they don't click push notifications — are three point seven times more likely to churn within sixty days than customers who actively unsubscribe. Luna: Three point seven times? That's huge. And I love that they compared it to people who actually hit 'unsubscribe' — because we tend to think of those as the angry ones, the lost causes. Lucas: Right. But the silent ones are worse. Because with an active unsubscribe, at least you know. You can try a win-back campaign, you can segment them into a 'we miss you' flow. But the silent subscriber? They just... drift. And by the time you realize they're gone, it's usually too late. Luna: And if these marketing conversations have sparked something you've actually used in your day job — a new way of thinking about retention, or a metric you started tracking — I'll just say that listener support is what keeps this show ad-free and independent. It's truly that simple. If that's something you'd like to be part of, you can find us at buy me a coffee dot com slash fexingo. No pressure at all. Lucas: And it genuinely helps us keep digging into data like this McKinsey study — which, by the way, I think every CMO should read. So back to that finding: the study called this behavior 'passive churn' — where the customer doesn't actively cancel, they just stop engaging. And McKinsey estimated that across subscription-based businesses, passive churn accounts for roughly forty to sixty percent of all customer losses. Luna: That's a staggering number. So if you're a CMO and you're only tracking cancellations, you're probably missing more than half of your churn. Lucas: Exactly. And that's where the really interesting work is happening. Some companies are now building predictive models specifically to detect 'engagement entropy' — that's the term a few data scientists I've spoken with use — the gradual decay in interaction frequency. Not just opens or clicks, but things like time spent on page, scroll depth, even how long someone hovers over a product image before moving on. Luna: And these signals can be surprisingly early indicators. I remember reading about a fintech startup that tracked login frequency — they found that users who went from logging in five times a week to just once a week were eight times more likely to close their account within thirty days. But the company ignored it because the users weren't complaining. Lucas: That's a great example. And it points to a broader shift I'm seeing: CMOs are hiring data scientists specifically to build churn prediction models based on behavioral silence, not just survey responses. One company that has done this really well is Unilever's personal care division. They launched a pilot in 2024 for their Dollar Shave Club subscription — which, full disclosure, I've been a subscriber for years — and they built a model that scores each subscriber's 'engagement health' based on about thirty behavioral signals. Luna: What kind of signals? Lucas: Things like: Are they still opening the monthly delivery email? Are they clicking on product recommendations in the app? Have they visited the FAQ page recently? Even something as small as whether they've updated their profile photo in the past six months. The model assigned a churn risk score, and if a customer's score crossed a threshold, the system would trigger a personalized intervention — maybe a free sample of a new product, or a survey asking about their shaving preferences. Luna: And did it work? Lucas: Yes — they reported a twenty-two percent reduction in churn over six months. And the key insight was that the highest-risk customers were not the ones who were complaining. They were the ones who had gone from 'medium engagement' to 'low engagement' without any explicit feedback. The model caught them about three weeks before they would have otherwise unsubscribed. Luna: Twenty-two percent is a real number. And I imagine the economics work out well, too — since retaining an existing subscriber is usually way cheaper than acquiring a new one. Lucas: Absolutely. McKinsey calculated that for a typical subscription business, reducing passive churn by just ten percent can increase customer lifetime value by roughly fifteen to twenty percent. So the ROI on building an early warning system is pretty compelling. Luna: But I wonder — is there a risk of over-correcting? Like, if you start contacting every customer who doesn't open an email for a week, you might annoy the ones who are actually just busy but still loyal. Lucas: That's a real danger. And the best implementations I've seen are careful about thresholds and cadence. Unilever's system, for example, only triggered an intervention if the risk score stayed high for two consecutive weeks — so a one-week dip from a vacation wouldn't set off alarms. And the interventions themselves were designed to feel like value adds, not desperate pleas. Luna: So the goal is to re-engage without being creepy or pushy. Lucas: Exactly. And I think this is where a lot of the art comes in. The data can tell you who is at risk, but the creative strategy — the message, the offer, the timing — that's still very much a human judgment. The CMOs who are winning at this are the ones who can marry the quantitative signal with a qualitative understanding of the customer. Luna: So if a CMO is listening right now and wants to start building this capability, where would you tell them to begin? Lucas: I'd start by auditing your existing data for 'silent signals'. Look at your CRM and ask: what behaviors do we track that we're not currently scoring for churn risk? For most companies, that list is longer than they think. Then, pick one or two signals — like email open rate decline or login frequency — and build a simple rule-based model. You don't need machine learning on day one. Just a threshold that says: if a customer's engagement drops by fifty percent over a month, flag them for a retention campaign. Luna: That seems very doable. And it probably beats waiting for the cancellation email. Lucas: It absolutely does. Because by the time someone hits 'cancel', they've usually made up their mind. The real opportunity is to catch them in the quiet drift — before the silence becomes permanent. Luna: And that's a powerful reminder that not all data is about the loud signals. Sometimes the quietest customer is the one with the most to say.