5 early warning signs a high-value customer is about to churn (and how to stop it)

By the time a customer stops buying, they've usually been drifting for weeks. Purchase frequency drops. Emails go unopened. Basket size shrinks. The signals are there, but most retail brands only notice after the customer has already left. At that point, winning them back costs significantly more than keeping them would have. According to Bain & Company research cited in Harvard Business Review, acquiring a new customer can cost anywhere from five to twenty-five times more than retaining an existing one. The math is unambiguous: intervening early is almost always cheaper than chasing customers who've already walked out the door.

Why reactive win-back is the expensive option

Most retail brands treat churn as a win-back problem. A customer goes quiet, eventually gets dropped into a re-engagement flow, and receives a discount or a "we miss you" email. Sometimes it works. Often it doesn't, and research from Lexer's retail clients shows that more than 66% of lapsed customers have already opted out of email by the time a win-back campaign reaches them, severely limiting your ability to re-engage them through your primary channel.

The better approach is to act before a customer lapses. That requires knowing which signals to watch, and having the customer data infrastructure to detect them at the individual level, not just as aggregate trends in your reporting.

1. Purchasing frequency drops below their personal baseline

Customer churn in retail is gradual, not binary. It rarely happens in a single moment, but rather happens over a series of missed purchase cycles. When a customer who used to buy every six weeks goes twelve weeks without a purchase, that gap is the first and most reliable early warning signal.

The critical word here is personal. A customer who buys twice a year lapsing at four months looks very different from a high-frequency buyer going four months without a purchase. Aggregate reporting hides this distinction entirely, which is why brands relying on overall repurchase rate metrics routinely miss individual customers who are quietly drifting.

What to do about it: Set recency thresholds at the segment level, not the brand level. For each customer value tier, define what a meaningful gap looks like given their purchase history, and build automated triggers that fire when a customer exceeds that threshold. The intervention at this point doesn't need to be a discount. A personalised "new arrivals in categories you love" email, timed to their expected purchase window, is often enough to pull an early-drifting customer back without eroding margin.

2. Email engagement declines without an unsubscribe

A customer who unsubscribes is telling you something explicitly. A customer who stays subscribed but stops opening is telling you something far more important: they haven't consciously decided to leave yet, but they're mentally checking out.

Declining email open rates on their own are a noisy signal. Inbox algorithms, Apple Mail Privacy Protection, and send-frequency changes all affect opens in ways that aren't customer behaviour. But when declining email engagement correlates with other signals, particularly extending recency and smaller basket sizes, it becomes a reliable early indicator of drift.

What to do about it: Treat declining email engagement in a high-value customer as a trigger for a channel shift, not just a subject-line refresh. If a customer has opened your last eight emails but hasn't clicked in three months, move a portion of your communication to paid social, where you can reach them based on first-party audience data rather than hoping email lands in the right tab at the right moment. Audience activation that connects your CRM data to paid channels gives you a second path to re-engagement that doesn't depend on inbox placement.

3. Average basket size shrinks without an obvious cause

A high-value customer who starts buying less per transaction is signalling a change in their relationship with your brand. They might be testing alternatives. They might be reducing discretionary spend. They might have found something they prefer at a competitor. Whatever the reason, a sustained decline in basket size, particularly when it's not explained by seasonal trends or promotional timing, is worth paying attention to.

This signal is especially telling when it appears in customers who previously bought across multiple categories. When a customer who used to buy across womenswear, accessories and footwear starts buying only accessories, that category narrowing (covered in signal four below) often starts with a basket-size decline.

What to do about it: Use basket composition data to understand what's being dropped, not just the total value. If a customer has stopped buying from a specific category, a targeted "you haven't tried our new [category]" message, referencing their past purchase history, is more likely to re-engage them than a generic promotion. PAS Group uses preference-based segmentation to do exactly this: matching customers to products based on their demonstrated affinity rather than what the brand wants to sell.

4. Category engagement narrows

A customer who shops across four or five categories is deeply embedded in your brand's range. A customer who retreats to one category is starting to disengage, and in many cases, the category they cling to last is the one where you haven't yet been replaced by a competitor.

Category narrowing is one of the clearest signals of customer drift that is routinely missed by brands without unified purchase data. If your eCommerce and in-store data live in separate systems, you won't see the full picture of what a customer is buying.

What to do about it: Cross-category purchase history is only useful if it's unified. Brands that connect their in-store POS data with their eCommerce transaction history can build a complete view of each customer's category engagement over time, and identify the moment a customer starts retreating from their full range of purchases. Identity resolution that links in-store and online transactions to a single customer profile is the foundation of this capability. Without it, you're always working with a partial picture.

Once you can see category narrowing at the individual level, the intervention is straightforward: a personalised campaign featuring product recommendations from the categories the customer has stopped exploring, based on their historical preference signals rather than generic bestseller lists.

5. Recency extends past the point of natural return

Every customer has a natural purchase cycle. Some buy every three weeks. Some buy twice a year around specific events. When a customer's recency extends meaningfully beyond their established pattern, not by days, but by weeks, you've crossed from "on their normal cycle" to "showing early churn signals."

Recency is the foundation of RFM analysis (Recency, Frequency, Monetary value) for good reason. Among the three dimensions, recency is consistently the strongest predictor of whether a customer will buy again. A customer who bought yesterday is far more likely to buy next month than one who bought six months ago, regardless of how many times they've purchased in the past or how much they've spent.

What to do about it: Tier your intervention by customer value. For your top-decile customers, extended recency warrants a more personalised and more generous response than it does for mid-value segments. This might mean a direct outreach from a CRM team member for your very best customers, a personalised product curation email for the tier below that, and a standard re-engagement campaign for the broader mid-value segment. The cost of intervention should match the value at stake.

How to tier your response

Not every at-risk customer warrants the same intervention. The cost and depth of your response should reflect the value you stand to lose.

• High-value at-risk customers deserve a personalised, multi-channel response: a specific product recommendation based on their purchase history, a channel shift to paid social if email engagement has dropped, and in some cases a direct outreach from your CRM or clienteling team. The margin on retaining a high-value customer typically justifies a meaningful offer, but make it feel personal, not promotional.
• ‍Mid-value at-risk customers warrant an automated response: a well-timed personalised email, a product recommendation based on category affinity, and a modest retention offer if they haven't responded after two touches. The automation does the heavy lifting; personalisation is driven by the data.
• ‍Lower-value at-risk customers often aren't worth a significant retention investment. A standard re-engagement flow is appropriate; if they don't respond, let them lapse rather than expending budget on customers who may not justify the cost.

Building an early-warning system

Detecting these five signals manually, at scale, across a customer base of tens of thousands, isn't realistic. An early-warning system requires three things:

• Unified data. Purchase history, email engagement, loyalty activity, and channel behaviour all need to live in the same place, tied to the same customer identity. If your data is fragmented across systems, you'll always be seeing partial signals.
• ‍Segment-level thresholds. Churn signals are only meaningful relative to a customer's own baseline. That requires segmentation that reflects individual purchase patterns, not brand averages.
• ‍Automated triggers. Once you've defined the signals and the thresholds, the system should run without manual intervention, surfacing at-risk customers automatically and feeding them into the appropriate retention flow.

The earlier you act, the cheaper it gets

Churn prevention is most effective in the window between "early drift" and "fully lapsed." Once a customer has gone genuinely cold, the cost and difficulty of re-engagement rises sharply. The five signals above give you a reliable view of that window. The question is whether your data infrastructure lets you see them in time to act.

For brands that can identify at-risk customers early and respond with a personalised, tiered intervention, the impact compounds: lower win-back costs, higher LTV from retained customers, and a retention programme that pays for itself in the margin it protects.

Book a demo to see how Lexer's customer analytics helps retail brands identify at-risk customers before they churn, and build automated retention flows that act on the signals before it's too late.