Distribution customers almost never resign. There is no cancellation email, no awkward call. The fortnightly order becomes three-weekly. The basket quietly loses four lines to another supplier. The value drifts down over two quarters — and then one week the order simply does not come, and by the time anyone rings, the shelf space is gone. The relationship ended months before the ordering did; the ending just was not announced.
The good news is that declining customer accounts telegraph themselves, in data every distributor already holds: order records. This article sets out the signals worth watching, why each customer must be measured against their own history rather than an average, and what to actually do in the window between the first flag and the last order — because detection without a playbook is just watching churn in higher resolution.
Why the decline is invisible from the office
Three habits hide it. Revenue is reviewed in aggregate, so twenty small declines vanish inside one good new account. Attention follows noise, so the customers who ring get service and the ones going quiet get nothing — precisely backwards, since going quiet is the signal. And ordering history is often splintered across a phone log, an inbox and a rep's memory, so nobody can see one customer's full pattern in one place. The fix begins with a boring prerequisite: every order, on every channel, landing against one customer record.
Six signals worth watching
- Cadence stretch. The gap between orders lengthens against the customer's own rhythm. A shop that has ordered every 7–8 days for a year and is now at 11 is telling you something — even though 11 days would be perfectly normal for a different customer. This is often among the earliest signals, and the easiest to compute: days since last order divided by that customer's median gap.
- Basket narrowing. Total value holds, but the number of distinct lines falls. Losing the bread but keeping the milk can mean a second supplier has won part of the call — though range rationalisation, your own availability gaps or the customer's own trading can produce the same shape. Where it is a split, part-supplied customers are the natural next full loss.
- Value drift. The same lines, thinner quantities, quarter on quarter. Sometimes the customer's own trade is shrinking; sometimes yours is. Either way the account needs a different conversation than the one it is getting.
- Promotion-only ordering. An account that once bought the range and now surfaces only for deals can be signalling that you are becoming a backup or deal-only supplier. The revenue can look loyal while the relationship is not.
- Exception and dispute uptick. Shorts, refusals and invoice queries can precede defection — not because customers engineer them, but because service friction is often the reason they started shopping around. Check the exception history of every flagged account before assuming the cause is price.
- Payment behaviour change. A previously prompt account drifting to the edge of terms can signal the customer's own distress, a dispute working through, or changed payment priorities and processes at their end. Each deserves attention, for different reasons.
Before anyone phones: the false-positive check
Every signal above has innocent explanations, and a rep dispatched at a false positive spends credibility. Rule out the benign causes first:
| Signal | Benign explanations to rule out | Check first |
|---|---|---|
| Cadence stretch | Seasonality; an agreed change of delivery frequency; holiday, refit or temporary closure | Same period last year; schedule-change notes |
| Basket narrowing | Range rationalisation; a category delist; your own availability gaps | Stock-out history on the lost lines; range reviews |
| Value drift | The customer's own trade shrinking; a planned supplier split | Rep notes; the customer's footfall story |
| Payment slippage | A temporary credit hold; process or personnel change at their end | Credit-control notes; who now pays the bills |
Measure against the customer's own baseline
Averages across the customer base bury every signal above. A weekly account at 12 days is an emergency; a monthly account at 12 days is early. The workable pattern is a per-customer baseline built from that customer's trailing history — median order gap, typical line count, typical value — and a watchlist rule expressed as ratios against it. A deliberately simple starting rule, tuneable once you see your own data:
The playbook when an account flags
First, check your own performance. Pull the account's exception and dispute history before anyone phones. If the last quarter gave them two shorts and a mispriced invoice, the win-back conversation is an apology with a fix attached, and leading with a discount would miss the actual wound.
Then have a visit, not a survey. A rep call — in person where the account merits it — with the ordering history in hand: which lines went, when the rhythm changed. Customers respond differently to “we noticed and we came” than to a retention email; the noticing is itself the message.
Make the commercial response specific. If the basket narrowed, the conversation is about the lines that left and what would earn them back — range, price on those SKUs, or the service moment that pushed them elsewhere. Blanket discounts reward the wrong behaviour and teach accounts that going quiet is how you get one.
Close the loop. Record the reason discovered against the account. A quarter of these records is a defection-cause table for your operation — and the honest input for deciding whether the systemic fix is range, service reliability or price architecture.
Where RouteMagic fits
The prerequisite — one complete ordering record per customer — is what a single order pipeline provides: every channel's orders, deliveries, exceptions and payments land against one account, so the pattern exists in one place instead of five. Sales reporting puts per-customer trends — frequency, lines, value — on a standing view rather than a quarterly spreadsheet dig, and the account's visit, dispute and payment history sits alongside it in the customer record for the check-your-own-performance step. For the playbook's legwork, field sales beats put the flagged account on a rep's planned round with the history on their device — so “we noticed and we came” happens while the intervention window is still open.
Conclusion
Customers who leave without telling you often leave signals — stretched cadence, narrowing baskets, deal-only orders, a run of service friction — written in ordering data months before the last drop. Reading it takes three commitments: every order against one customer record, each account measured against its own baseline rather than the book's average, and a standing review that treats a quiet account as more urgent than a noisy one. Then the playbook: check your own service history first, visit with the numbers in hand, make the offer specific to what was actually lost, and record the cause. Start this month with a single list — every account whose days-since-last-order exceeds, say, one and a half times its own median gap (a starting threshold to tune by segment and seasonality) — and work it. The names on it are still customers. The question the list answers is for how much longer.
