Sign in

Blog · Coverage and territory

Lost customer review template: the timeline from the data, before anyone explains what happened

A template for reviewing a lost or sharply declining customer: build the timeline from the ledger, the activity log, the service record and the contract file first, then hold the conversation. This page gives the timeline table, the eight questions the data can answer before the meeting, the three that only people can, how to classify the cause, how to turn one review into a rule that finds the next account early, and a copyable one-page form.

The short answerReview a lost customer by building the timeline from data before anyone tells the story. Lay out, month by month for the eighteen months before the loss: revenue and order frequency, categories or products bought, touches by the account owner, service issues, price changes, contact changes at the customer, and contract dates. The data answers eight questions on its own, including when ordering first fell below the customer's own pattern and when the owner last spoke to them. Only then ask the three questions people must answer: what the customer says happened, what we knew and when, and what we would do differently. Classify the cause, then write the rule that would have flagged the account at the first signal and run it across the current base. The review is worth doing only if it produces that list.

A large customer has gone, or halved. The usual review is a meeting in which the account owner explains. This one starts with a table nobody has to explain.

The timeline

Eighteen months before the loss, one row per month, from exports the business already has.

Month Revenue Orders Categories bought Owner touches Service issues Price or terms change Customer contact change Contract
−18 $52,000 9 6 2 0
−12 $50,000 8 6 1 2 late deliveries
−11 $49,000 8 6 0 1 credit dispute, open 40 days
−9 $44,000 7 5 0 +4% list increase
−8 $41,000 6 5 0 Buyer replaced
−6 $30,000 4 4 1
−4 $18,000 3 3 1 Notice window opens
−2 $6,000 1 1 3 Discount offered
0 $0 0 0 2 Not renewed

Read down the columns. Service issues at month −12 and −11; no touch from the owner for four months after; a price increase into that silence; a new buyer who inherited a grievance and no relationship; a category dropped at −9; and the owner active again only at −2, with a discount. The discount was offered eight months after the first signal.

Eight questions the data answers

  1. When did order frequency first fall below the customer's own pattern? Month −9.
  2. When was the first category or product dropped, and which? Month −9; consumables.
  3. When did the account owner last speak to the customer before the decline began? Month −12.
  4. How long was the longest gap in contact, against what the tier is owed? Four months; owed monthly.
  5. Were there service failures in the prior year, and how long did they stay open? Three; one open for 40 days.
  6. Did a price or terms change precede the decline? Yes, by one month.
  7. Did a contact change at the customer? Yes, month −8. Did we meet the new contact? First logged meeting, month −6.
  8. How many contacts did we have? One, until month −6.

None of these needs anyone's recollection.

Three questions for people

What does the customer say? Ask them; someone other than the owner. They will usually talk. Compare with the timeline.

What did we know, and when? Was the credit dispute escalated? Did anyone know the buyer had changed?

What would we do differently? Specific and procedural, not a resolution to try harder.

Classifying the cause

Cause Signs in the timeline Preventable?
Neglect Long contact gaps on a valuable account; decline follows silence Yes
Service failure Issues cluster before the decline; slow resolution Yes
Price or terms Decline follows a change closely; categories with alternatives go first Partly
Contact change New buyer; single-threaded before Partly: multi-threading
Competitor action Sudden, not gradual; one category, then all Sometimes
Customer circumstances Site closed, acquired, product line ended; decline matches their business No
Poor fit from the start Never reached the pattern of similar customers At qualification

Most losses have two causes, a trigger and a vulnerability. Here: service failure and price increase were the triggers; neglect and a single contact were why they were fatal. Record both.

From one review to a rule

The review pays for itself only here. Write the earliest signal as a rule and run it on the whole base.

Finding in this review Rule Accounts flagged today
Category dropped nine months before loss Tier A or B account that stopped buying a category it bought in each of the prior four quarters 14
No owner contact for 120 days on a monthly cadence Tier A account, no two-way touch in 60 days 9
Service issue open 40 days, no follow-up Any account over $100,000 with an issue open more than 14 days 6
New buyer not met for two months Contact change logged; no meeting with new contact in 21 days 4
Price increase into an unresolved complaint Accounts with an open issue excluded from the increase until resolved Process change

Thirty-three accounts that look today like this customer looked nine months before it left. That list is the deliverable. The dormancy worked example and the churn versus dormancy comparison cover the first of these signals in depth.

A copyable one-page form

Customer: [name]. Revenue, prior 12 months: [value]. Tier: . Owner: . Outcome: lost / declined by [x]%. Date: .

Timeline attached: 18 months; revenue, orders, categories, touches, issues, price, contacts, contract.

From the data: First fall below own pattern: . First category dropped: . Last owner contact before decline: . Longest contact gap: . Service issues, longest open: . Price change before decline: . Contact change; days to first meeting: . Contacts known: .

From people: Customer's account: . What we knew and when: . What we would do differently: .

Cause: trigger ; vulnerability . Rule written: . Accounts flagged by the rule today: . Owner of that list: .

Where it goes wrong

Story first. The competitor's price, accepted, and nothing learned.

Review of the final month. The discount that failed, not the silence that made it necessary.

Blame. Owners stop logging honestly, and the next timeline is empty.

No rule. An interesting meeting; the next loss looks identical.

Only the big ones. Patterns among mid-sized losses, where most of the money goes, never examined.

The short version

Timeline first, from the data. Eight questions it answers, three that people must. A trigger and a vulnerability. Then a rule, run across the base, producing a list with an owner. For losses in businesses with no contracts, see B2B churn without a subscription. Covirage builds the timeline for any account from the ledger, activity and service exports, and runs the resulting rules across every account each week.

Questions people ask

Why build the timeline before the conversation?

Because the first explanation offered is nearly always about something outside the business: a competitor's price, a change of buyer, the economy. The timeline usually shows the account went untouched for four months, or a service problem three quarters earlier, before any of that. Starting from the data keeps the review from becoming a defence.

Which customers deserve a review?

Any loss or decline of more than half among the top tier, and a sample of the rest. Five a quarter is plenty. The aim is not to explain every loss but to find the two or three patterns that recur, since each pattern becomes a rule that protects the accounts you still have.

How early was the first signal, typically?

In most reviewed B2B losses, the first measurable signal, order frequency slipping, a category dropped, usage falling, a contact leaving, appears four to nine months before the revenue is gone. That interval is the opportunity. The review exists to find what the signal was for this customer and whether anyone was in a position to see it.