Notes for owners · Channel and customer growth
How to identify customers at risk of leaving
The signals appear months before the notice does: falling order frequency, a changed contact, complaints that stop, and a renewal nobody has discussed.
The GullySales team · Updated 15 Sept 2026 · 6 min read
On this page
The customer who leaves almost never surprises you in hindsight. Order frequency drops, and the person you dealt with moves without anybody introducing you to the replacement. Complaints stop being raised because they stopped expecting a fix, your calls go to voicemail, and the renewal date arrives with nobody having discussed it. Identifying risk is a matter of tracking a handful of signals on every account, reviewing them monthly, and giving one named person the job of acting when the pattern appears. None of it requires software you do not already have.
The signals worth tracking
Nine signals cover most Indian SME situations, and they fall into three groups.
Buying behaviour. Order value or frequency falling against the same period last year. A product line they used to buy every month that they now buy occasionally. Payment slowing for no stated reason, which often means your invoice moved down their priority list.
Relationship behaviour. Your main contact has changed roles or left. Nobody attends the review you schedule. Requests now come from a junior person rather than the decision maker. Your emails are answered in one line, days later.
Service behaviour. Complaints that spiked and were resolved slowly. Complaints that stopped entirely from a previously vocal account. An open issue older than thirty days. A request for a rate comparison or a fresh quotation in a year when nothing changed, which usually means somebody else has already quoted.
A scoring sheet you can build in a spreadsheet
| Signal | Green | Amber | Red |
|---|---|---|---|
| Order frequency against last year | Same or higher | Down up to 25 per cent | Down more than 25 per cent, or nothing in two cycles |
| Range bought | Same or wider | One line dropped | Reduced to a single line |
| Payment behaviour | On terms | Slipping by two weeks | Consistently late, or disputed |
| Main contact | In place and responsive | Changed, introduction made | Changed, no introduction |
| Review meetings | Held as agreed | Postponed once | Not held in two cycles |
| Open service issues | None over 30 days | One | More than one, or one over 60 days |
| Complaint pattern | Raised and resolved | Rising | Silent after a period of complaints |
| Renewal or contract date | Discussed, more than 90 days away | Inside 90 days, not discussed | Inside 30 days, not discussed |
| Competitor activity | None known | Asked for a comparison | A competitor has quoted or visited |
Two reds, or four ambers, and the account goes on the list for a visit. Keep the sheet to one row per customer so a person can read it in five minutes.
A worked example
For example, imagine a facility management firm in Bengaluru with forty contracts across tech parks and apartment complexes. The details and figures below are illustrative. Its highest-value contract looks healthy on paper: the invoice is paid, the manpower is deployed, no complaint has been raised in three months.
The sheet tells a different story. The admin head who signed the contract moved to another building in January and nobody from the firm has met the replacement. The quarterly review was postponed twice. Two housekeeping staff were replaced without informing the client, which is how the previous three contracts were lost. The renewal is eleven weeks away and has not been mentioned.
That account is red on four signals with no complaint in sight. An illustrative ₹2.4 lakh a month contract is sitting on a decision nobody from the firm has been part of. The remedy is a meeting with the new admin head in the next fortnight, taking the service record and a proposal for the coming year. That has to happen well before the procurement team starts collecting three quotations.
What to do when an account turns amber
Act on the specific signal rather than on the score. A changed contact needs an introduction meeting, not a discount. Falling order frequency needs a conversation about what changed at their end, which is often their own demand rather than your service. An old open issue needs closing, with a call from someone senior, because the reason it is old is that nobody owned it.
Three rules make the difference. Meet in person where the account justifies the travel, because a video call does not surface the thing they were reluctant to say. Bring your own service data, since arriving empty handed puts the customer in the position of having to explain your business to you. Do not lead with price, because you cannot recover a relationship problem with a rate revision, and the offer tells them their bargaining position is stronger than they thought.
Build the habit into the month
Fifteen minutes at the end of each month to update the sheet. One hour a month for the retention review, in which every red and every amber account has a named owner and a dated next action. One quarterly question in the management meeting: which customers did we lose, what was the first signal, and how many months before the exit did it appear.
The last question is what improves the sheet over time. After a year most businesses find that two signals predict almost every loss, and they can then watch those two closely.
What to do next
List your top twenty customers by revenue and score them against the nine signals this week. The exercise takes an afternoon and normally surfaces two accounts nobody had worried about. If you want the sheet built into your CRM and the review made a monthly routine, tell us what you need.
