What changed · Retail
You sell through stores, so you assume you have no customer data
Almost every conversation about customer data assumes an e-commerce backend. Brands that sell through stores hear that and correctly conclude none of it applies to them, then wrongly conclude they have nothing.
20 August 2026 · 6 minute read
A brand selling through kirana, modern trade or its own outlets usually believes customer segmentation is an e-commerce exercise it cannot run. In practice most such brands already hold thousands of phone numbers, scattered across a billing system, warranty registrations, a WhatsApp business inbox, exchange and festival offer entries, and a loyalty programme somebody set up and abandoned. A phone number with a date attached is enough to know who has stopped coming, which is the single most valuable thing segmentation produces. The barrier is not data. It is that the data sits in four systems owned by three people and nobody has ever put it in one file.
Where the numbers already are
Ask around inside almost any retail-led consumer brand and these turn up. None of them was collected as marketing data, which is exactly why nobody thinks of it as marketing data.
| Source | What it gives you | Who usually owns it |
|---|---|---|
| Billing or POS at the counter | Phone number, date, amount, sometimes items | Finance or store ops |
| Warranty or product registration | Number, product bought, date | Service or customer care |
| WhatsApp business inbox | Number, and the questions people asked | Whoever answers it |
| Exchange and festival offers | Number, and proof of a purchase occasion | Trade marketing |
| Loyalty or membership | Number, and often a visit history | Usually nobody, any more |
| Service and repair records | Number, product, and a reason they contacted you | Service |
| Distributor secondary sales | Which outlets sell what, not who bought | Sales |
The last row is the one people mention first and it is the least useful for this purpose. It tells you about outlets, not about people. Useful for distribution decisions, useless for knowing whether a customer stopped coming.
What you can do with a number and a date
More than you would think, and much less than an e-commerce brand can. Being clear about both halves of that matters.
- Recency. When did this person last buy anything. This alone identifies who is drifting away, which is the most valuable group in almost every consumer business.
- Frequency. How often, if the record covers more than one purchase. Billing data usually does; warranty data usually does not.
- Value. If the amount is captured. Often it is, and often nobody has looked.
- Category. What they bought, which tells you what to say next far better than any demographic.
What you cannot do: attribute a sale to an ad, follow a browsing journey, or run the kind of tight loop an online brand can. Anyone promising you that from offline data is selling something.
The one-file exercise
Before any tool, get one spreadsheet with three columns: phone number, last purchase date, amount. Pull from every source above and de-duplicate on the number. Most brands find they have between five and fifty thousand rows and had no idea. The exercise usually takes a week of somebody chasing four colleagues, and it is the whole unlock.
Why this does not happen on its own
Because no single person is inconvenienced by the current state. Finance has what it needs for billing. Service has what it needs for warranty. The person answering WhatsApp is answering WhatsApp. The absence only hurts marketing, and marketing rarely has the standing to ask three departments for an export.
It is an organisational problem wearing a technical costume, which is why buying a customer data platform before doing the one-file exercise reliably fails. The platform arrives and there is still nothing to put in it.
The consent question, which is real
A number captured at a billing counter for a bill is not automatically a number you may market to. This varies by how it was collected and what the customer was told, and it is worth getting a straight answer from whoever advises you rather than a comfortable one from a vendor.
The practical route most brands take is to re-permission: reach out once through a channel the customer already used with you, say plainly who you are and why you have their number, and let them opt out easily. The list gets smaller and becomes worth something.
An e-commerce brand knows the moment a regular customer stops buying. A brand selling through stores usually finds out a year later, from a sales chart, when it is far too late to do anything about that person.
If you are an agency
For a retail-led client, the one-file exercise is the highest-value opening engagement available, and almost nobody offers it because it looks like data plumbing rather than marketing. It also gives you something no competitor pitching that account will have: a real number for how many contactable customers the brand actually has. Scope it as a fixed-fee two-week piece and it will pay for the relationship.
How we do this ourselves
Nothing in the platform requires an e-commerce connection. A spreadsheet of numbers and dates, a scheduled file drop, or a warehouse table all work the same way, and WhatsApp is a first-class channel for reaching the people it identifies. We do not integrate with POS systems, and if that is what you need first we will say so.
How the grouping works