Customers who have not bought from you in months: spotting it in time
In September everyone looks ahead: the next campaign, the quotes, the new customers. But there is one list almost nobody opens that is usually worth more than any campaign: the customers you already have who have not bought from you in months.
The mental image we have of a lost customer is someone who gets angry, complains and leaves. In practice that rarely happens. What happens far more often is that the customer simply stops calling. They tried another supplier for an urgent job, it went fine, and they stayed. Or the person who placed the orders changed and the new one does not know you. Or you just did not come to mind. There is no grievance to resolve: there is a silence nobody noticed.
Why we do not notice
Because no small business has an alarm for things that stop happening. If a complaint comes in, you see it. If a customer stops buying, nothing arrives: no email, no call, no noise. And since the month's turnover comes out roughly as usual — because the other customers make up for it — the leak is not noticed until a year has passed.
There is a second, more prosaic reason: the information is scattered. Orders are in the invoicing program, conversations in the salesperson's chat app, incidents in the inbox. Nobody has the full picture, and assembling it by hand every month is a job that always gets left for next week.
The first step is not technical: decide what "silence" means
Before thinking about any tool, answer one question: how often does a normal customer buy from you? The answer varies a lot by business.
- For a distributor that delivers every week, three weeks without an order is already a signal.
- For a maintenance workshop, the signal is not having come in for the annual service due in March.
- For an accountancy firm, it may be a client who has not responded to two campaigns in a row.
Write your threshold down in one sentence. Without it, any automation will give you a useless list: it will either alert you about everyone or about no one.
Turning it into a list that fills itself
Once you have the threshold, what you need is not a pretty report: you need someone or something to put five names in front of you every Monday. Five, not two hundred. Nobody calls a list of two hundred names.
This is where a simple automation does a lot of work: cross the date of the last order with the last contact, apply your threshold, and send you a weekly email with the cases that are due, sorted by what they used to bring in. No need to change invoicing software or build anything large.
A concrete example. A catering-supplies company serves around 120 restaurants. In March they reviewed which customers had gone more than two months without ordering: 19 came up. Of those, 11 had not gone anywhere — they had changed chef or purchasing manager and the contact had been lost in a chat thread. With two calls a week for a month they recovered 7. No discount, no campaign: just noticing in time.
What the automation must never do
This is where most projects go wrong. Three rules:
- Do not send automated "we miss you" emails. They can be spotted a mile off and they burn the customer you could still have recovered. The automation detects; a person calls.
- Do not offer a discount by reflex. Before cutting the price, it is worth knowing why they left. Often it was not the price.
- Do not confuse "not buying" with "does not want to hear from you". If someone asked not to be contacted, that must be marked and they stay out, always.
And your customers' data?
We are talking about purchase history and contacts: sensitive information about your business and about third parties. It is worth setting conditions from day one. Data stored inside the European Union. Not used to train third-party models. Each person on the team sees only what they need. And deleting a contact is a one-click action, because the GDPR is not a recommendation.
If you cannot answer where the data is and who has access to it, it is not a tool you should use with your customer base.
This is what a contact-tracking tool does. As an example, MemorIA — which I built — gathers the contacts that reach you by form, chat and email, keeps their status, and sends a weekly summary of who has gone too long without a reply. The design is written up on the project page.
This article is general information, not legal advice for a specific case.
About this article
iaetica.cat is a space for writing in the open: what I am learning about artificial intelligence, its ethics and its law.
— Laura Graells