How a Mobile Repair Business Reactivated 2,000 Dormant Customers in 30 Days
A transparent mobile-repair reactivation case study: how to segment dormant customers, sequence contact, protect permission and measure what really returned.
Advanced AI frontier lab and business growth agency. Helping UK businesses deploy agentic AI systems.

- 01The starting problem
- 02Step 1: make the audience explainable
- 03Step 3: sequence contact without creating pressure
- 04What the case study measures
- 05What another repair business should copy
This case study describes a 30-day reactivation exercise built around a mobile-repair customer database of 2,000 dormant records. It is a worked example of the method and the decisions, not a guarantee that another repair business will recover the same number of customers or revenue.
The important lesson is not the headline number. It is the discipline required before a message goes out: understand what each record means, choose a relevant repair or upgrade reason, separate channels and record objections.
The starting problem
The business had a familiar leak: customers who had previously paid for a repair were difficult to see once the original job was closed. The database contained names, contact routes and service history, but “dormant” mixed together several different situations:
- a customer whose device might fail again;
- someone who had bought a replacement elsewhere;
- a one-off visitor with no ongoing need;
- a record with an unclear permission or contact preference;
- a customer who had already asked not to be contacted.
Treating all 2,000 as one audience would have made the campaign easier to schedule and harder to defend.
Step 1: make the audience explainable
The first working table used fields the business could verify:
| Field | Why it matters |
|---|---|
| Last service or purchase | Helps identify a plausible future need |
| Device or service category | Prevents irrelevant messages |
| Recency band | Separates recent customers from genuinely old records |
| Preferred channel and permission | Determines whether contact is allowed |
| Objection or suppression | Stops the next send |
| Customer status | Separates active, unresolved and closed relationships |
The ICO's electronic-mail guidance says marketing emails, texts and other stored electronic messages to individuals need consent or the limited own-customer soft opt-in. A phone number in a repair database is not a universal licence to send a campaign.
Step 2: choose a reason to return
The campaign did not begin with “we have a promotion”. It grouped customers around a reason the business could explain:
- a likely maintenance or repair question;
- a complementary service related to the original job;
- a seasonal or practical device problem;
- a simple invitation to ask whether help was needed.
If a segment could not support a relevant reason, it stayed out of the first cohort. That is a feature, not a failure.
Step 3: sequence contact without creating pressure
The 30-day plan used a deliberate order:
- Preparation: reconcile permission, deduplicate records and remove objections.
- First message: email or the recorded preferred channel, with one clear question.
- Follow-up: only for eligible non-responders and only when it added information.
- Conversation route: give the customer a way to reply to a person rather than forcing a booking form.
- Suppression: record a purchase, objection, complaint or invalid address immediately.
SMS or WhatsApp were not automatic “next touches”. They were separate channel decisions with their own permission and operational owner.
What the case study measures
The useful result table is wider than revenue:
| Measure | Question |
|---|---|
| Eligible records | How many could lawfully and meaningfully be contacted? |
| Delivered messages | Did the channel and data work? |
| Replies | Did the reason to return make sense? |
| Qualified repair conversations | Was there a real service need? |
| Completed jobs | What happened after the conversation? |
| Margin after delivery and staff time | Was the work commercially useful? |
| Opt-outs and complaints | Did the campaign damage trust? |
The published case headline uses 2,000 records and 30 days. It should not be read as independent research, a typical benchmark or a promise. A repair business considering the method should substitute its own database, permissions, capacity and margin.
What changed operationally
The lasting improvement was visibility. The business could now see which records were:
- eligible for a defined message;
- waiting for a human answer;
- suppressed permanently or temporarily;
- linked to a service history;
- ready for a later, different reason to return.
That is more durable than one campaign. A reactivation programme should leave the database clearer than it found it.
What another repair business should copy
- 01Define one segment
- 02Check permission
- 03Choose a reason
- 04Route replies
- 05Measure margin
- 06Suppress objections
- Start with one service category and a named owner.
- Write the eligibility rule before writing the message.
- Make the customer reason specific and honest.
- Give every reply a human route and a response target.
- Stop after an objection, complaint or completed job.
- Measure completed work and margin, not only clicks.
- Keep the case-study result separate from your forecast.
For the legal and channel detail, read email, SMS or WhatsApp reactivation. For the broader strategic choice, read database reactivation vs cold outreach.
FAQ
Can a smaller repair database use the same approach?
Yes, if the segment is explainable and the team can handle replies. Smaller volume does not remove the need for permission and suppression checks.
Is a dormant repair customer automatically eligible?
No. Recheck the original notice, consent or soft-opt-in conditions, product similarity and objections before contacting the person.
Do I need email, SMS and WhatsApp together?
No. Choose the channel the customer is eligible to receive and the team can support. Add another only for a clear reason.
What if the case result does not repeat?
That is normal. Use the case as a method, then measure your own eligible audience, response quality, completed work, margin and complaints.
If the database is full of “maybe” records and you need a safe first segment, Get unstuck.