Five 'Personalised' Mailers Under the Knife: A Direct Mail Audit Framework for UK Marketers
Photo: Smokeonthewater, CC BY-SA 4.0, via Wikimedia Commons
Personalisation has become the default claim of British direct mail. It appears in the agency deck, on the campaign brief and, invariably, in the post-mortem email that follows a disappointing response rate. Yet when a mailer is placed under any real scrutiny, the word usually turns out to describe a first name printed above a generic proposition.
Over the past quarter we collected five direct mail pieces received by members of our team and by a small panel of UK marketers. They span retail, charity, financial services, B2B and subscription. The senders are anonymised, the failures are not. Each was scored against a framework we use with clients, and each is instructive in a different way.
What 'personalised' should actually mean
A useful working definition: a mailing is personalised only where the data used has changed the offer, the timing or the message in a way the recipient could not otherwise have received. Printing a name is addressing, not personalisation. Segmenting by recency is a start. Triggering a piece off a specific behaviour, with a proposition that only makes sense for that person, is the real thing.
Most UK mail sits at level one and describes itself as level three. That gap is where budget leaks.
The five-point audit scorecard
We score each mailing out of 25, five points per dimension. It takes roughly ninety minutes per pack and can be run entirely in-house.
| Dimension | What good looks like | The usual failure |
|---|---|---|
| Data integrity and permission | Accurate name, salutation and address; documented lawful basis; suppression file checked | 'Dear Valued Customer'; duplicated records; no evidence of consent or legitimate interest |
| Segment logic | The variant reflects something demonstrably true about the recipient | One creative, one offer, mailed to everyone |
| Offer clarity | A single, dated, unambiguous proposition with a clear next step | Three asks competing for attention; no deadline; terms buried |
| Creative and production craft | Appropriate stock, legible typography, accessible contrast, a reason to open | Dense copy, poor paper, QR code pointing to a generic homepage |
| Measurement readiness | Unique response route, source coding, and a holdout cell | No control group, no unique URL, response attributed by guesswork |
The five mailers
Retail: the loyalty mailer that was not loyal to anyone
A fashion retailer's member mailing arrived with the recipient's name, member number and a headline promising 'exclusives chosen for you'. The hero product was the same item featured on the homepage that week, and the offer, 20% off, had been available to non-members through a paid social campaign running concurrently. Score: 14/25. The data was clean and the print quality high, but the segment logic was cosmetic and the offer undermined the loyalty premise it was built on.
Charity: excellent craft, three asks, no coding
A legacy giving pack arrived addressed to 'Dear Supporter', which is a missed opportunity given that the charity could reasonably infer long-standing support. The emotional creative was genuinely strong and the copy was well written. The problem was structural: three distinct calls to action on a single response form, no source code on the reply slip, and no way to link the mailing to the donor record. Score: 15/25. A good piece of fundraising communication that will be almost impossible to evaluate.
Financial services: strong data, buried proposition
A mortgage retention mailer was the strongest of the five on data. The recipient's product, rate expiry window and branch were all correct, and the variant logic clearly differed from a neighbouring customer's pack. The offer, however, sat beneath nine lines of regulatory small print, and the deadline appeared only in the terms. Score: 18/25. The best personalisation discipline in the sample, undone by a design decision that a compliance team probably insisted on and a marketer should have argued with.
B2B: the purchased list problem
A professional services firm mailed a 'personalised insight report' to a list that had, on inspection, been bought. Job titles were wrong in roughly a third of the sample we saw, one recipient had left the company two years earlier, and the covering letter referenced a sector the addressee did not operate in. Score: 9/25. Beyond the data failure, the mailing carries a compliance risk under UK GDPR that no response rate can justify.
Subscription: the win-back without a control
A subscription business sent a lapsed-customer reactivation pack with a genuinely compelling offer and a well-produced mailer. It scored 16/25. The reason it did not score higher is simple: no holdout group was retained, and the QR code pointed to a generic landing page with no source parameter. The business will know how many people reactivated. It will not know whether the mailer caused any of it.
The pattern: measurement is where they fall over
Four of the five mailings had no control cell. Three had no unique response mechanism. Two could not reconcile responses to the original mailing file without manual work. This is the most consistent weakness we see in UK direct mail, and it is the cheapest to fix: a holdout of 10% of the file, a unique URL or code per drop, and a pre-agreed definition of what counts as a response.
Running the audit before your next drop
Pull a sample of fifty records from the mailing file and check them by hand. Read the brief and ask which decision the data changed. Confirm there is one offer, one deadline and one action. Set the control cell before the file is signed off, not after. Brief fulfilment on response capture at the same time as you brief the printer. And agree the headline KPI in writing with whoever holds the budget.
None of this is sophisticated. All of it is routinely skipped.
The bottom line
Personalisation is a data discipline long before it is a creative one. The five packs in this audit were not badly made; several were beautifully produced. They failed because the personalisation claim was never tested against a definition, and because the measurement was designed after the drop rather than before it. Both are fixable within a single campaign cycle.
