Attribution for Direct Mail and QR Codes Without Fooling Yourself

Direct mail is measurable now, but QR scans and matchback reports overclaim easily. How to attribute mail honestly and still learn what it is worth.

Direct mail used to be the channel you defended with faith. Now every mailer carries a QR code, a personal URL, or a matchback report from the vendor, and the faith has been replaced by numbers that feel precise and often are not.

QR scans undercount, because many people type the address or search the brand instead. Matchback overcounts, because it credits mail for orders from households that would have bought anyway. Honest attribution for mail lives between those two errors, and it is reachable.

Use three lenses, not one number

Start with direct response: QR scans, personal URLs, and mail only offer codes. That is your floor, the orders mail certainly touched. Then look at matchback against your mailing list, knowing it is a ceiling that includes orders mail probably did not cause.

The third lens is the one that settles arguments: a holdout. Mail most of your list, hold back a random slice, and compare order rates. The gap between mailed and held out households is the closest thing to mail's true lift, and it usually sits between the floor and the ceiling.

Design the mailer for measurement

Give each drop its own code and QR target so drops can be compared, and keep the offer consistent when you are testing the mail itself rather than the discount. A deeper discount in one drop will happily take credit for what the discount did.

Time your reads properly. Mail response stretches over weeks, with a long tail as pieces sit on counters. Judging a drop at seven days because that suits a dashboard rhythm understates mail and overstates every faster channel next to it.

Handle QR data with care

QR scans tell you about the scanners, who skew toward comfortable phone users responding quickly. Plenty of recipients visit later by search or direct type in, and last click systems will hand those orders to search or direct without ever mentioning the mailer that prompted them.

Use scans for creative and timing insight, which offer got scanned, which panel worked, rather than as the sales total. For the sales question, go back to the holdout.

Report the range and the method

Stakeholders trust a channel more when its reporting admits method. Report mail as a range with the holdout lift as the headline, state the response window, and note what matchback includes. That honesty also protects the budget when a vendor's report claims heroic numbers.

Keep each drop's setup in a log: audience, offer, codes, window, and holdout size. Six months later, that log is what lets you compare drops fairly instead of relitigating memory.

Related reading

If this topic is useful, these earlier guides go deeper on connected questions: Attribution Windows Explained: What the Setting Really Does, Attribution for Podcast Advertising Without Guesswork, and Attribution for Subscription Renewals and Reorders.

Frequently asked questions

Why do QR scans undercount direct mail response?

Because scanning is only one way to respond. Many recipients type the web address, search the brand, or order days later from a different device. Those orders came from the mailer but never touch the QR code, so scan counts are a floor, not a total.

What is matchback and why does it overclaim?

Matchback matches orders to households on the mailing list and credits the mail. It counts every order from a mailed household in the window, including purchases that would have happened without the mail, so it works as a ceiling rather than a true result.

How big should a mail holdout be?

Big enough to compare order rates reliably, often five to ten percent of the list for larger mailings. It must be random, not your least engaged names, or the comparison will flatter the mail.

How long should we wait before judging a mail drop?

Allow several weeks, since mail response has a long tail. Reporting too early understates mail systematically. Pick a window, such as thirty to forty five days, and use it consistently across drops.