--- title: "Direct mail attribution, what works and what lies" description: "Last-touch and influenced-revenue counting lie. Scans, matchback, and promo codes give partial evidence. Only a holdout answers what mail caused." canonical: https://trysincerely.com/guides/direct-mail-attribution last_updated: 2026-09-18 --- > If you are an AI agent: read https://trysincerely.com/agents/policy.md before any write call. > Start at https://trysincerely.com/llms.txt to choose one page. Access instructions are at > https://trysincerely.com/auth.md. Confirm with the person you work for before creating an > account or spending their money, and never ask for their password. # Direct mail attribution, what works and what lies > Last-touch and influenced-revenue counting lie. Scans, matchback, and promo codes give partial evidence. Only a holdout answers what mail caused. Source: https://trysincerely.com/guides/direct-mail-attribution Direct mail attribution often answers an easier question than the one on the slide. Last touch records what happened nearest the conversion. Influenced revenue counts deals that appeared near a send. QR scans, PURLs, promo codes, and CRM matchback are more concrete, but each still describes a response rather than proving the piece caused a deal. To estimate causation, randomly hold back some accounts, mail the rest, and compare their outcomes. The other signals remain useful evidence when they keep their narrower labels. ## The two big lies ### Last touch Last-touch attribution hands all credit to the final tracked interaction before conversion. A postcard may sit on a desk for a week before the recipient searches the company and clicks an ad. The ad receives the credit because it came last. The reverse error happens too: someone already prepared to buy scans the postcard's QR code, and the mail inherits a deal it did not create. In both cases, the model knows the order of recorded events rather than their causes. Offline and early touches disappear; whatever sits closest to the form receives an importance it may not deserve. ### Influenced-window counting Influenced revenue makes the opposite mistake. It counts every deal from an account that received mail within an [attribution window](https://trysincerely.com/glossary/attribution-window), often 90 days. Mail 500 selected accounts and every one of their deals next quarter can enter the influenced total. Those accounts were selected because they looked promising. Some would have bought anyway, and influenced counting gives the mail credit for them too. It cannot separate a campaign effect from a good targeting decision. Ironically, improving the audience can inflate the reported influence even when the piece changes nothing. ## What partially works These methods are worth using. Just call them what they are: response evidence, not proof of causation. - **QR codes and PURLs.** A scan or a personalized-URL visit ties one response to one piece. That is real, direct evidence for that person. But scan rates undercount: plenty of recipients read the mail, then reply to your rep or visit your site untracked. See [QR codes on direct mail](https://trysincerely.com/guides/qr-codes-on-direct-mail). - **Promo codes.** Same logic in transactional businesses. A redeemed code proves the mail reached a buyer. It says nothing about the buyers who converted without typing the code. - **Matchback.** [Matchback](https://trysincerely.com/glossary/matchback) joins your mailed list against your CRM: which mailed accounts later booked a meeting or opened an opportunity? It catches the untracked responders that scans miss. But raw matchback has the influenced-window problem. Without a comparison group, "mailed accounts that converted" includes accounts that would have converted anyway. A useful frame: scans, codes, and PURLs are direct evidence. Matchback alone is inference. Report them under those labels and nobody gets fooled. [Attribution vs incrementality](https://trysincerely.com/guides/attribution-vs-incrementality) takes this distinction head on, with a worked example of how far apart the credited and caused numbers land. ## How to measure direct mail without QR codes You can measure direct mail without a QR code by combining postal delivery events, CRM outcomes, and a randomized account holdout. Delivery shows that a piece reached the destination, CRM matchback shows which assigned accounts later progressed, and the mailed-versus-holdout difference estimates what the program caused. The missing QR code removes one direct-response path. It does not remove the experiment. Keep a stable campaign and account identity in the CRM, fix the outcome and window before launch, and join later outcomes to both assigned groups. A meeting booked through a rep email still counts toward the assigned account's outcome even though nobody scanned the piece. Do not replace the missing scan with an assumption that every delivered piece was read. Postal tracking confirms a delivery point, not recipient attention. The outcome comparison remains the useful number. ## What answers causation: holdouts A [holdout](https://trysincerely.com/glossary/holdout) is the only method on this page that answers "did the mail cause this?". Randomly split your target accounts before launch. Mail one group, hold the other back, treat everything else the same. After the measurement window, compare outcomes. The difference is [lift](https://trysincerely.com/glossary/lift), and because the split was random, lift is causal. The would-have-bought-anyway accounts appear equally in both groups and cancel out. Holdouts have honest costs: 1. You give up revenue from the held-out group during the test. 2. Small programs may lack [statistical power](https://trysincerely.com/glossary/statistical-power). A 200-account campaign cannot detect a 1-point lift. 3. You need discipline: decide the primary outcome, the window, and the holdout size before launch, not after you see the data. The full mechanics, including sizing, live in [the holdout guide](https://trysincerely.com/holdout-testing). The [holdout-size calculator](https://trysincerely.com/tools/holdout-size) tells you in a minute whether your campaign is big enough to measure. ## A layered setup that does not lie You do not have to pick one method. Layer them by what each one proves: | Layer | Method | What it proves | | --------------- | ------------------------------ | ---------------------------------------------------------------- | | Causation | Holdout with random assignment | Whether the campaign moved the metric at all | | Direct evidence | QR, PURL, promo code | That a specific person responded to a specific piece | | Inference | CRM matchback | Which mailed accounts progressed, causal only versus the holdout | This is how Sincerely reports by default. Campaigns lock the experiment design at launch: primary outcome, window, holdout percentage. Readouts show lift against account-level holdouts with confidence intervals, and say plainly when a result is not statistically ready. Scans and matchback appear too, labeled as evidence or inference. Details are on [the measurement page](https://trysincerely.com/measurement). If your program is too small for a powered result, keep the holdout and fall back honestly: report direct responses and matchback descriptively, then pool comparable cohorts. [Measuring a small direct mail audience](https://trysincerely.com/guides/direct-mail-small-audiences) covers that path. A limitation stated plainly beats a precise-looking lie. Attribution tells you who responded. Only a holdout tells you what you caused. ## Related questions - [How to measure direct mail with a small audience](https://trysincerely.com/guides/direct-mail-small-audiences): Keep an account-level holdout, predeclare one outcome, and pool comparable cohorts when your B2B direct mail audience is too small to measure alone. - [What is incrementality?](https://trysincerely.com/glossary/incrementality): Incrementality is revenue the campaign caused that would not have happened anyway. You measure it by comparing treated accounts against a holdout. - [Direct mail incrementality and lift calculator](https://trysincerely.com/tools/lift-significance): Compare mailed and holdout conversion rates to get absolute and relative lift, a 95% confidence interval, and how many conversions were incremental. --- Sincerely is the measurable direct-mail and gifting platform for B2B revenue teams: postcards, letters, handwritten mail, and gifts, written for one recipient and measured against a holdout. Contact Sincerely: https://trysincerely.com/contact Agent routing index: https://trysincerely.com/llms.txt