How to Tell Whether Your Postcard Actually Worked

Ask most local businesses whether their last mailing worked and you get a feeling rather than a number. The phone seemed busier. Somebody mentioned a card. That is not enough to decide whether to do it again, and it is why so many businesses try mail once and quietly stop.

The measurement problem is solvable, but only if you set it up before the card prints. Afterwards is too late.

Why mail feels unmeasurable

A card lands in about five thousand homes across a few days. Some people call immediately. Some keep it on the fridge and call in six weeks. Some look you up online instead of calling the number on the card, so the response shows up as a search visit with no obvious source. Some mention it, most do not.

All of that is real response, and none of it is automatically attributed to the card. Meanwhile normal business continues, so the signal is buried in noise.

The fix is not clever analytics. It is putting something on the card that only exists on the card.

The mechanisms that work

An offer that exists only on the card. A specific figure, a named package, a code to quote. If somebody claims it, they can only have got it from the card. This costs nothing, needs no technology, never expires, and still works when the card comes off the fridge a year later. It is the strongest single thing most local businesses can do, and the most commonly skipped.

Asking. Put a line on the card asking the reader to mention it, and ask the question again when the phone rings. It undercounts, because plenty of people forget to say it, so treat what you capture as a floor rather than a total. Cheap enough that there is no reason not to.

Your own call volume. The least clever and often the most honest: count the calls in the three weeks after the drop against the three weeks before. It captures everybody, including the ones who never mention the card and never claim the offer, which is the population every other method misses.

A tracked phone number, if you want one. A separate number bought from a call-tracking provider and forwarded to your normal line, so calls to it can only have come from the card. Know the trap before you commit: it is a monthly cost per number for as long as the card is in circulation, and a card can sit on a fridge for a year. Release the number and it does not go dead, it gets reassigned to somebody else - so a late caller reaches a stranger holding your postcard.

Use at least two. The offer catches the people who act on it and your call volume catches everybody else. The gap between the two tells you something about your audience.

Where you send them

This is where most campaigns quietly lose their measurement and a good share of their response.

A code pointing at your homepage makes the reader start over. They came from an ad about one specific thing and arrived somewhere that talks about nine. Most will not hunt for it.

A code pointing at the page on your own site that matches the offer - same headline, same figure, same fine print, with a form or a phone number - converts far better and produces a clean number. If you are on a shared card, ask whether the code is yours alone and where it sends the reader.

What to count, and over how long

Count redemptions, codes mentioned and your own call volume, and count them for at least eight weeks.

The two-week number will understate the drop badly. Mail keeps working: a card on a fridge is still a prospect in month two, and for high-value trades the lag between seeing a card and needing the service can be long. Judging on week one is the most common way a working card gets written off.

Then convert the count into the only figure that matters: cost per lead, and if you can, cost per job. A slot at four hundred and ninety-nine dollars that produced twelve identifiable leads cost forty-two dollars a lead. Whether that is good depends entirely on what a lead is worth in your trade, which only you know.

Attribution honesty

Two failure modes, in opposite directions.

Under-counting. Somebody sees the card and searches your name, then calls from the search result. That is a mail response, and it will not appear in any of your three mechanisms. If you see a bump in branded searches or direct traffic in the fortnight after a drop, that is almost certainly the card.

Over-counting. Somebody who was going to call you anyway happens to use the code. That is not new business, it is a discount on business you had.

Neither is fixable precisely. Both are worth knowing about, so you treat your number as an estimate with error bars rather than a fact.

Reading a quiet result

If a drop produces very little, resist the urge to conclude "mail does not work". Work through the four things that could have caused it, in order.

The zone. Were those the right households? Did response cluster in one part of it? If you can see which streets the redemptions came from, split the zone and mail the half that worked.

The offer. Was there one, and was it specific enough to act on?

The timing. Did it land before the season or during it? Did it land in a month when a chunk of the neighborhood was empty?

The follow-up. Could somebody answer the phone that week? A surprising share of "failed" drops are answered-phone problems.

Only after all four should the channel itself be on trial.

Set it up before you print

Every mechanism in this piece has to exist before the card goes to press. An offer code the caller has to mention costs nothing and takes one line on the ad. Pointing the code at the right page on your own site takes a minute. An offer code takes one line on the ad but has to be specific enough to be unmistakable.

The cost of setting all three up is small. The cost of not doing it is that you will run a drop, get a feeling, and have no basis for the next decision - which is the actual reason most local businesses believe direct mail cannot be measured.

Build the measurement into the schedule

Set three dates when you book the drop, not after it.

Mail date plus two weeks. First read. Expect the largest single burst here, and expect it to be a fraction of the total.

Mail date plus six weeks. The real number for most trades. Fridge-kept cards are converting by now.

Mail date plus twelve weeks. The final figure, and the one to compare against the next drop.

Diarise them. The most common measurement failure is not a missing mechanism, it is nobody looking at the numbers after the initial excitement.

What good looks like

There is no universal benchmark, and anybody quoting you a response-rate percentage without asking about your trade is guessing.

What is useful is your own baseline. Your first measured drop sets it. The second one tells you whether a change - a different zone, a different offer, a different month - moved it. That comparison is worth more than any industry average, because it controls for your business.

If you want a rough sanity check before you have a baseline: for a high-value trade, a handful of identifiable leads from five thousand homes is a card that paid for itself several times. For a small-ticket business, a handful is not enough, which is the arithmetic from the start.

Ask the phone, properly

Whoever answers should ask how the caller heard of you, and should record the answer somewhere retrievable. Not "did you see our card" - that leads the witness. "How did you hear about us?" and then write down what they say.

Two weeks of that costs nothing and consistently surprises businesses, because the answers rarely match what they assumed. It also catches the responses no mechanism can, such as the person who saw the card and mentioned it to a neighbor who then called.

Keep the card itself

Sounds trivial. File a physical copy of every drop with its date, its zone and its offer.

Three drops later, when you are trying to work out why one performed and another did not, having the actual pieces in front of you makes the comparison concrete. Offers blur in memory and the differences that mattered are usually small ones you will not otherwise recall.

What to do with a good result

Repeat it before you change it.

The temptation after a card that worked is to improve everything at once - new offer, wider zone, bigger format. Then the next one underperforms and you cannot tell which change caused it.

Run the same card to the same zone again. If it holds up, you have something repeatable, which is far more valuable than a single good month. Change one variable at a time after that.

What to do with a bad one

Change one thing, not everything, and be specific about which.

If the response clustered in one half of the zone, the zone is the variable. If nothing came back anywhere, it is the offer or the ad. If plenty called and nothing converted, it is the offer or the follow-up.

The mechanisms in this piece are what let you tell those apart. Without them, every disappointing drop looks the same and the only available conclusion is that mail does not work - which is usually wrong, and always expensive to believe.

Comparing mail against your other channels

Once you have a cost per lead from a drop, you can put it next to your other channels honestly - and the comparison is usually more favorable than businesses expect.

Search advertising is excellent at catching somebody who needs you today, and the cost per click in most home-service trades has climbed steadily. Social is cheap per impression and hard to aim at a specific five thousand households. Mail costs more per household than either and hits every door in a defined area exactly once, whether or not that household is online.

The point is not that one wins. It is that until you have a measured number from a drop, you are comparing a channel you have data for against one you have a feeling about, and the feeling always loses. Measuring the card is what lets it compete on its merits.

One number worth asking your operator for

If somebody else is printing the card, ask what measurement they actually provide and who owns it. Some operators offer tracking, some do not, and an operator who offers none but tells you to run your own coded offer is being straight with you. What should worry you is a card-wide number presented as yours: sixteen advertisers cannot divide one figure sixteen ways and learn anything from it.

Ask where any printed web address points, too. It should be your own domain, not a page the operator owns. A page on somebody else's domain stops working the day you stop advertising with them, and any search value it earns belongs to them rather than to you.