One goal, one number

Impressions. Reach. Frequency. Clicks. Click-through rate. Cost per click. Cost per thousand shown. Page views. Conversions, counted three different ways. Cost per conversion. Return on spend. That is one campaign, one week, one screen — and every one of those numbers is accurate. None of them is the reason you ran the campaign. You had a reason. It was a sentence in your head before any money moved, and it is nowhere on that screen.

The wall exists because a platform reports what it is able to measure, which is a completely different thing from what you decided to find out. It has no idea whether you were trying to get in front of people who have never heard of you, or trying to get last month's visitors to finally sign up. So it hands you everything, in the reasonable hope that something in there is what you meant. The wall is not a trick. It is just an answer to a question nobody asked, printed at length.

What it does to you, though, is specific and worth naming. Give a person eleven numbers and they will read them the way anyone reads a horoscope: skimming for the line that matches the mood. On a hopeful Monday the click-through rate is up and the campaign is working. On a nervous Thursday the cost per signup is up and the campaign is a disaster. Same campaign, same week, same data. The number you land on is being chosen by your mood, which means the report has stopped being evidence and started being a mirror.

So the fix is to make the choice before the data exists, and to make it exactly once: the objective picks the number, and everything else on the screen is scenery. The shapes are simpler than the dashboard makes them look. If the campaign exists so that people who never heard of you hear of you, the number is how many of them arrived — visits, from your own side of the glass, not the platform's count of eyeballs. If the campaign exists to bring people to a specific page, the number is what a visit cost. If it exists to produce signups or sales, the number is cost per signup, or revenue against spend, and nothing above it in the funnel gets a vote.

when you watch five numbers, one of them is always up. a campaign you cannot lose is a campaign you cannot learn from.

The discipline is not the picking, which takes ten seconds. It is the ignoring, which is genuinely uncomfortable. An awareness campaign that produced two signups did not fail at signups — it was never asked about signups, and grading it there is like marking a spelling test on handwriting. A conversion campaign with a beautiful click-through rate and no signups is not partly working. It is failing at its one job while being good at a job you did not need doing. Both of those sentences are easy to write down in advance and remarkably hard to say out loud a week later, in front of a number that is finally green.

One extra number is allowed, and only one, and it is a guardrail rather than a second goal. A guardrail is a limit you would stop for, not a result you are chasing: total spend, or cost per click going somewhere your economics could never survive, or an unsubscribe rate climbing past what you consider the ordinary price of showing up. The difference matters because a goal and a guardrail are read in opposite directions — the goal is the thing you want to move, the guardrail is the thing you want to stay still. Two goals is zero goals. A goal and a limit is a decision you can actually make on a Thursday.

Then check that your chosen number can move at the size you are actually operating at, because this is where good intentions quietly break. A conversion goal on a hundred and thirty visits is a number built from two or three events, and a handful of events is noise wearing the costume of data — the same reason a first hundred dollars of ad spend buys you a real opinion about your ad and almost none about your product. If your budget cannot produce enough of the thing you picked, move up the funnel on purpose: judge this run on what a visit cost, decide in advance how much spend has to accumulate before the signup number is allowed to speak, and stop reading it until then.

Pick a number you can see on your own side, too. The platform will happily report conversions it credits to itself across windows it chose, and a view count that means something different on every service — a gap wide enough that ten thousand views and ninety visits can both be true about the same afternoon. That does not make the platform's numbers useless; they are the right tools for making the work better, and the wrong tools for deciding whether the work was worth doing. When the ad account says twelve and your own screen says three, your goal number is three, because three is the one your bank account agrees with.

This is roughly the shape we build toward. The metrics your connected tools report land on one screen next to spend, you can keep a small chosen set in front of you instead of everything a provider is willing to offer, and when something is not connected the panel says so plainly rather than drawing a zero you might read as a result. Campaigns drafted here arrive paused and spend nothing until a person starts them, and that quiet minute before starting is the cheapest moment there will ever be to write the sentence down: this campaign exists to move this number, and I will not grade it on anything else.

Because a campaign without one chosen number does not have no goal — it has whichever goal the report offers you on the day you look. That is a worse position than being wrong, since being wrong at least teaches you something by the end of the month. Choose the number while you are calm, write it where you will see it, and let the other ten be what they always were: true, expensive to read, and none of your business this week.