September 2026 · 4 min read
Campaign names should carry the hypothesis
If the ad account can't be read back into the experiment that produced it, the learning dies in a spreadsheet.
Most growth teams run more tests than they can remember. A new hook, a different offer, a landing page variant. Each one starts with a reason. A month later the reason is gone, and what's left is a campaign called “Prospecting_Q3_v4_FINAL” and a slide that says the test performed well.
Performed well against what? Nobody wrote down what they expected before the money went out, so any result can be explained after the fact. A lift confirms the idea. A drop gets blamed on seasonality or the algorithm still learning. Either way the next test starts from the same place as the last one.
Write the prediction first
A test needs three things before spend: what you believe, what you predict will move and by how much, and how long you'll wait before reading it. “Price-anchored creative will lower CPA by 10% to 15% against the current control within two weeks at the same budget.” That sentence can be wrong, which is what makes it useful. A prediction that can be wrong is one you can learn from.
The discipline is simple and it rarely survives a busy quarter. The hypothesis lives in a doc, the campaign lives in the ad account, and the results live in a dashboard. Three places, three owners, and nothing that joins them.
Put the key in the name
The ad account already has one field every campaign has to fill in: the name. If the name carries a test ID, the account becomes readable. “ML-042_price-anchor_prosp_CA” tells you which experiment produced this spend, and a script can pull the results back against the prediction without anyone rebuilding a spreadsheet.
It's a naming convention, and it's also the cheapest piece of measurement infrastructure a team can install. Once the names carry the link, better questions open up. Which of our predictions were wrong? Which kinds of ideas beat their forecasts? Who on the team calls tests most accurately?
Why we built Marketers Lab
Marketers Lab is the ledger between the idea and the account. You record the belief and the prediction, it issues the test ID that goes into the campaign name, and it joins the account's results back to that row. What you believed, what you predicted, and what happened stay together.
It doesn't write ads and it doesn't pick winners. It keeps a record honest enough that the next test starts smarter than the last one. Marketers Lab is in beta.