The four stages of a paid ads program that compounds

Research to briefs, briefs to tests, tests to decisions, decisions back to briefs.

Otter Growth Advisory operating method

An ad account gets stuck the same way almost every time. Creative gets made because someone had an idea, several things change between one ad and the next, results come back unreadable, and the next batch starts from opinion again. Nothing compounds, so month twelve looks like month one with a bigger budget. The fix is not better ideas. It is four stages that hand each other something specific.

Stage one: derive the creative strategy, do not brainstorm it

The most common failure is deriving every motivator from one source, usually customer interviews. Interviews are worth running and they systematically miss a whole class of motivation, because everyone who agrees to a thirty-minute research call is already thinking hard about the problem, while the person scrolling the feed is not. So six kinds of research run, each one catching what the others cannot, and every motivator gets scored on the same six weighted dimensions before five are confirmed by name.

The rule that makes it real: a kind of research that returns nothing was not done. That is a gap in the research, not a finding about the market. Letting one come back empty quietly turns a six-source method back into a one-source one.

Stage two: write briefs that can be read as tests

A backlog is not a list of ideas. It is a set of concepts, each with one base ad and a set of variations, where every variation changes exactly one thing from that base ad. Opening line, format, person, offer, one of them, never several at once. The moment two things change together the result stops being readable, and an unreadable result is the same as no result while still costing the media spend.

The column that does the work is the one naming what the row changes from the base ad. If a row cannot fill it, the row is another ad rather than a test, and it belongs in a different budget line.

Stage three: classify before you decide

Every live ad lands in one of four groups, and each group carries its decision before anyone looks at it. Better than the target range, scale and build variations. Inside the target range, keep running. Worse than the target range, cut. Too little spend to judge, no decision in either direction, which is the group most accounts get wrong.

Killing an ad that has not spent enough to judge is not discipline, it is noise. It has not gathered enough conversion volume for the difference to mean anything, and cutting it teaches the account nothing while removing a candidate that might have won.

Stage four: report so the numbers argue

Most ad reporting answers what happened and stops, which leaves the reading to whoever opens it first. That is where confident wrong conclusions come from. Every metric should carry its target, whether it is on or off, its trend, and for the metric that needs a decision, the cause and the move in plain words.

Two rows belong on every ads dashboard and are almost never there. Share of spend sitting on the best ads, because an account can be full of good ads and still spend most of its budget on average ones. And how many creatives are still in learning, because that number explains a bad cost per result better than the cost per result does.

Why the loop matters more than any single stage

Each stage hands the next one something it can act on, and the last hands back to the first. The classification tells you which angles earned more variations. The variations are briefs. The briefs are traceable to a motivator and a scenario, so when one wins you know what won and not merely which file won. That traceability is the whole difference between an account that compounds and one that restarts every month.

None of these four stages is expensive on its own. Running three of them and skipping one is what costs money, because the skipped stage is where the readability was going to come from.