Why Equaliser
Advice that keeps its receipts
Most tools stop at the recommendation. Equaliser is built around what happens after it: each recommendation is projected, observed, and graded against what actually happened, and the grade publishes either way, whatever it says.
The loop is young. Grades publish as changes complete their observation windows, and the ledger grows from there. We would rather show one honest grade than a hundred simulated ones.
01
The recommendation states its claim
Every recommendation names the metric it expects to move, the direction, the size, and the observation window it should be judged over. The window is named on the recommendation itself, typically one to four weeks depending on how quickly the metric settles. Not “improve performance”: a falsifiable sentence.
02
The baseline is frozen
Before anything changes, the platform records what the metric was doing. The comparison is fixed in advance, so nobody gets to pick a flattering window afterwards.
03
The window runs
You decide whether to act. Equaliser connects with read-only scopes and does not execute changes on your ad accounts; if per-action execution ships in the future, every action will require explicit human approval, under scopes you grant at that point and can revoke. Acted on or not, the clock runs and the platform watches the metric it named.
04
The grade publishes, whatever it says
Worked, partly worked, no effect, or worse: the grade publishes either way, whatever it says. It lands on the recommendation, in the account view, and in the client report, with a miss shown at the same prominence as a hit, and agents that miss get corrected.
Why this is rare
Grading your own advice is uncomfortable. A dashboard that only presents data can never be wrong; an AI that never checks its predictions can claim anything. Closing the loop means some grades read “no effect” or “worse”, on the record, in front of the client. That is the deal, and it applies to every change.
The same discipline runs through the numbers themselves: when the data cannot support a figure, Equaliser says so instead of printing one. Zero conversions over real spend reads “tracking gap”, never “healthy”.
See the loop on your own account data.
Start with the Paid Media Diagnostic