It asks every morning what an expert would ask
An experienced account manager asks the same questions every morning: which keywords take clicks and never sell, which product is advertising while its stock runs out, which placement is burning money. The engine asks those questions every day for every product. Not creative — meticulous: no rule guesses, each one compares a measured number against a threshold.
Every finding has a recorded reason
A suggestion you cannot audit is a suggestion you cannot trust. Each one states the rule, the threshold and the measured values that crossed it.
That also makes the engine arguable. If the threshold is wrong for your account, you can say so once instead of rejecting the same card every week.
When no card appears, there is a reason for that too
Silence is ambiguous: it can mean everything is fine or it can mean nothing was measured. The two call for very different reactions.
A product with no strategy assigned produces no suggestions — and says so, at the top of the screen, with the ad spend it is running without supervision.
The engine measures its own accuracy
Rules that look right in theory can be wrong for a catalogue. The reports show how often each rule’s prediction held.
That turns tuning into evidence rather than opinion: a rule with a poor record is a rule to change, not a rule to argue about.
Frequently asked questions
What kinds of suggestions does the engine produce?
Bid waste, unprofitable targets, negative candidates, search term harvesting, placement multipliers and stock-versus-advertising conflicts.
Why am I not getting suggestions for a product?
Most often because no strategy is assigned to it. The engine needs to know whether you are optimising for profit, launch, growth, liquidation or defence before it can recommend anything.