The finding
This brand's Amazon account looked fine from the top: roughly $371,000 in revenue over a 59-day window, an 18.3% TACOS, a 13.5% store conversion rate. Nothing in the headline numbers said anything was wrong.
When we ran this brand's Account Diagnostic, we cross-referenced the full search-term report against the campaign structure. The problem only showed up when we put two reports side by side: the full search-term report and the campaign structure. Auto campaigns had surfaced 453 unique search terms that had each already generated two or more orders, at profitable ACOS levels, and not one of them had ever been promoted into a Manual campaign. The brand had a list of proven winners it had already paid to discover, and it was leaving every one of them on autopilot.
The data
- 453 unique search terms with 2 or more orders each, never promoted to Manual Exact
- "bar peeler": 36 orders, $950 in sales, 11% ACOS
- "lemon zester": 50 orders, $706 in sales, 36% ACOS
- "mixed drink stirrer": 3 orders at 0.2% ACOS
AUTO TODAY (no bid control)
MANUAL (bid-controlled)
453 proven terms sat in the left state. The fix moves them to the right.
The leak mechanic
This is the part most audits miss, because it is invisible unless you join the search-term data to the campaign structure. Auto campaigns cannot be optimized at the keyword level. You cannot push more budget toward a single winning term inside an Auto campaign without raising the budget on everything that campaign is buying, including the losers.
So these 453 proven terms were stuck at whatever the Auto algorithm chose to spend on them. The brand had no lever to put more money behind its own best demand. The winners were converting, profitably, and capped. Every week they stayed in Auto was a week of demand the account had proven it could win, and was choosing not to scale.
The recommendation
Harvest them. Pull every Auto converter with two or more orders. Send true keywords to Manual Exact campaigns and ASIN targets to Manual Product Targeting, each with controlled bids set to a 35% ACOS target.
Then add the harvested terms as negatives in the source Auto campaigns, so Auto stops paying for what is now handled elsewhere and goes back to hunting for the next batch of winners. This is a one-time structural fix that turns a static list of capped winners into a set of campaigns the brand can actually scale.
The projected impact
Modeled on bid-controlled scaling of the proven converters, this single lever projects roughly $19,500 in incremental revenue over 90 days, on demand the account had already validated. This is a 90-day projection from scaling proven terms. It is not recovered waste, and it is not a guarantee.
This is the kind of finding a single Account Diagnostic surfaces.
Most accounts are hiding at least one. Seven days, a 30 to 50 page report, and a prioritized action list.