Ongoing Optimization, Performance Diagnostics, and Account Organization
The diagnostic ladder
Work down the funnel. No impressions: check eligibility, budget exhaustion, bids below the floor, targeting too narrow. Impressions but low CTR: irrelevant targets, weak main image, poor price display, unhelpful placement. Clicks but low CVR: detail page, reviews, price, delivery promise, variation confusion. Sales but high ACOS: bid too high, thin margin, match too broad. Each symptom has a specific lever, and changing several at once destroys the ability to learn.
The recurring optimisation cycle
Harvest converting search terms into exact-match ad groups; negate waste; shift budget toward campaigns with headroom and away from out-of-budget winners only after confirming that scale stays profitable; adjust placement multipliers using placement-report evidence; refresh creative when CTR decays; and re-check retail readiness continuously.
Account organisation
Segment campaigns by purpose so reporting stays readable: brand defence, category and non-brand prospecting, competitor conquesting, automatic discovery, and margin-based tiers. Use consistent naming (marketplace, targeting type, product line, match type, goal), group campaigns into portfolios by budget owner, and separate branded from non-branded traffic so TACOS and halo effects remain interpretable.
Traps
Pausing a keyword after a handful of clicks, cutting bids on the ad group that actually drives new-to-brand sales, letting one campaign serve eight unrelated ASINs and then blaming the algorithm, and scaling budget on a campaign whose margin is already negative.
Rule of thumb
Set a target ACOS from unit margin - a 30% margin implies a break-even ACOS near 30% - then optimise toward it in steps of 10 to 20% on bid, budget or multiplier, giving each change a full attribution window before judging it.