The other day I audited an account spending about $100K a month on Amazon Ads. Buried in the Search Term Report: a long tail of queries with exactly one click and zero sales and together they added up to 40% of their budget. Now this might be a more extreme case but this is definitely something worth your time to investigate!
I’m sure you figured Amazon was still “finding new keywords.” What’s really going on is simpler: In the past Amazon tested terms to experiment new terms for you, which is exactly how auto, broad, and even phrase matches work. However, Amazon’s follow up for these terms has been poor as they show your product once for a search query and then seem to forget that term exists.
When one shopper clicks on your ad once and you never get the chance to appear for that term again, this creates a graveyard of terms that provide you with no results, no data, and worse returns than you deserve. Now when this happens a handful of times, that’s okay. When it happens so often it takes up half of your spend, this is something that needs to change.
Below: a quick glossary if Amazon Ads is new to you, why this happens, and what to do this week.
Quick glossary
• Search term report — The list of what shoppers actually typed (what your ads matched before that click).
• Match types — How loosely Amazon maps your bids to searches. Broad and Auto cast wide nets; Phrase sits in the middle; Exact is the tightest.
• Negative keyword / negative target — Rules that tell Amazon not to show you for certain queries/placements—so budget flows toward shoppers more likely to buy your catalog.
Why single-click / zero-sale is so common now
Amazon is built to probe “will this query convert?” Under broad/auto/phrase, you get lots of shallow tests: low impression clusters, one paid click, no order—and limited repeat shots at the same pattern. That’s exactly where you land on one click’s worth of data … and no way to optimize with confidence.
You still need to do two jobs:
• Cleanup — Address damage already done.
• Prevention — Use negative target lists so future exploration bumps into boundaries where you won’t donate budget.
Best-practice anchors (widely echoed in account work)
• Review search terms on a rhythm that matches spend (weekly for heavy accounts; at least monthly for many).
• Don’t confuse “junk intent” vs “thin data.” Obviously wrong searches → negatives. Possibly valid noise → judgment and rules—not fake certainty off one click.
• Put negatives where exploration is widest (especially broad- and auto-heavy campaigns) so learning runs in clearer lanes.
• Leverage more exact phrase terms where you want your focus to be.
This week: see it, judge it, list it
Step 1 — Ads → Reporting → Search term report (often last 30 days; stretch to 60 if ordering volume is thin).
Step 2 — Filter Clicks = 1, Purchases / Orders = 0. Sort by cost (high to low).
Step 3 — For each row ask: Would a sane shopper who wants what I sell type this? If no → add negatives (often exact for surgical cuts; sometimes phrase if an entire phrase family is wrong). If maybe → don’t bulldoze the account over one click; note bigger spend rows first.
Step 4 — Maintain logical negative lists (“Irrelevant category,” “Brands we don’t chase,” etc.) so you apply them cleanly to exploratory campaigns—not one-off negatives you forget in three months.
Speed this up with AI
Paste your Search Term Report CSV into Claude (strip sensitive IDs!):
Paste this Prompt: Analyze this Search Term Report. Show rows with exactly 1 click and 0 orders. Rank by highest CPC/Spend. Flag rows clearly irrelevant to [one-sentence: what you sell]. Recommend negative exact vs negative phrase where obvious. Leave ambiguous rows labeled “investigate,” not negatives.
Spend your human time checking edge cases, not scrolling every row manually.
---
Next Tuesday: Campaign structure decoded—intent buckets × ad formats so dashboards stay readable for months.
