There's a category of advertising waste that has nothing to do with strategy, creative or competition. It's simpler and stupider than that: money spent on search terms that have never, in the entire history of the account, produced a single sale. Not underperformers to optimise. Zeros. In one recent audit from our own book, over £3,400 of a year's ad spend had gone to hundreds of search terms with zero sales between them. Every account has a version of this number, and almost nobody knows theirs.
At rankhouse, building the kill list is week-one work on every new account, because it's the fastest legitimate money in Amazon advertising: a pure cost with no offsetting benefit, removable in an afternoon. Here's the full method.
Why zero-sale spend accumulates in every account
It isn't incompetence. It's structure. Auto campaigns and broad match exist to discover new search terms, which means they are designed to spend on unproven queries. That's fine, that's their job. The failure is in the second half of the loop: nobody comes back to judge the experiments. Terms that failed keep receiving budget, month after month, because the review cadence that should catch them doesn't exist. Discovery without judgement is just leakage with a research budget.
The leakage compounds in a specific way, too. Zero-sale terms are usually individually small, a few pounds here, twenty there, which is exactly why they survive. No single term is worth a meeting. Collectively, across hundreds of terms and twelve months, they're a real number, and they're often concentrated in the campaigns nobody has opened since they were built.
| Step | Filter | Action |
|---|---|---|
| 1. Pull the data | Twelve months of search term reports, every campaign | One sheet, every term, spend and orders side by side |
| 2. Find the zeros | Meaningful clicks and spend, zero orders | These are the leaks |
| 3. Judge, then cut | Is it genuinely irrelevant, or just early? | Negative exact the irrelevant; watch the undecided |
| 4. Redeploy | Freed budget | Move it to terms already converting below breakeven ACoS |
| 5. Repeat | Every week, not once a year | Waste regrows; the loop is the discipline |
Building the kill list, step by step
- Pull the search term report for the longest sensible window. Twelve months where the account history allows. Short windows produce false zeros on genuinely slow terms; a year of clicks with nothing to show is a verdict.
- Filter for spend above a materiality floor with zero orders. The floor depends on your price point: a term that has spent more than one expected cost-per-acquisition with nothing to show has had its chance. Below the floor, leave the experiment running.
- Read the list before you cut, because the kill list is also a diagnosis. The zeros cluster into patterns, and each pattern names a different disease:
- Irrelevance: terms from a different intent entirely, leaking in through loose match types. Disease: targeting hygiene.
- Wrong fight: relevant terms where shoppers compared you and chose cheaper. If a premium product is spending on bargain-intent queries, no bid change fixes that. Disease: positioning, treated in our premium pricing piece.
- Listing failure: perfect-intent terms that clicked and didn't convert. The ad worked; the page lost. Disease: main image, price or reviews, not PPC at all.
- Negate with precision. Negative exact for specific dead queries; negative phrase only for whole dead families, checked against the report first so you don't block a converting variant. Sloppy negation is how brands cut waste and volume together.
- Redeploy, don't bank. This is the step that turns hygiene into growth. The recovered budget has a known size; move it deliberately into proven exact-match winners, into underfunded ingredient terms, or into the experiments that earn their keep. An account that kills waste and banks the savings has optimised its way to standing still.
Want this level of thinking inside your account?
rankhouse is a boutique specialist agency: experienced Amazon professionals, no juniors, led personally by founder Dan Whalley. Fixed fee from £1,500 + VAT a month. Every engagement starts with a free profit audit built from your own data.
Get the free audit at rankhouse.co.uk →Keeping the list dead: the maintenance loop
A kill list built once is a spring clean. The value is in the standing process, because auto and broad campaigns will faithfully manufacture new zero-sale terms every month, that's their nature. The working cadence across our accounts: a monthly zero-sale sweep at a lower materiality floor, inside a weekly review of the search term report for harvesting winners. Kill and harvest are the same motion in opposite directions: unproven terms that failed get negated, unproven terms that converted get promoted to exact match with their own bids. Run both and your targeting sharpens every single month; run neither and it dissolves at the same rate.
What the number tends to say about the account
The kill list total is a useful proxy for something bigger: how much judgement the account has been receiving. A small number says someone has been reading the reports and making calls. A large one almost always comes with siblings, blended reporting that hides per-product losses, campaigns structured in 2022 and never revisited, bids set against ACoS instead of per-product breakeven TACoS. Waste is rarely lonely.
Which is why we open every engagement by finding it. The free audit we run at rankhouse puts a hard number on your zero-sale spend, along with where every other pound of the ad budget actually went and what we'd do about it in the first 90 days. It costs nothing, it comes from your own data, and for most brands the kill list alone pays for the first year of working together. That's not a sales line. It's arithmetic.
Questions we get asked about this
How much zero-sale spend is normal in an Amazon account?
Every account carries some, because discovery campaigns exist to spend on unproven terms, and that's legitimate research cost. What separates hygiene from leakage is judgement cadence: in accounts reviewed weekly, zero-sale accumulation stays modest, the cost of live experiments; in accounts nobody has swept for a year, we routinely find four figures attached to hundreds of terms that failed long ago and kept billing. In one audit from our own book, over £3,400 of a single year's spend sat on hundreds of terms with zero sales between them. The number itself matters less than what it proxies: how much judgement the account has been receiving, because heavy zero-sale spend rarely travels alone.
What spend threshold should trigger a kill?
Anchor it to your expected cost per acquisition, not to a round number. A term that has spent more than one CPA-worth with nothing to show has had a fair trial; below that, the experiment is still running and killing it early trades real discovery for tidy-looking reports. Price point changes everything: a £12 product might warrant killing at £8 of fruitless spend, a £60 product at £25. Two refinements make the rule robust: use the longest sensible data window, ideally twelve months, so slow-but-real converters aren't executed as false zeros, and hold a lighter trigger for terms that are obviously irrelevant on sight, because those needed no trial at all.
Negative exact or negative phrase, and how do I not block good traffic?
Negative exact for specific dead queries, which is surgical and safe. Negative phrase only for whole term families you're confident are dead, and only after searching the report for every variant containing the phrase, because one converting variant inside the family means phrase-blocking costs you real sales in exchange for tidiness. The discipline that prevents accidents: negate from evidence, not intuition; keep a log of what was blocked and why, so future reviews can reverse mistakes; and re-check the report a fortnight after big negation passes, because volume that shifts unexpectedly is the symptom of an over-broad block. Sloppy negation is how accounts cut waste and growth in the same motion.
Where should the recovered budget actually go?
Somewhere specific, decided when you cut, because recovered spend that drifts back into the general pool achieves nothing measurable. The three destinations in rough priority: proven exact-match winners currently constrained by budget, where the marginal pound has known economics; underfunded generic ingredient terms where new customers live and share is available; and fresh discovery in genuinely new territory, new match types, new products, new placements, which restocks the pipeline the kill list depends on. The kill list and the harvest are one motion in two directions: failed experiments negated, successful ones promoted. Run both monthly and targeting sharpens continuously; run neither and it dissolves at the same rate.
Can't I just use automated rules to do all this?
Automation handles the arithmetic well and the judgement badly, and the kill list is one-third arithmetic. Rules can flag zero-sale terms past a spend threshold reliably. What they can't do is read the list as a diagnosis, distinguishing irrelevant leakage from wrong-fight positioning problems from listing failures where the ad worked and the page lost, and each of those patterns demands a different fix that isn't a negation at all. Our working setup uses automation for detection and a human weekly for the verdicts, which takes minutes once the model exists. Full autopilot produces tidy accounts that stop learning; full manual produces learning that stops happening. The blend is the method.
Almost every Amazon account quietly funds search terms that have never produced a single sale, not underperformers, zeros, because discovery campaigns manufacture unproven terms by design and nobody runs the judgement loop that should follow. One audit from our own book found over £3,400 across hundreds of zero-sale terms in a single year. The method: pull twelve months of search term data, filter for spend above roughly one cost-per-acquisition with zero orders, read the list as a diagnosis before cutting because the zeros cluster into irrelevance, wrong fights and listing failures, then negate with precision and redeploy the budget into proven winners rather than banking it. Then make it a standing loop: monthly kill sweeps beside weekly harvesting of converting terms, the same motion in opposite directions. The kill list total is really a measure of how much judgement the account has been receiving, and waste is rarely lonely.
The next step is twenty minutes.
If any of this reads like your account, the fastest way to find out is the free audit: per-product profitability, where the ad spend is leaking, and what we would fix first. No pitch deck, no obligation.
Book a chat with Dan →
About the author
Daniel Whalley is the founder of rankhouse, a boutique specialist agency for Amazon-focused growth in FMCG, health, wellness and beauty brands. He has spent 10 years inside Amazon accounts, generating £100M+ for the brands he works with, and manages £500k+ a month in ad spend across the UK, Europe and the US. He writes from inside the accounts he runs, not from the sidelines. Connect on LinkedIn → · amazon@rankhouse.co.uk