Every January we sit in the same meeting. A brand pulls up November and December, ACOS is three or four points worse than October, and the room starts producing explanations. Bids were too aggressive. The restructure didn’t take. The new campaigns dragged the average.
Almost none of those meetings begin by asking the only question that matters first: did we actually lose ground, or did the field move?
ACOS cannot answer that. ACOS is an absolute metric — your spend over your sales — and in Q4 every input underneath it changes at once. CPCs run well above the annual average as the whole market bids into the same ten weeks. Conversion rates shift because the traffic mix shifts. A campaign can get more competitive on every measure that matters and still post a worse ACOS than it did in a quiet October.
Search term impression share is the one Sponsored Products metric that holds you against the competition instead of against yourself. It is also the one almost nobody pulls before peak, which means it’s the one nobody can use in January.
What the report actually gives you
The search term impression share report for Sponsored Products is built in the ads console under Measurement & Reporting → Create Report. It returns two metrics per search term:
Search term impression share (STIS) — the percentage of all Sponsored Products ad impressions for that search term that you won over the period. If you show 20% for a term, you took one in five of the sponsored impressions available on it.
Search term impression rank (STIR) — where you sat against other advertisers on that term by impression volume. A rank of 3 means two advertisers got more Sponsored Products impressions than you did. Lower is better.
That’s a competitive read you cannot get anywhere else in the console. Your ACOS tells you what your traffic cost. Your impression share tells you how much of the available traffic you were even in the running for — and unlike every other number in the account, it moves when other people move.
Three caveats that make brands misread it
We’ve watched more than one account get restructured off a misread of this report, so before the tactics, the limitations.
1. Impression share is account-wide, but the rows are per campaign. This is the big one. The share figure for a search term reflects your whole account’s impressions on that term, not the impressions of the campaign whose row you’re looking at. If a term appears in four campaigns, you’ll see the same account-level share repeated across rows that have different spend, different bids and different ACOS.
Brands read those rows side by side and conclude one campaign is winning share and another is losing it. They’re looking at the same number four times. If your account isn’t tightly segmented — one term genuinely living in one place — impression share is a portfolio metric, not a campaign metric, and treating it as a campaign metric produces bid changes with no relationship to the thing being measured.
2. There is no lost-impression-share-by-reason breakdown. A lot of the advice circulating on this imports vocabulary from Google Ads, where you get lost impression share split into budget and rank so the platform tells you which lever to pull. Amazon doesn’t do that. You get share and rank. The reason you lost is something you derive, not something you read — and derive it you must, because adding budget when the constraint is bid does nothing, and raising bids when the constraint is budget makes the problem more expensive.
3. The data window rolls. Search term data in the console is widely reported across the practitioner community to be available for roughly the trailing 65 days, and once it ages out there’s no way to get it back. Confirm what your own account will hand you rather than taking a number from a blog post, including this one — but plan for the shape being true.
That third point has a consequence most brands don’t see coming until they need it: if you want to compare your Q4 impression share to last Q4’s, that comparison does not currently exist unless somebody exported it a year ago. Reports have retention windows. Your own archive doesn’t.
Deriving the loss reason in three questions
You’ve got a term where share fell or rank slipped. Work down this list in order — it takes about ten minutes per term and it’s the part the report won’t do for you.
Question one: were you out of budget? Pull the out-of-budget report, or check time-in-budget on the campaigns carrying that term. A campaign that stops delivering at 2pm cannot win impressions between 2pm and midnight, and in Q4 the back half of the day is not a quiet window. If you’re capping, the constraint is budget. Fund it and re-measure before you touch a bid — a campaign that has never run a full unconstrained day has never produced a full-day number to bid against.
Question two: are you present but placed badly? Pull the placement report and look at top-of-search impressions and top-of-search ACOS on that campaign. If you’re getting impressions but almost none at top of search, you’re in the auction and losing position, which is a bid or relevance problem, not a budget one. If top-of-search ACOS is materially better than the campaign average, that’s the classic case for pushing more weight to that placement rather than raising the base bid across everything.
Question three: are you present, placed, and simply not being clicked? If impressions held and clicks fell, the auction isn’t your problem. Something in the card changed — yours or somebody else’s. A 2026-specific note: the text layer of the search results page has moved twice this year, with the title character cap and with item highlights rendering as their own line on desktop. A CTR decline across that period is not automatically a creative failure. Check what changed on the listing before deciding what to change about it.
Why this matters more in Q4 than any other quarter
Because in Q4, a rising ACOS and a falling impression share mean opposite things, and they look identical on a dashboard.
ACOS up, share flat or up: you’re paying more for the same ground. That’s the market repricing, not your account underperforming. The decision is a margin decision — can we afford this traffic at this price — and it’s a conversation with the P&L, not a reason to cut bids.
ACOS up, share down: you’re paying more and losing ground. That’s a real competitive problem and it usually resolves into one of the three questions above.
ACOS flat, share down: the one nobody catches, because nothing looks wrong. Your efficiency held because you retreated. You bought less traffic at a similar rate while somebody else took the volume — and since velocity feeds organic position, you’ll pay for that in January, on a metric that lives in a different report.
On a brand doing $200K a month with $30K in ad spend, a couple of points of ACOS is a number worth a conversation. Share of the queries that produce the business is a number worth a strategy, and it’s the one that isn’t in the monthly deck.
What to do in the next two weeks
Holiday deal submissions for Prime Big Deal Days close September 8, Black Friday and Cyber Monday submissions close October 20, and peak fulfillment fees run from October 15 through January 14. Everything below wants to happen before those dates rather than after them. Confirm the specifics against your own Seller Central notifications.
Pull the baseline and save it with the date in the filename. Search term impression share and impression rank for your top revenue ASINs, on a normal August traffic pattern. This is a fifteen-minute job and it is the single most important item on the list, because it’s the only one that becomes impossible later.
Schedule the report to run recurring. The console will run it on a schedule. Set it weekly through the end of January so the archive builds itself and nobody has to remember.
Pick the ten to fifteen terms that matter. Not every term — the ones carrying the revenue. Write down current share and current rank for each. That’s your scoreboard.
Fund the chronic cappers now, before CPCs inflate. A campaign that caps in August will cap harder in November at higher prices. Fixing budgets is cheap, and it’s the one lever that needs a clean week of unconstrained delivery afterwards to produce a usable number.
Diary an early-October and an early-January review. Not a restructure — a read. Compare share and rank against the August baseline, and decide what changed before anyone touches a bid.
Don’t restructure. A structural change made in September stabilises in October, which lands the learning period on the most expensive traffic of the year. That window is closed. Budgets, bids and reports are cheap. Rebuilds are not.
FAQ
Where exactly do I find the search term impression share report?
Ads console → Measurement & Reporting → Create Report → Sponsored Products (or Sponsored Brands) → Search Term Impression Share. You can set the date range and schedule it to run on a recurring basis, which is what you want here.
Is impression share available for Sponsored Brands too?
Amazon has published impression share reporting on the Sponsored Brands side as well, and the Brand Category Benchmark reporting gives a different competitive read again. The mechanics differ, so read each one on its own terms rather than assuming the Sponsored Products definitions carry across.
What’s a good impression share?
There isn’t one, and be suspicious of anyone who hands you a number. Share is a function of how many advertisers are competing on a term and how much they’re willing to spend, both of which vary enormously by category and by week. The useful reading is your own trend on your own terms — which is exactly why the baseline matters more than the benchmark.
Should I be chasing 100% impression share on my branded terms?
Usually not, and it’s an expensive instinct. Defending your own name at a low ACOS is one of the better uses of spend in most accounts, but the last few points of share on any term are the most expensive points to buy. Watch rank on branded terms — if you’re not first on your own name, that’s the finding.
Our agency reports ACOS and TACoS monthly. Is that enough?
For a normal quarter, it’s defensible. For Q4 it isn’t, because both of those metrics move for reasons that have nothing to do with how well the account is being run, and neither one contains any information about the competition. Ask for share and rank on your top terms alongside them. If the answer is that nobody has been pulling it, the data for last year is already gone — start the archive now.
—
The uncomfortable version: most Q4 post-mortems are conducted entirely with metrics that describe the seller and none that describe the market, which is why so many of them end with a restructure that fixes nothing. The fix is a fifteen-minute report pulled in August and saved with a date on it.
If you’re looking for a team that manages every lever — creative, advertising, and operations — Velocity Sellers works with brands doing $100K+/month on Amazon. Contact us for a free account audit.
