# Amazon Advertising Reports: Which One, When

URL: https://aiadking.com/blog/amazon-advertising-reports-explained
Updated: 2026-09-18

The six Amazon ads reports that answer real questions — search term, targeting, placement, advertised product, purchased product, over-time — what each is fo…

## The map: six questions, six reports
Two orientation notes before the details. First, these are the Sponsored Products reports — SB and SD have parallel sets with different column meanings (and different attribution generosity, as the ad-types guide covers). Second, the console's dashboard graphs are summaries of these reports, not substitutes for them: the dashboard shows you that ACoS moved; only the underlying report shows you which fifty search terms moved it.
## The search term report: the account's ground truth
One report earns the "read weekly, no exceptions" tier. The search term report shows the actual queries (and product pages) your ads served against — as opposed to the targets you bid on — with clicks, spend and orders for each. It is simultaneously:
The leak detector: sorted by spend, its top rows expose the proven wasters — in measured live accounts, about one ad dollar in nine sits on terms this report had already convicted.
The discovery feed: converting queries you never targeted are your graduation candidates.
The relevance mirror: what your auto campaign matched reveals what Amazon thinks your listing is about.
The trap: the report's recent days are attribution-incomplete — a term clicked Tuesday shows its orders days later. Judge nothing on the freshest slice, and never negate a "non-converting" term whose clicks are three days old. Pull 60 days, act on the settled majority, and re-run weekly so the fresh edge keeps rolling into judgment range.
## Targeting and placement: the two bid-setting reports
The targeting report answers the question the search term report can't: how are the things I actually bid on performing? A broad keyword aggregates dozens of queries; the targeting report scores the keyword as a container — the input for glide-path bid steps. Its trap is aggregation itself: a "mediocre" broad target is often one great query subsidising junk, which only the search term report can see. The two reports are a pair: targeting decides how much, search terms decide what.
The placement report splits each campaign's results across top-of-search, rest-of-search and product pages. Its single job is keeping placement modifiers honest: a top-of-search modifier is a percentage bid raise for the premium slot, and this report is the only evidence that the slot converts well enough to earn it. Its trap is small-sample drama — placement slices divide already-thin data three ways, so judge on long windows and remember the modifier compounds on top of every bid change you make elsewhere. Raising bids 20% and the modifier 30% in the same week is a 56% raise nobody decided on.
## The two product reports: advertised vs purchased
The pair that sounds redundant and isn't:
Advertised product report — performance per ASIN you advertise. This is where budget allocation gets honest: which products convert ad traffic at viable ACoS against their own break-even (not the account average — margins differ per product), and which are passengers absorbing spend because they're in the same campaign as a winner.
Purchased product report — what shoppers actually bought after clicking, including products you didn't advertise. Three genuinely useful signals hide here: halo sales (ad on product A, purchase of product B — real revenue your ACoS never credits), variation drift (clicks on the blue one, purchases of the black one — a listing-order signal), and accidental winners (a product selling steadily off other products' traffic, telling you it deserves its own campaign). It's also the report that keeps TACoS conversations grounded when someone claims ads "only" produce their attributed sales.
## The 25-minute weekly routine
The routine's design principle: every read produces a written output — a negation list, a bid-step list, a budget move, or an explicit "no action this week". That discipline is what separates reporting from scrolling, and it's also what makes the routine auditable: six months of weekly outputs is a decision log you can learn from, which no amount of dashboard-glancing produces.
Cadence notes: the search-term step is weekly without exception; targeting steps follow the settled-window rule so some weeks legitimately output "waiting"; placement is closer to monthly because its data accumulates slowly; and the whole routine doubles in value during Q4 — where it should run against event-adjusted expectations rather than normal baselines.
## Reports the routine deliberately skips
Equal honesty about the rest of the menu:
Impression-share / rank reports: interesting context, dangerous objectives. Optimising for impression share buys auction positions your conversion rate may not support — visibility is a means, and the targeting report already scores the end.
Hourly views: the data behind dayparting is real but console-level hourly slices are thin and attribution-smeared; block-level analysis over weeks beats hour-squinting every time.
Campaign-level summaries: the level where everything true about targets gets averaged into mush. Campaigns are containers; decisions live one level down.
Anything read daily: daily reading of any report mostly harvests attribution noise and anxiety. The one legitimate daily glance is budget pacing — did anything cap out — which is a status check, not a report.
None of these are useless; all of them are optional. The six-report core plus the budget glance covers every decision the optional set could trigger, with less noise on the way.
## One week's routine, with real-shaped outputs
To make "every read produces an action" concrete, here's a composite week for a ten-product account:
Step 1 output (search terms): a written list — negate exact: 4 terms (each 15+ clicks, zero orders, plausibly relevant no more); negate on sight: 2 clearly irrelevant; graduate: "bottle with straw lid" (3 orders, 22% ACoS) to owner campaign + seal upstream; watch-list: 3 terms at 8-12 clicks, revisit next week. Ten minutes because the report was sorted by spend and only the top forty rows matter this week.
Step 2 output (targeting): step down 5%: two exact terms whose settled ACoS sits above target with position to spare; step up 5%: one term consistently profitable but losing impression share; no action: eleven targets inside the noise band. The "no action" majority is normal and correct — the glide path is patient by design.
Step 3 output (placement, this month): no change — top-of-search sample still too thin post-restructure. Written down anyway, because "checked, insufficient data" is an output; silent skipping is how the monthly check dies.
Step 4 output (products + budgets): raise budget 30%: hero SKU capped out three days at 24% ACoS; trim: one passenger ASIN absorbing spend at 2× its break-even inside a shared campaign — split it out next week.
Total: 26 minutes, eleven written decisions, four of them "deliberately nothing". A quarter of those weekly lists, kept in one running document, becomes the most honest performance review an account can have.
## From reports to a system
Notice what the routine actually is: extract evidence, apply thresholds, write actions, repeat — the definition of mechanisable work. That's not an argument against doing it manually (the routine is genuinely how to run an account well by hand); it's an honest description of why it decays, and why AIAdKing's nightly cycle is essentially this routine with the skip-a-week failure mode deleted: search terms harvested and sealed nightly, bids stepped on settled evidence, placements and budgets paced from their own reports, and — the part manual routines never produce — every action logged with its reasoning into a change history you can audit in shadow mode before it ever applies. The reports stay yours to read; the drudgery of turning them into actions stops being a weekly tax. Flat fee, details here.

## FAQ

Q: Which Amazon advertising report is most important?
A: The search term report, by a wide margin: it shows the actual queries your ads served with spend and orders per query, making it simultaneously the leak detector (proven non-converting terms to negate), the discovery feed (converting queries to graduate to exact match), and the relevance mirror for your listing. Read it weekly, sorted by spend, on a 60-day window.

Q: What is the difference between the search term report and the targeting report?
A: The targeting report scores what you bid on (keywords, product targets); the search term report shows what those bids actually matched (real shopper queries). A broad keyword is one row in targeting but dozens of rows in search terms. Use targeting to size bids, search terms to decide negations and graduations — they're a pair, not alternatives.

Q: How often should I check my Amazon ads reports?
A: A 25-minute weekly routine covers it: search terms (weekly, non-negotiable), targeting on settled windows, placement roughly monthly, advertised-product budget checks weekly. Daily report reading mostly harvests attribution noise — the one daily glance worth taking is whether any winning campaign hit its budget cap.

Q: Why do my recent days show high ACoS in every report?
A: Attribution lag: Amazon credits orders to the click that produced them for days afterward, so the freshest days in any report show spend whose orders haven't landed yet. This systematically overstates recent ACoS. Judge on windows that exclude the last two to three days, and never negate a term based on its freshest clicks.

Q: What does the purchased product report show?
A: What shoppers bought after clicking your ad — including products you didn't advertise. It reveals halo sales (ad on one product, purchase of another), variation drift (clicked one variant, bought another), and accidental winners earning steady sales from other products' traffic — often the strongest signal that a product deserves its own campaign.