Reading search query data: the funnel per keyword
- You read the search query funnel as a diagnosis rather than a report.
- You know which gap between two stages points to which problem — and the matching action.
- You know the four false conclusions people routinely draw from share data.
- You have a ranking for which search query to fix first.
Most sellers know two numbers per keyword: search volume and their own rank. Neither says WHERE a keyword breaks for you. That is exactly what Amazon's search query data delivers — as a complete funnel for every individual query. Read it properly and you stop guessing which problem matters right now.
It requires Brand Registry with the brand representative role; the reseller role grants A+ and Brand Stores but not Brand Analytics (evidence level: Amazon, see Listing L16). Without a brand this tool is unavailable — then Growth L4 with the business reports is your substitute, only coarser. That lesson introduced the funnel; this one turns it into a prioritised work list.
1What the report gives you — and what it is not
The Search Query Performance report shows four funnel stages per query and your share of each (evidence level: Amazon):
| Stage | What it counts | Your share answers |
|---|---|---|
| Impressions | How often results were served for this query | How visible am I in this search at all? |
| Clicks | How often those results were clicked | How often is MY tile the one chosen? |
| Cart adds | How often a product went into a basket | Does my detail page convince after the click? |
| Purchases | How often a purchase happened | Does the basket make it through checkout for me? |
Three properties separate this report from everything else you have:
- It is keyword-level, not ASIN-level. Sessions in the business report tell you visitors dropped — this tells you which query lost them.
- It shows the market, not only you. The totals per stage belong to the query as a whole; your share is your slice of it.
- It is temporally coarse. Data arrives weekly, monthly and quarterly, not same-day. That is an advantage: daily figures would be too noisy to act on (Growth L11).
How long Amazon keeps the history is not officially documented; roughly 18 months is reported (evidence level: practice). Practical consequence: export regularly. Year-over-year comparisons need your own copies — you notice that at the first seasonal comparison, when the data is no longer there.
Picture a shopping street. Impressions are the people walking past your street. Clicks are the ones who stop at YOUR window. Cart adds are the ones who pick something up. Purchases are the ones who reach the till. The skill is not knowing every number — it is seeing at which door people turn around. That is exactly what this report measures, street by street.
2The gaps are the diagnosis
The value is not in the four numbers but in the three transitions between them. Where your share drops noticeably from one stage to the next, that is your problem — and every gap has a different remedy:
| Gap | Meaning | What to do |
|---|---|---|
| Impression share low | You are barely visible in this search | Relevance problem: check indexing and ranking (Listing L12), is the keyword placed properly at all? Advertising as a bridge. |
| Impression share high, click share much lower | You are seen but not chosen | Tile problem: main image, opening words of the title, price impression, star rating, delivery promise — everything visible in the results list. |
| Click share high, cart-add share much lower | They arrive, they do not buy | Detail page problem: image stack, bullets, A+ content, unanswered purchase doubts, missing dimensions or compatibility. |
| Cart-add share high, purchase share much lower | The basket breaks | Checkout problem: relative price, delivery time, Featured Offer lost (Listing L14), availability. |
For the query “pepper mill ceramic grinder” the report shows the AURELO set an impression share of 11 %, a click share of 3 % and a cart-add share of about 3 % as well. The diagnosis is unambiguous and has nothing to do with ranking: visibility is there, but of eleven percent of visibility only three percent of clicks arrive — the tile loses. Working on the bullet copy here repairs a part that is not broken. What needs fixing is the main image and the opening of the title. That cart-add and click share end up level says something extra: whoever gets in does get convinced. The detail page is fine.
Compare SHARES, not absolute numbers. Absolute clicks move with search volume and season; the share tells you how you did against everyone else in the same search — with the same weather for all.
3The four false conclusions
Share data invites conclusions it cannot carry. These four cost money regularly:
- Reading branded queries as success. On searches for your own brand name your share is naturally high. Those rows belong in a separate view, otherwise they flatter the average and hide the generic queries that actually matter.
- Mixing advertising and organic. The funnel does not distinguish paid from organic impressions. A rising impression share may simply be a bigger ad budget. Always cross-check against advertising data (lesson 3).
- Over-reading small queries. With a few hundred impressions in the period, shares swing wildly. The rule from Growth L11 applies here too: small samples lie loudly.
- Confusing share with cause. A falling purchase share may be about you — or about a competitor running a promotion. The report shows movement, not its reason.
Sorting the file by “purchases” and starting at the top. The top is where the queries that already work sit — you optimise the healthy. Sort by what is there to win: large volume times large gap. Anything else is busywork.
4The rhythm: when to look
Search query data is not a daily tool. A workable rhythm looks like this:
- Monthly, 30 minutes: walk the ten most important queries per product, mark the gaps, pick one job. More than one at a time destroys attribution (lesson 1).
- Quarterly, 2 hours: quarterly data against the previous quarter and the same quarter last year. This is where seasonal shifts and slow share erosion show up that the monthly view swallows.
- After every major change: judge only in the period AFTER NEXT. The current one is incomplete, the next one mixes before and after.
After the AURELO set's new main image, the click share for “pepper mill ceramic grinder” rises from 3 to 6 % — measured on the first complete week after the swap. Impression share stays at 11 %, cart-add share follows to 6 %. Two conclusions: the tile really was the bottleneck, and doubling the click share did not overwhelm the detail page — visitor quality held. The next lever is therefore no longer the image but that 11 % impression share.
5The other Brand Analytics reports in one line each
| Report | What it is actually good for |
|---|---|
| Top search terms (search frequency rank) | Market size and seasonality per term — not your own success. The rank is a ranking, not a volume. |
| Item comparison & alternate purchase | Who your competitor really is. Often not who you think — the basis for product targeting. |
| Market basket analysis | What is bought together: bundle ideas and candidates for line extension (Growth L10). |
| Repeat purchase behaviour | Whether your product creates repeat buyers — which decides how much a first customer may cost. |
| Demographics | Rough buyer structure. Aggregated and patchy; usable for imagery, too coarse for hard decisions. |
- Brand Registry role checked: brand representative, not just reseller.
- Export routine in place — your own copies instead of trusting the history.
- Branded queries separated out of the analysis.
- Ten key queries per product noted with their three gaps.
- Ranking built on volume times gap, not on absolute purchases.
- One job per cycle, effect judged in the period after next.
- Advertising data placed alongside before celebrating an impression jump.
6Expert insight: the leverage calculation — which query first
The question that separates professionals from the merely diligent is not “where is a gap?” but “which gap is worth the most?”. That needs arithmetic, not intuition — and it fits in one spreadsheet column.
The leverage of a query is what you would gain if you were as good at this stage as you are at the stage before it:
| Step | Calculation | AURELO example |
|---|---|---|
| 1. Size the gap | Previous stage share minus this stage share | 11 % impressions − 3 % clicks = 8 points |
| 2. Convert to units | Gap × the stage total for the period | 8 % × 40,000 clicks on the query = 3,200 clicks |
| 3. Convert to money | × your conversion rate × contribution per unit | 3,200 × 12 % × €8.20 ≈ €3,150 per period |
| 4. Discount to realism | × the achievable share of the gap (rarely above 50 %) | ≈ €1,570 — and THAT is the number you rank by |
These four lines order your jobs more honestly than any gut feeling. A large gap on a small query drops down the list, a small gap on a giant term rises. Three rules keep the maths honest:
- A gap never closes fully. A 100 % share at any stage is impossible — competitors exist. Calculate without a discount and you promise yourself twice the truth.
- Gaps earlier in the funnel weigh more. A percentage point of impression share carries every stage after it; a point of purchase share carries only itself. At equal euro value, fix the earlier stage first.
- Cheapest gap first. A new main image costs a day. A better impression share costs months of ranking work or permanent ad budget. At similar leverage the cheaper action always wins — leverage divided by effort is the real ranking.
One overlooked special case deserves its own line: queries with a high impression share and low volume. They look magnificent in the report — you dominate! — and earn almost nothing. They are still valuable, but as evidence rather than as a job: your relevance demonstrably works there. Compare how those terms sit in your listing with the terms where you are invisible. The difference between the two groups is the most concrete instruction this report can give you — far more concrete than any general keyword rule.
Shares can rise while your business shrinks: if the market drops 30 % and you only lose 20 %, your share goes up — and the report looks better than the bank account. So every share column belongs next to the query's absolute total. Share without volume is a number that keeps you content in a shrinking market.
The pro track assumes the other three and repeats nothing from them. It opens once you have completed all three in full — every lesson quiz at 80 % or better, and every final quiz passed.
Not logged in? Your progress only counts with a free account.
Check yourself
6 quick questions — one at a time, instant feedback. With a free account your progress is saved.
Frequently asked
I have no Brand Registry — what replaces the report?
Business reports plus your own ranking measurement. That gives you sessions and conversion per ASIN but not per query — you have to assemble which term breaks from ranking histories and advertising data. It is coarser and slower, but it replaces the diagnosis in direction if not in precision.
How many impressions does a query need before the shares are reliable?
There is no official number. It becomes usable in the four-digit range per period; below that, shares swing so much that any movement can be chance. For small terms read quarterly rather than weekly data — the longer period substitutes for the missing volume.
Can I read from the report whether a change worked?
Only with discipline: one change per cycle, and judge only in the period after next. The current period is incomplete, the immediately following one mixes before and after. Break those two rules and you are measuring your own impatience.
The free Listing Check scores any ASIN from 0 to 100 in a minute — the fastest way to see whether the mechanics from this lesson actually hold on your own listing.
Go deeper: the complete guide to optimizing Amazon listings →
Everything in this academy comes from day-to-day selling practice — the same playbook behind Listimo, the tool that turns product photos into complete Amazon listings.