How to analyze an Amazon SQP report to optimize keywords (at scale)

Amazon's Search Query Performance (SQP) report shows the exact search terms shoppers used to find your products, along with impressions, clicks, click-through rate, and conversion data for each term. It's the closest thing Amazon gives you to a keyword-level view of your own catalog, and the only Amazon-native source of proven keyword revenue.

SQP data lives in Seller Central under Brand Analytics (brand-registered sellers). Unlike third-party keyword tools that estimate volume, SQP shows your real performance per search term: how often your listing appeared, how often shoppers clicked, and how often they bought, and at what price.

Every other keyword tool (Helium 10, DataDive, Jungle Scout, SellerSprite) works the same way: they estimate search volume and difficulty from sampled or modeled data, then hand you a list of keywords ranked by those estimates. That's useful for discovery: finding terms you hadn't thought of. But it's all indirect. Nobody outside Amazon actually knows what shoppers type, click, and buy.

SQP is different because it's direct data from Amazon itself. It's not an estimate or a model; it's the actual record of what shoppers did on your listings:

• What they searched: the exact query strings, straight from Amazon's search logs
• What they clicked: whether your listing (or a competitor's) got the click
• What they bought: whether the click turned into a purchase, and at what price
• What they didn't buy: the queries where shoppers saw you, clicked you, and still bought elsewhere

That last point is the one no other tool can give you. A keyword tool can tell you "fishing net" has 33,580 monthly searches. Only SQP can tell you that your family got 24,436 of those impressions, 613 clicks, and 180 purchases, and that shoppers searching "swimming pool net" are clicking your competitors instead of you. That's not a model's guess. That's what actually happened, in your account, last month.

So the hierarchy is simple: keyword tools tell you what people might search. SQP tells you what people actually searched, clicked, and bought (or didn't buy) on your listings. When you're deciding where to spend your optimization effort, the direct data wins every time.

Most sellers analyze SQP the wrong way. They look at impressions, chase the biggest search terms, and ignore the data that actually matters. Three common mistakes:

1. Analyzing the catalog as one blob. A kitchen gadget and a pet product convert on completely different terms. Segment by product.
2. Chasing impressions. A keyword with 50,000 impressions and £0 in sales is noise. A keyword with 500 impressions and £2,000 in proven revenue is gold.
3. Never re-pulling the report. SQP is a snapshot, not a one-time study. The market moves monthly.

The core insight: volume ≠ value. SQP is the only report that tells you which keywords are already making you money, and which ones could make you more. That's the method this article teaches.

This is the analysis method Perci runs in production on every catalog. It's a 7-step pipeline that turns a raw SQP export into a ranked list of what to fix first. We'll walk through it with a real report: a UK fishing-net listing family (4 color variants), July 2026.

The raw report:

258 unique search queries, 192,086 impressions, 6,851 clicks, 2,331 purchases across the family.

Step 1: Pull and combine the data

Pull the SQP report per ASIN. If your product is a variation family (like this one: green, blue, pink, and yellow nets), combine the child ASINs' data per keyword before analyzing. A shopper searching "kids fishing net" doesn't care which color they land on; the family competes as one product.

What this step pulls out: one clean row per keyword for the whole product family, instead of four fragmented rows per color.

Step 2: Clean the noise

Not every keyword in the report is worth acting on. The method filters out:

• Other brands' keywords (shoppers searching a competitor's brand)
• Misspellings ("fishing net" vs. "fishing nett")
• Foreign-language terms
• Irrelevant or incorrect keywords (queries that don't match the product)

Misspellings and foreign terms aren't deleted; they're demoted to backend search terms, where they still capture traffic without cluttering your visible copy.

What this step pulls out: a clean keyword list where every remaining term is a real, relevant opportunity.

Step 3: Calculate proven revenue per keyword

For every keyword, compute current revenue = your purchase count × median price. This is the anchor of the whole method: the keywords that are already paying your bills.

Here's what that looks like for the top keywords in the fishing-net family:

What this step pulls out: the revenue each keyword already generates. "Kids fishing net" isn't just a popular search; it's a £1,503/month keyword for this family.

Step 4: Find the three gaps (estimated upside)

This is the heart of the method. For each keyword, look for the three ways it could grow, each expressed in pounds:

Each estimate combines the size of the gap with your own conversion rates and price. The keyword's estimated upside = the largest of the three gaps, the strongest lever to pull first.

What each gap actually means, and how fixing it lifts sales:

• Visibility gap. Shoppers searching this term buy from you when they find you, but Amazon isn't showing you to enough of them. Your purchase share is higher than your impression share, which means you're under-served. Fixing it means improving your ranking for the term (better keyword placement, more sales velocity, stronger backend relevance) so Amazon shows you more often. Every extra impression you win is one you already convert at your proven rate, so the revenue estimate is grounded in what you've already demonstrated you can do.
• Click gap. Shoppers see your listing in the results but click a competitor instead. Your click share trails your impression share, which usually points at the title and main image: they're not compelling enough, or they don't match what the searcher is looking for. Fixing it means rewriting the title to match the term's intent and testing the main image. The clicks you win are clicks you already convert at your proven rate, so again, the estimate is grounded in your own numbers.
• Purchase gap. Shoppers click your listing but buy from someone else. Your purchase rate trails the query average, which points at the listing content itself: bullets, price, reviews, or A+ content aren't converting as well as the competition. Fixing it means improving the content that turns a click into a purchase. The conversions you win are conversions you're already earning clicks for; the traffic is there, the conversion just isn't.

The three gaps are really three different places in the funnel where you're leaking sales: not shown enough (visibility), not clicked enough (click), not bought enough (purchase). Each one has a different fix, and each fix lifts a different part of the funnel. That's why the method separates them: "improve your listing" is too vague to act on, but "you have a click gap on 'swimming pool net'" tells you exactly what to change and what it's worth.

Here's the gap analysis for the fishing-net family's top keywords:

What this step pulls out: the money left on the table per keyword. "Kids fishing net" already earns £1,503/month, and the visibility gap says there's another £752/month available if Amazon showed the family more often. "Swimming pool net" barely earns anything (£18) but has a £456 click-gap: shoppers see the family but click competitors. Different problems, different fixes.

Step 5: Rank every keyword with a priority score

Now combine proven revenue + upside into a single priority score, weighted by confidence:

• Confidence (High/Medium/Low) reflects how much data backs each estimate: a gap measured on thousands of impressions is more trustworthy than one measured on a handful.
• Low-volume queries are penalized so tiny keywords don't crowd out real ones.
• The top of the ranking survives; the long tail is noise.

The exact weighting behind these scores is the result of a lot of calibration against real catalogs. The concepts are simple, the calibration is the hard part.

Here's the ranked result for the fishing-net family:

What this step pulls out: a single ranked list where proven revenue and real upside beat raw volume. "Swimming pool net" has a big upside (£456) but low confidence and low current revenue, so it ranks 9th, not 2nd. The ranking tells you where to spend effort.

Step 6: Drop what's already optimized

Match each keyword against the listing's existing copy (title, bullets, description, backend search terms). Keywords already covered drop out; they're not copy opportunities, they're already working. What remains is the actual opportunity set: keywords with revenue or upside that the listing isn't currently targeting.

Here's what that looks like with the real fishing-net listing copy:

The result for this family: 123 of 258 keywords are already covered by the listing copy; they carry £8,274/month of the family's proven revenue. The 135 remaining keywords are the opportunity set, worth £1,580/month in estimated upside:

What this step pulls out: the pattern jumps out: the listing targets the core terms (fishing net, butterfly net, kids) but misses the adjacent use-cases: pool, beach, crabbing, pond, toddler. Those are worth £1,580/month and aren't in the copy at all. That's the actionable opportunity set.

Step 7: Rank the ASINs, not just the keywords

Aggregate the opportunity-set keywords up to the ASIN level. The key insight: a £500 opportunity on a £1,000/month listing outranks a £500 opportunity on a £50,000/month listing. The opportunity score weighs each listing's upside against its existing sales, so you fix the listings where the upside matters most relative to what they already earn.

Here's the ASIN-level ranking for the fishing-net family, using only the opportunity set:

What this step pulls out: which listing to fix first. The green net has the highest opportunity-to-sales ratio (score 62.6), so it's the clear first move. The pink net has the largest raw opportunity (£911) but a lower score because it's a bigger seller, so the same £911 matters less relative to its existing revenue. The score normalizes for size so you fix the listings where the upside matters most.

Run all 7 steps and you get:

• £9,126/month in proven revenue already attributed to specific keywords
• £6,135/month in estimated upside across all keywords: the money left on the table
• 123 keywords already optimized (in the listing copy); they're working, don't touch them
• 135 keywords not in the copy: the opportunity set, worth £1,580/month: pool, beach, crabbing, pond, toddler
• A ranked ASIN list (green net first, with the highest opportunity-to-sales ratio)
• A clear fix per keyword: visibility gap → improve ranking/impressions; click gap → improve title/CTR; purchase gap → improve conversion; not-in-copy → add to title, bullets, or backend search terms

That's the difference between "analyzing your SQP report" and using it.

The manual version of this method works for a handful of listings. It breaks at catalog scale: 100, 500, or 1,000+ ASINs, each with their own SQP data, each needing all 7 steps.

That's what Perci automates, end to end:

1. Perci pulls SQP reports automatically for every listing in your catalog, so you never export, combine, or clean anything by hand.
2. Perci runs this exact analysis for you: combine, clean, revenue, gaps, rank, dedupe, ASIN priority. The opportunity report merges SQP with your Sales & Traffic data, so the ranking reflects your actual business, not generic estimates.
3. Perci layers in traditional keyword research alongside SQP: root keywords and competitor listings. SQP finds the hidden terms shoppers already use for your products; keyword research finds the terms you should be targeting that you haven't shown up for yet. Together they build a complete keyword list, not just an SQP slice.
4. Perci updates every listing at scale to take advantage of the hidden SQP keywords: the new terms flow into titles, bullets, and backend search terms across the whole catalog, not one listing at a time.
5. Perci publishes the changes directly to Amazon and monitors them: it confirms each listing updated correctly, flags anything that failed, and tracks rank so you can see whether the changes moved the needle.

In short, Perci does everything in this article for you: it finds the keyword opportunities, fixes the copy in bulk, publishes it, and monitors the results. You review and approve; the pipeline does the rest.