The search term report for Amazon PPC is a clicked-traffic report, not a complete list of every query where your ads appeared. For Sponsored Products, Amazon includes search terms that generated at least one ad click; the customer search term can also represent an ASIN or a term inferred in a non-search context. Read the row’s targeting, reporting period and attribution fields before treating it as a shopper’s typed query or a finished sales result.
This guide explains report interpretation rather than negative-keyword selection. For that workflow, use how to find negative keywords for Amazon PPC; for account-wide checks, use the Amazon PPC audit.
Documentation checked November 6, 2026: definitions below come from Amazon’s Search term report for Sponsored Products, using its public help text captured for this date. The help page carries a May 18, 2026 update date. This is a documentation-based reference, not a report exported from a seller account.
Search Term Report Amazon PPC: Define the Dataset First
Amazon’s help states that only search terms resulting in at least one ad click are included, checked November 6, 2026. Consequently, report impressions may not match campaign manager impressions. The report can contain impressions associated with its included terms without being an inventory of all impressions across all terms.
That distinction changes what you can conclude from a missing row:
| What you see | Supported interpretation | What it does not establish |
|---|---|---|
| A term has a report row | It qualified for inclusion in this report’s scope | Every appearance was a literal typed search |
| A term is absent | No corresponding qualifying row appears in the selected data | Nobody searched it, or the ad never appeared for it |
| Report impressions differ from campaign impressions | The datasets may have different inclusion rules | The campaign total is necessarily wrong |
| Clicks exist but sales are zero | No sales are recorded in those selected fields at export time | The traffic can never produce an attributed sale |
| The same term appears more than once | More than one reporting context may be represented | The rows are automatically duplicates |
Before doing keyword analysis, write down the ad product, account, marketplace, currency, date range, time unit and export date. A comparison across mismatched scopes can look like a missing-data problem when it is actually a different dataset.
If you are investigating no delivery, start with campaign-level impressions instead. A report designed around clicked terms cannot answer the separate question of whether every unclicked opportunity received an impression.
Read the Row as Targeting Plus Observed Traffic
Targeting describes the instruction; Customer Search Term describes the reported traffic context. Amazon’s column definitions identify Targeting as the campaign targeting clause and Customer Search Term as the term used by a customer or inferred by Amazon in non-search contexts, checked November 6, 2026.
Keep both fields. Replacing the target with the query destroys the connection between what you selected and what the campaign reached.
| Field or field group | What it contributes | Reading rule |
|---|---|---|
| Campaign Name, Ad Group Name | Advertiser-defined grouping | Preserve them when comparing repeated terms |
| Targeting | The targeting clause | Do not relabel it as the shopper’s query |
| Match Type | Keyword matching context | Interpret it with the target, not as a standalone explanation |
| Customer Search Term | Reported search or inferred context | Check whether it is text or an ASIN |
| Start Date, End Date | Summary reporting period | Keep them with an aggregated export |
| Impressions, Clicks | Reported exposure and interaction | Remember the report’s click-based inclusion rule |
| Cost or Spend, Currency | Advertising expenditure and its unit | Keep currency attached to every monetary calculation |
| Purchases, Units, Sales | Different conversion measures | Preserve the exact column name and attribution definition |
| Promoted and halo fields | Different purchased-product relationships | Do not treat every attributed sale as the advertised SKU |
For a hypothetical lunch bag campaign, a selected keyword might be insulated lunch bag while a reported term is small insulated lunch bag. Those values answer different questions. The first identifies the targeting instruction; the second describes traffic associated with the row. The PPC match types guide explains matching behavior without turning these two fields into synonyms.
When a column is unavailable or blank in your export, keep that absence explicit. Do not manufacture a keyword match type for a product-targeting row just to make a spreadsheet filter work.
Why ASINs and Non-Search Terms Appear
An alphanumeric Customer Search Term entry is not automatically corrupted data. Amazon’s help explains that ASIN-related entries correspond to ASINs and the associated product detail pages where the ad displayed, checked November 6, 2026. The help limits these entries to Sponsored Products automatic-targeted campaigns and product-attribute-targeted ad groups.
Treat an ASIN as a product-context identifier, not a phrase to split into words. Retain the original value and label its row type separately. The product targeting guide covers the targeting decisions that follow; this report reference stops at identifying what the row represents.
Text entries need care too. Amazon’s current help says search terms can be inferred as a best match to the product in non-search contexts, giving off-Amazon social-site placements as an example, checked November 6, 2026. Therefore, the column name does not guarantee that a person typed the displayed phrase into Amazon search.
Keep three interpretations separate:
- Search context: a customer-used term as described by the report.
- Product context: an ASIN associated with a product detail page placement.
- Non-search context: a term inferred by Amazon rather than necessarily typed.
The definition alone does not give you a reliable row-by-row placement classifier for every text entry. If the export does not identify which context applies, leave that distinction unresolved. Do not assign an entire text row to Amazon search, social traffic, or a particular placement merely because its wording sounds plausible.
Daily, Summary and the 65-Day Lookback
Amazon lists summary and daily as the report’s two time units and gives a 65-day lookback window, checked November 6, 2026. That retrieval limit is not the sales attribution window. One concerns how far back you can request report data; the other concerns which purchases count after an ad click.
Choose summary for a period-level reading. Choose daily when the question requires a date-level comparison, such as checking whether traffic changed around a targeting edit. Do not combine the daily rows with a summary covering those same dates and then sum both: they represent overlapping activity, not additional traffic.
For ongoing analysis, archive exports on a regular schedule rather than assuming old data will remain retrievable indefinitely. Store the untouched file alongside its request settings. A refreshed file and an older export for the same dates are different snapshots; preserve their export dates so you can investigate changes instead of silently overwriting the evidence.
If a term appears in several campaigns or ad groups, decide what you want to compare before grouping it. Account-wide totals can be useful, but preserve the underlying campaign and target context for any later action. Otherwise, one combined term total can conceal very different targeting instructions.
Attribution: Keep Window, Product Scope and Metric Together
A conversion field is not fully defined by the word Sales. Amazon’s help includes 7-day definitions and separately named 14-day fields, checked November 6, 2026. Because the public text includes seller and vendor sections, inspect the exact columns available for your account rather than assuming every export has identical fields.
The help defines Sales (promoted) around attributed orders where the purchased SKU matches the advertised SKU within seven days of the click. It defines Sales (halo) around purchases of a different SKU within that seven-day window. It also lists 14 Day Advertised ASIN Sales and 14 Day Total Sales with their respective 14-day definitions.
These distinctions matter when comparing results:
- Promoted versus halo: a sale attributed to advertising need not be a sale of the advertised SKU.
- Orders versus units: a count of purchases and a count of units measure different things.
- Seven versus fourteen days: different windows admit different conversion opportunities.
- Recent versus older clicks: a recent click has not had the same elapsed opportunity to convert as an older click.
Suppose you export yesterday’s clicks today. A zero-sales row is a snapshot taken before an entire seven-day post-click window could elapse. Preserve that timing in your interpretation instead of declaring the row permanently unproductive. This is not a promise that a conversion will appear later, nor a claim about a fixed reporting refresh delay.
Record the comparison as: metric name + attribution window + product scope + currency + export date. If those components differ between files, explain the mismatch before comparing ROAS or declaring that an optimization improved performance. Attributed revenue is also not profit: product cost, fulfillment and other business expenses are separate inputs.
Recalculate Ratios Instead of Averaging Them
Use the underlying totals when combining rows. Amazon defines CTR as clicks divided by impressions, CPC as cost divided by clicks, and ROAS as sales divided by advertising spend, checked November 6, 2026. ROAS is a multiplier, not a percentage.
The following is a hypothetical example, not seller-account data. Both rows use the same currency, attribution window and sales definition.
| Row | Impressions | Clicks | Cost | Sales | CTR | CPC | ROAS |
|---|---|---|---|---|---|---|---|
| A | 100 | 10 | $20 | $100 | 10.00% | $2.00 | 5.00 |
| B | 900 | 10 | $10 | $20 | 1.11% | $1.00 | 2.00 |
| Combined | 1,000 | 20 | $30 | $120 | 2.00% | $1.50 | 4.00 |
Combined CTR is 20 divided by 1,000, not the average of the two CTR percentages. Combined ROAS is $120 divided by $30, not the average of 5 and 2. Compute each ratio from the matching numerator and denominator after aggregation.
If a denominator is zero, retain an undefined or blank result instead of forcing a numeric zero. A missing conversion result and an undefined ratio are different observations. Also keep monetary aggregation within one currency unless you explicitly convert it using a documented exchange rate.
An Import Checklist Before You Make Changes
Preserve these columns or metadata in your working file:
- Original export filename, export date and requested time unit.
- Account, marketplace, currency and reporting dates.
- Campaign, ad group, target, match type and unmodified customer term.
- Raw impressions, clicks, cost and conversion fields.
- Exact attribution definitions used for the comparison.
- A separate interpretation column for text, ASIN or unresolved context.
Keep analysis labels outside the raw fields. A label such as relevant or irrelevant is your classification, not a replacement for the original customer term. Likewise, a grouped term total should not erase which campaign generated it.
When the dataset is understood, move to the appropriate next task: the negative keyword workflow for exclusions or the PPC audit for broader account diagnosis. Reading a row correctly comes before deciding what to change.
Frequently Asked Questions
Is the search term report the same as Search Term Impression Share?
No. Amazon’s help lists Search Term Impression Share as a separate Sponsored Products report. This guide concerns the search term report’s rows and definitions; it does not treat those rows as an account’s share of all available impressions.
Can I use missing rows to find keywords with no demand?
No. The click-based inclusion rule means an absent term is not a demand measurement. It does not establish search volume, organic traffic, or whether the campaign received unclicked impressions for that term.
Does a high ROAS prove the advertised product is profitable?
No. Check which sales field supplied the numerator, including promoted versus halo scope, and compare the result with your actual costs. A revenue-to-ad-spend ratio does not include every expense or establish incremental profit.