Choose Amazon PPC reports by the decision you need to make and what each row represents. Use campaign or portfolio exports for allocation, and seek more detailed traffic or placement dimensions when the question requires them. Keep those datasets separate: the same advertising activity can appear at different levels, so adding their totals can count it repeatedly.
This is a report-selection framework, not an exhaustive catalog of report types or a search-term analysis tutorial. Official export and metric guidance was data checked 2026-11-07. The workflows below are based on public documentation, not an advertising account we operated.
Amazon PPC reports: choose by question and grain
Grain means the combination of dimensions that defines one row. A campaign summary, a campaign-day row, and a campaign-target row are different grains even when all contain clicks and cost.
Write the business question before downloading anything. Then choose the least detailed dataset that can answer it, plus a drill-down only when you need to explain the result.
| Business question | Minimum useful grain or dimension | Decision it supports | What it cannot establish alone |
|---|---|---|---|
| Where is spending concentrated? | Campaign or portfolio over a defined period | Which area needs investigation | Which individual query is responsible |
| Did performance change over time? | Same campaign scope, separated by reporting date | When to investigate a change | What caused that change |
| Which targeting instruction needs attention? | Campaign, ad group, and target context | Which target to review | The exact shopper phrase from the target name alone |
| What traffic context produced the clicks? | Search-term detail with its targeting context | Which observed traffic to examine | Every unclicked search opportunity |
| Is a placement adjustment worth investigating? | A placement dimension for the relevant campaign scope | Where to compare cost and outcomes | The effect of a bid change before it is tested |
| How should a business group be reviewed? | Portfolio label plus campaign identity | Allocation across grouped campaigns | A separate set of sales to add to campaign totals |
These are data requirements, not a promise that every ad product offers identically named reports or columns. Check the options and definitions available for your ad product and account. If the necessary dimension is unavailable, keep the question unresolved rather than treating an aggregate as the missing detail.
Amazon’s performance metrics help, checked 2026-11-07, describes campaign-manager controls for columns, date range, and export. Amazon’s portfolio report help, checked 2026-11-07, documents exports from campaign or portfolio views and campaign and keyword exports with a portfolio column.
For query-level interpretation, continue to the Sponsored Products search term report guide. This page stays focused on choosing and reconciling the datasets, rather than repeating its column-by-column tutorial.
Export a defined dataset, not an unexplained total
A spreadsheet needs an identity card. Without one, a later reviewer cannot tell whether a change reflects advertising performance or simply different filters.
Amazon’s portfolio export guidance, checked 2026-11-07, says a downloaded report includes the data filtered in the table at the time of download. It also describes using Columns and Customize columns to choose performance metrics before exporting. Treat the visible filters as part of the dataset definition.
Record this information alongside each raw file:
| Identity-card entry | What to preserve |
|---|---|
| Account and marketplace | Which advertising population the file describes |
| Ad product and scope | Campaigns, portfolios, products, or other included entities |
| Reporting period | Selected start and end dates, plus time zone where specified |
| Row definition | The dimensions that make a row distinct |
| Monetary unit | Currency associated with cost and sales |
| Outcome definition | Exact purchase, unit, and sales fields and their attribution scope |
| Snapshot | Export timestamp and any known reporting delay |
| Filters | Included and excluded statuses, names, and other selections |
Keep the original column names in the raw file. If you create friendlier labels in an analysis tab, retain a mapping back to the source. Renaming Purchases, Units, and Sales to the same conversions label removes distinctions you may need later.
Amazon’s portfolio help also documents a Sponsored ads reports route through Measurement & Reporting, followed by Create report, report configuration, and Run report, checked 2026-11-07. Use the current options in your account to define the export; this public sequence does not establish a universal list of available types, lookback periods, or delivery times.
Save the unmodified download separately from the working analysis. A reusable filename can contain the account, marketplace, reporting period, scope, and snapshot date. The goal is to recover the exact evidence behind a decision without relying on somebody’s memory of the dashboard.
Keep campaign, target, and search-term rows separate
Different views of the same activity are not independent activity. A campaign summary and a target breakdown may both describe the campaign’s cost. Combining the summary row with the breakdown rows does not create a larger valid spending total.
Use one authoritative dataset for each headline measure. Then use the other datasets as explanations or reconciliations, not extra contributions to that headline.
A practical workbook can separate:
- Campaign totals: the scope used for the overall review.
- Target detail: the instructions associated with performance inside that scope.
- Traffic detail: observed search contexts, interpreted under that report’s inclusion rules.
- Portfolio labels: a grouping attribute for campaigns where available.
Do not assume similarly named columns give every table identical coverage. The search term report guide explains why a clicked-traffic dataset is not an inventory of every impression. A detail table with narrower inclusion rules is not automatically expected to reproduce every parent metric.
Guard against duplicate joins
Before merging two exports, decide which columns make a row unique in each. Prefer stable identifiers when available, and preserve names as readable labels. If identifiers are unavailable, document the full contextual key rather than assuming a campaign name alone is globally unique.
A common spreadsheet error is joining one campaign total to several target rows, then summing the copied campaign cost. The campaign value has been repeated, not newly measured. Keep that total outside the target sum, or aggregate the target table to the required compatible grain before joining.
Daily data introduces another trap. A period total and its daily components represent alternative summaries of the period. Keep time grain explicit and avoid stacking both into a single pivot as though they were disjoint observations.
After any merge, compare the unchanged headline measure against its original source. If the merge increased total cost without adding new, nonoverlapping activity, inspect the relationship between the tables before making a budget decision.
Align reporting clocks before comparing outcomes
An export is a snapshot, not necessarily a final settlement of recent performance. Keep the selected reporting dates, the metric’s attribution definition, and the retrieval time separate.
Amazon’s performance metrics documentation, checked 2026-11-07, says sales data can take up to 48 hours to populate and can be incomplete in Yesterday. Its Purchases definition separately notes up to 12 hours for order metrics to update, with possible delays in Today. The same page says Cost can be updated until three days after the click or viewable impression during traffic validation.
Those statements describe reporting updates for particular metrics. They do not define one attribution window for every campaign type. Preserve the definition attached to the actual field you export instead of labeling all purchases with a universal window.
For a recurring review, compare elapsed periods with comparable opportunity for attribution to develop. Retrieve both comparison datasets using the same definitions and keep their snapshot dates visible. A newer snapshot of an older period may differ from a file saved earlier even though you selected the same dates.
Keep purchases, units, and sales value separate too. The PPC conversion rate guide explains why a purchase-per-click measure is not interchangeable with units per session. Choosing a convenient column changes the question, even if the spreadsheet formula still works.
Build a repeatable review packet
A useful reporting packet has a headline, an explanation, and a next action. Downloading every available export creates more reconciliation work without necessarily improving a decision.
Start with a campaign-level baseline for the selected scope. Identify a specific change or concentration that matters, then request only the dimensions needed to investigate it. Keep the raw download, identity card, analysis, and action record together.
| Review finding | Supporting detail to seek | Action record |
|---|---|---|
| A business group needs allocation review | Campaigns with portfolio labels | Which group, which campaigns, and why |
| A campaign spends without mature purchases | Relevant product and traffic detail | The hypothesis and affordable exposure |
| Performance differs across placement contexts | Compatible placement detail | Proposed adjustment and later review point |
| Several entities need an operational change | Verified entity list and intended settings | Separate analysis evidence from editing instructions |
Amazon’s portfolio help, checked 2026-11-07, says campaign performance history stays at campaign level. Use portfolio labels to organize the review without treating a portfolio export as a second independent source of sales. The PPC portfolios guide covers the grouping workflow.
If the finding concerns placement, move from the comparison to a defined test using the placement adjustments guide. A report can identify an association worth investigating; it does not by itself prove that a particular adjustment will improve the result.
Keep reporting exports distinct from operational changes. Before translating an analysis into multi-entity edits, use the separate Amazon Ads bulk operations guide and verify the intended entities and settings. An analysis workbook is evidence for a change, not proof that the change has been applied.
Troubleshoot disagreement between exports
When two numbers disagree, first ask whether they were supposed to match. Reconcile their identity cards before looking for a reporting fault.
Work through these checks in order:
- Scope: Are the account, marketplace, ad product, campaigns, and filters identical?
- Time: Are the reporting dates, time grain, and retrieval dates compatible?
- Metrics: Are both files using the same cost, purchase, and sales definitions?
- Coverage: Does the detail dataset include the same activity as the summary?
- Transformation: Did a merge repeat rows, a pivot hide categories, or a formula include a total twice?
Recreate a small export with matched settings if the difference remains unexplained. Start with a single campaign and an elapsed reporting period. A narrow comparison is easier to inspect than an account-wide workbook containing several ad products and currencies.
Document an expected difference instead of forcing totals to agree. For example, a filtered table export and an unfiltered baseline describe different populations. Amazon’s filtered-export guidance, checked 2026-11-07, supports checking that distinction before treating the discrepancy as missing activity.
End the review with one of three outcomes: compatible datasets reconciled, a documented scope difference, or an unresolved discrepancy with the raw files preserved. Only the first two support a confident interpretation. An unexplained gap should remain visible in the decision record, not disappear inside an adjustment formula.
Frequently Asked Questions
Which report should I download first?
Begin with the campaign or portfolio view that matches the allocation question. Add finer detail only when a specific finding requires it. A small, defined reporting packet is easier to reproduce than a folder of unrelated exports.
Can I use one report for every ad product?
Use common review questions, but verify the dimensions and metric definitions for each ad product. Similar column labels do not guarantee identical attribution, coverage, or available detail.
Does downloading a report change my bids?
The documented export steps retrieve performance data. Keep any later operational change as a separate, verified workflow, with its own record of affected entities and settings.