Before increasing an external campaign’s budget, reconcile three things: which ads were measured, how much those ads actually cost, and whether the Amazon sales window is mature enough to evaluate. Then calculate the contribution left after product costs and campaign-specific expenses. A profitable-looking campaign can change direction when a missing ad account, reused tag or creator invoice enters the calculation.
The practical output is a short decision record: keep the current budget, run a bounded test, or stop spending until a specific problem is resolved. It should explain the numbers behind that decision and identify what remains uncertain.
This workflow is for Amazon sellers sending traffic from external advertising to their Amazon listings or Store. It assumes an existing Amazon Attribution setup. The reconciliation method and decision controls below are analytical recommendations, not Amazon requirements or a claim that one reporting tool can prove advertising caused incremental sales.
What exactly are you trying to reconcile?
Reconcile a defined campaign and reporting period, rather than trying to make every dashboard total identical. Each source has a different job. The advertising platform records media delivery and spend. Amazon Attribution measures shopping activity associated with tagged traffic. Product-cost records explain the economics of those sales. Invoices capture expenses that media-spend totals may omit.
Choose one marketplace, currency and campaign first. Write down the selected reporting dates, export time, account IDs, campaign IDs and the sales metric being used. Keep campaign names for readability, but do not rely on names alone: names can change, and different accounts can contain identically named campaigns.
Your working file needs a place for every relevant spend source, including sources that have no matched Attribution record. An unmatched row is an exception to investigate. Deleting it makes the report cleaner while making the decision less reliable.
| Question | Record to check | What counts as a resolved check |
|---|---|---|
| Which traffic belongs here? | Live ads, account and campaign IDs, destination links | In-scope ads have a documented measurement route |
| Was all media spend included? | Platform delivery exports and connected-account list | Included spend plus explained exclusions matches the source total |
| What sales were measured? | Amazon Attribution export and metric definitions | Same campaign scope and clearly stated product coverage |
| Is the period ready to judge? | Window, reporting basis and extraction time | Mature period separated from provisional activity |
| What did the campaign cost beyond media? | Creator, agency and production invoices | Relevant costs recorded once, with an allocation basis |
Do this at the level where you will make a decision. If several ads share one measurement tag and cannot be separated reliably, a campaign-level decision may be supportable when a creative-level winner is not.
How do you prove that the right ads and tags are included?
Start from the ads that actually ran, then trace them to measurement. Starting only from the Attribution dashboard can hide campaigns whose links were never tagged.
Export delivery and spend from each advertising account used for the campaign. Include paused and archived ads that spent money during the selected period. Add any agency-managed account, second brand account or boosted post paid for outside the account you usually review. Check who controls each account and who can provide the complete export.
For each in-scope ad or creative group, record the destination URL, the Attribution campaign or tag it is intended to use, and the date the link was checked. Inspect the live destination and any redirect path, not just a planning spreadsheet. If someone replaced the URL midway through the campaign, record the affected interval separately.
Amazon’s Attribution guide describes tags as links between external clicks and later shopping activity. For traffic discrepancies, it recommends checking date ranges, tagged-ad scope and unexpected tag reuse.
A useful test is to follow a sample of live links through the routes customers actually use. Confirm the intended marketplace, product or Store destination, and preservation of the expected tag parameters. A successful link test verifies routing; it does not guarantee that every customer journey will be observable.
Record three separate outcomes: correctly mapped, intentionally grouped, and unresolved. For an intentionally shared tag, document the combined spend that belongs to it. For a wrong or missing tag, preserve the original spend and label the affected period as incompletely measured. Fixing the link today does not justify inventing historical attributed sales.
How do you check spend completeness without counting costs twice?
Match spend to the campaign’s delivery period and scope, then explain the difference between that spend and cash billed by the platform. An invoice may include a different billing cutoff, credits or taxes. It is evidence for a bridge, not automatically the right media-spend denominator.
Suppose the platform export shows $1,850 for a campaign: $1,400 in the main account and $450 in an agency account. A report that sees only the main account is missing 24.3% of the campaign’s media cost: $450 ÷ $1,850 × 100. The difference remains important even if the missing account produced few measurable sales.
In sellerboard, the campaign-level Attribution profit view can match daily spend from connected advertising accounts to Attribution data by campaign, ad group and creative. Supported connections include Facebook, TikTok, Google, Snapchat, Pinterest and Bing. Verify that every relevant account is connected and compare its spend with the platform export before relying on the combined result.
For this check, use a simple spend bridge:
In-scope media spend = matched spend + identified unmatched spend
Keep excluded spend in a separate column with a reason, such as a campaign serving another marketplace. The sum of included and excluded rows should still tie to the source export. A small unexplained total can conceal a large error in one creative if another row offsets it.
Then review duplicate paths. Has a manually entered external-ad expense remained after the same cost began arriving through a connection? Does the agency invoice include media already paid directly to the platform, or only the management fee? Does a single production invoice support several campaigns? Resolve those definitions before subtracting anything again.
Do not assume a dashboard number includes every cost merely because its label says profit. Understand its cost basis, and add only expenses that are genuinely absent from the specific analysis you are making.
When is the Amazon sales window mature enough to use?
Use recent data for monitoring, and an older, sufficiently developed period for a scaling decision. Amazon Attribution uses a 14-day last-touch model: qualifying conversions can occur within 14 days after a click, and credit goes to the most recent qualifying click. Recent activity can therefore continue accumulating conversions.
Save an initial export and revisit the same defined reporting slice after its attribution window has elapsed and the available reporting has updated. Record both extraction times. Check whether the export organizes activity by click date, conversion date or another reporting basis before treating a selected date range as a fixed click cohort. Matching the calendar selectors alone does not establish identical populations.
A practical review can have two lanes. The live lane catches obvious problems such as spend running with broken links or an unavailable product. The mature lane evaluates campaign economics without confusing incomplete conversions with a final result. This does not require leaving a clearly broken campaign running while waiting for sales to arrive.
Avoid choosing a universal “safe” reporting lag beyond the stated attribution window. Reporting updates, returns and billing adjustments can still change the picture. Label which costs and sales are final enough for the decision, and name the next review trigger.
If you compare two campaigns, give their data equivalent time to develop. A campaign that ended two weeks ago and one launched yesterday are not comparable simply because both appear in this month’s dashboard.
Which Amazon sales belong in the calculation?
Use one explicitly named sales definition and match product costs to that definition. Amazon’s reporting definitions distinguish sales of promoted products from total product sales, which also include other products from the same brand. Those columns answer different questions; adding them together would double count the promoted-product component.
For a campaign designed to bring shoppers into a broader brand range, total product sales can be useful. The associated costs still need the actual product mix. Applying the advertised product’s margin to every attributed dollar can overstate or understate the result when customers buy different products.
Separate units from purchases, too. A purchase event can contain more than one unit. Cost of goods sold (COGS) and fulfillment economics normally need unit quantities and product identity, not a purchase count multiplied by the hero product’s cost.
The profit calculation on sellerboard’s Attribution page subtracts ad spend, COGS, Amazon fees and applicable value-added tax (VAT) from attributed sales. The feature requires Amazon Brand Registry and existing Attribution campaigns and tags; it does not create those links for you. Verify your product-cost inputs and avoid adding the same fee or tax adjustment a second time.
For a manual reconciliation, state how you handle refunds, promotions, other advertising costs and any referral credit. Include a credit only when its amount and treatment are supported by your records. A possible future credit should not be the unexplained number that turns an otherwise losing campaign into a winner.
What does a complete reconciliation look like in numbers?
The following hypothetical example uses one US campaign, USD, and a single consistent attributed-sales definition. It is a decision worksheet, not a reproduction of a particular dashboard or a published performance benchmark. Product costs below already include the applicable selling costs and identified refund adjustments, but exclude external media, agency fees and creative production.
At the first review, the seller sees $6,000 in attributed sales and $1,400 in connected media spend. Product-related costs are $3,900. The apparent contribution is $700: $6,000 − $3,900 − $1,400.
The reconciliation finds $450 of campaign spend in a second account. A later export of the same reporting slice shows an additional $900 in attributed sales and $585 in associated product costs. The campaign also incurred a $300 management fee and $250 of creative production.
| Reconciliation step | Change | Running contribution |
|---|---|---|
| Initial sales less product costs and visible media | $6,000 − $3,900 − $1,400 | $700 |
| Add missing media spend | −$450 | $250 |
| Add later reported sales and their product costs | +$900 − $585 | $565 |
| Add campaign management fee | −$300 | $265 |
| Add creative production cost | −$250 | $15 |
The completed arithmetic is:
Reconciled contribution = $6,900 − $4,485 − $1,850 − $300 − $250 = $15
The initial view overstated the eventual result by $685, even though additional sales arrived later. Waiting for conversions alone would not have solved the missing-cost problem.
Media return on ad spend (ROAS) is still approximately 3.73: $6,900 ÷ $1,850. That revenue ratio does not convey the campaign’s $15 remainder, which is only about 0.22% of attributed sales. The $15 is attributed campaign contribution before unallocated business overhead, not complete business net profit. Neither figure proves that the campaign created sales that would otherwise not have happened.
The bridge makes the operational diagnosis specific. The seller needs to fix account coverage, decide how much creative must be purchased again, and test whether another block of spending earns sufficient contribution. “The campaign has a good ROAS” no longer settles the question.
Should you keep spending, run another test or stop?
Base the next decision on complete historical economics and the costs of the next action. Keep those two views separate so an already-paid creative invoice is neither forgotten nor charged repeatedly to every future test.
In the example, the contribution before external media and campaign-specific expenses is 35% of attributed sales: ($6,900 − $4,485) ÷ $6,900. Suppose the next test would spend $500 on media and incur an additional $50 management fee, with no new creative production. At the same product mix and cost structure, the test needs about $1,571.43 of attributed sales just to cover those additional costs: ($500 + $50) ÷ 35%.
That is a planning assumption, not a promise that the next $500 will behave like the last $1,850. The implied media ROAS threshold is approximately 3.14. If the product contribution rate falls to 30%, required sales rise to $1,833.33, or about 3.67 media ROAS. Price changes, product mix and returns can consume a thin apparent margin quickly.
| Decision | Evidence that would support it | Guardrail to record |
|---|---|---|
| Keep the current budget | Complete, mature data and an acceptable contribution | Review date and maximum tolerated loss |
| Run a bounded test | Economics could work, but a key assumption remains uncertain | Spend cap, changed variable and success threshold |
| Pause to repair measurement | Missing account, broken link or unresolved mapping changes the answer | Exact repair and verification needed before restart |
| Stop the campaign | Complete data fails the seller’s required economics | Preserve results and the reason for stopping |
Set thresholds before seeing the next result. A test can have a learning objective, but “learning” should not become an unlimited exception to the spending cap. If a campaign is intended to create brand awareness rather than immediate contribution, document that objective and its approved budget separately.
Attribution is an allocation rule. Assessing incremental effect requires a credible comparison of what would have happened without the advertising, ideally through an appropriate controlled test. A rise in total account sales during the campaign is useful context, but promotions, seasonality, inventory availability and other advertising can also explain it.
Which mistakes make the reconciliation look better than it is?
Starting from only the connected accounts. Build the account list from actual spending responsibility, including agency and secondary accounts.
Forcing traffic and conversion totals to agree. Investigate differences in scope, reporting windows and measurement definitions; do not overwrite one system’s data to match another.
Removing unmatched spend. Keep it in the campaign’s cost total when it belongs there, and label the measurement gap.
Adding overlapping sales columns. Promoted-product sales and total product sales are alternative scopes, not independent revenue pools.
Treating production costs inconsistently. Show the completed campaign with its full costs, then state which costs recur for the next decision.
Scaling from a small positive remainder. Stress-test product costs and refunds before committing more money. A rounding error, missing invoice or change in product mix can erase a narrow buffer.
What do sellers commonly ask about Attribution reconciliation?
Can I reconcile a campaign if some ads were not tagged?
You can reconcile its spend and document the measurement gap. You cannot reliably reconstruct missing historical attributed conversions merely by applying the conversion rate of tagged ads. Repair the links and use a clearly marked new measurement period.
Must publisher clicks equal Amazon Attribution clicks?
No. Begin with matching dates, tagged-ad scope and tag implementation. Different counting rules and customer journeys can leave a residual difference. A discrepancy should have an explanation or an open investigation, rather than being silently treated as wasted spend.
Does zero attributed revenue prove the campaign generated no value?
No, but it also does not justify assuming hidden profitable sales. Verify tags, product coverage and reporting maturity, then decide whether a capped test or pause is appropriate.
Can I use campaign profit to forecast the return from doubling spend?
Only as an assumption to test. Additional spend may reach different shoppers or sell a different product mix. Reconcile the next spending increment and compare it with the threshold set beforehand.
How do I know the reconciliation is finished?
All in-scope spend is accounted for, the sales and cost definitions align, the period’s maturity is stated, and every material exception is resolved or explicitly carried into the decision. Record the decision, owner and next check so the same uncertainty does not reappear at the next budget review.
A campaign is ready for a larger budget when you can explain the path from paid traffic to measured sales to the contribution left over, and when the next spending increment has a defensible test. Reconciliation makes that explanation possible without pretending that every attributed sale is incremental or every dashboard total is final.
