Amazon Ad Spend Rose. Was It More Clicks or Pricier Clicks?

Start with a bill that can be compared

Pull two completed, equal-length periods from an authorized Amazon Ads campaign report for the same account, marketplace, currency and ad product. Retain campaign ID, impressions, clicks, spend and any attributed-sales field actually present. Record retrieval time, report date basis and attribution settings. Amazon seller reports accessed through SP-API do not supply an Ads campaign ledger merely because they contain seller sales. Separate Ads authorization is required.

Through 1stPage’s MCP connection, advertising_analytics can return campaigns and campaign_comparison datasets for an authorized Ads account. Ask Claude to recompute spend divided by clicks from those rows, then apply the volume-and-CPC bridge below. If the selected 1stPage account has no Ads capability, Seller Central sales rows cannot stand in for the missing clicks.

Sum spend and clicks across included campaigns for each period, then divide total spend by total clicks to get account CPC. Averaging campaign CPCs without click weights produces a different number. Check the report against its own subtotal and mark exclusions: a product type left out, a second currency, a campaign launched halfway through the week. If the two periods differ in coverage, stop the account-wide comparison or label it partial. A higher account CPC can reflect a shift toward expensive campaigns even when their individual CPCs barely move.

Price the extra clicks, then the new CPC

Synthetic example, one currency: Week A records 1,000 clicks at $1.20, so spend is $1,200. Week B records 1,200 clicks at $1.50, so spend is $1,800. The $600 change has a clean sequential bridge. At the old $1.20 rate, the extra 200 clicks cost $240. Apply the $0.30 CPC increase to all 1,200 Week B clicks and it adds $360. $1,200 + $240 + $360 = $1,800. Choosing the old CPC for the volume step defines this bridge; reversing the order would assign the interaction differently while preserving the same $600 total.

Do this first for campaigns present in both periods. Give newly launched campaigns their own new-spend bucket and ended campaigns a negative removed-spend bucket; neither has a defensible two-period CPC bridge. Sum the stable-campaign volume effects and CPC effects with those two buckets. The total must match the account spend movement. Only then inspect impressions and clicks per impression, alongside documented bids, budgets and targeting changes. The bridge says where the money moved, not why auction prices or shopper response changed.

Keep the sales columns in their own lanes

An Ads report credits sales according to its attribution rules. Seller ordered-product sales describe a different population, with their own time basis. Ad-attributed sales are not total seller sales; do not add the two columns or call the attributed figure the entire business. Seller Sales and Traffic can supply a separate view of business outcomes when authorized and aligned, but it cannot fill in missing campaign clicks or spend.

Extend the synthetic example with Ads-attributed sales of $4,000 for Week A and $4,500 for Week B. ACOS is $1,200/$4,000 = 30% in A and $1,800/$4,500 = 40% in B. ROAS is $4,000/$1,200 = 3.33 and $4,500/$1,800 = 2.50 after rounding. The attributed-sales increase is $500; no seller-wide revenue has been supplied. This is an advertising-efficiency reading, not a claim that total sales rose or fell. Confirm each report’s attribution configuration and allow recent conversions time to accrue before treating a period as final.

Give the biggest mover a question, not a verdict

Rank campaigns by their contribution to spend change. A campaign with the highest CPC might account for almost none of the $600. For the largest movers, place click count, CPC, spend share and attributed sales on the same line. Check change history for bids and budgets, and check offer availability from an appropriate inventory source. A launch, budget increase or stock interruption is a lead to verify, not an explanation the report proves by itself.

For zero clicks, CPC is undefined rather than $0. For positive spend and zero attributed sales, ROAS is zero while ACOS is undefined. Do not paper over either denominator. If attribution windows or ad products differ, split the comparison. Preserve campaign rows, period definitions and the unreconciled residue, if any, so another analyst can reproduce the bridge without accepting the narrative on trust.

If there is no Ads export

Request the account owner’s authorized Ads report and spell out the required account, marketplace, date range and fields. A seller sales export alone cannot reconstruct CPC, clicks or spend. The honest deliverable in its absence is a missing-input note, not a modeled Ads result. Keep customer identifiers out of a shared analysis under company policy. If the Ads source is unavailable, stop at the missing-input note.