Amazon advertising reporting: build the decision before the dashboard

By Naeela

Amazon advertising reporting should answer which decision changes next, not simply display another ROAS chart. Amazon made unified reporting generally available on June 8, 2026 and plans to sunset the older Sponsored Ads and DSP report locations by December 31, 2026. That migration is a useful forcing function for brands generating $1M+ a year on Amazon: rebuild the report around campaign, product and contribution decisions while the old exports are still familiar. A cleaner interface will not reconcile ad-attributed sales with total SKU sales or decide whether the next dollar of spend earns its place.

What decision should Amazon advertising reporting answer first?

Start with the choice on the table. "How did ads do?" is too vague to decide anything. "Should we raise bids on this search term while the advertised child SKU has 18 days of sellable cover?" names the campaign, constraint and consequence. The report must let the owner answer that question, even if the answer is to leave the bid alone and fix replenishment.

There are several legitimate reporting jobs, and they require different grains. A weekly budget decision needs campaign and contribution context. A keyword-targeted search-term decision needs the actual customer query, not merely the keyword you bought. A product decision needs the advertised ASIN, the purchased ASIN and total business sales for the same product and period. A creative decision needs a comparison with enough exposure to separate a useful signal from noise. Putting all of those into one executive scorecard generally makes the scorecard easy to admire and hard to use.

Amazon's campaign-reporting overview describes industry-standard metrics such as clicks and ROAS alongside Amazon-specific metrics including new-to-brand and Subscribe & Save. That is an inventory of measurement options, not a verdict on business value. ROAS can improve because a brand stopped buying growth. New-to-brand orders can rise while acquisition contribution falls. A campaign can be efficient and still accelerate a stockout. Each metric earns its place only if it changes an owned decision.

I would refuse a report request that stops at "make the numbers green." That costs an agency the comfort of an easy win: attractive slides are easier to deliver than a recommendation to slow spend. The refusal protects the client from mistaking a presentation for operating control. If you only need a clean export for a finance archive, a basic scheduled report is enough. Do not pay for a strategy layer you do not intend to use.

Which report belongs to which decision?

Choose the narrowest report that can settle the question. Amazon's sponsored ads guide identifies search-term, targeting, advertised-product, placement and purchased-product reports among its options. The categories are not interchangeable. A targeting report aggregates the target you configured; a search-term report can expose what a shopper actually entered for keyword-targeted traffic. The purchased-product view can reveal a different item bought after the click.

These are starting points, not automated rules. PPC Ninja's report guide covers much the same report menu and recommends a download cadence by type. Its blanket click-count and data-retention advice should not become a rule for every account. The useful part is the distinction between query, target, placement, advertised item and purchased item. Your account's current reporting access and history should be checked directly.

The executive summary can be one page. Behind it should sit an evidence trail: original export, time zone, marketplace, campaign type, attribution definition, pull date, calculation version and decision note. Without those, a changed number on Thursday becomes a meeting about whose spreadsheet is right. Nobody needs another 40-minute argument about whether Tuesday's sales were late-arriving attribution or a formula change.

What changed with unified reporting in 2026?

The important change is consolidation, not permission to blend unlike economics. Amazon's June 8 announcement says unified reporting is generally available in Ads Console. It allows one report across manager or advertiser accounts, ad products, countries, metrics and dimensions. Amazon also says daily or weekly history reaches 15 months and monthly, yearly or summary history reaches 6 years in this interface. Those limits describe the announced interface, not every legacy export or every account's data rights.

The same announcement says the older Sponsored Ads reports and Amazon DSP reports locations are due to sunset by December 31, 2026. It refers to a future automated subscription migration tool. Do not treat that mention as evidence that your scheduled reports have already moved. Build a small migration register now: report name, consumer, schedule, columns, filters, date grain, owner and whether a replacement has been tested. Run both versions for an overlap period and investigate differences before retiring the old one.

There is a real operating opportunity here. Standardized dimensions can shorten the time spent stitching files. An owner can inspect a cross-product or cross-country pattern without requesting five separate exports. But a standardized name is only a field definition. It does not prove a Sponsored Products order and a DSP-influenced conversion are equivalent, or that a combined ad-sales line is incremental revenue. Keep the product-specific source view available when a decision is disputed.

For migration planning, use the June general-availability announcement and your account's observed output, rather than an old beta description. Amazon says the new interface standardizes metrics, dimensions and attribution methodology across ad products. That is a reporting improvement. It does not make every credited order incremental or remove the need to check the product-specific definitions behind a budget decision.

How do you reconcile ad results with real product economics?

Build a bridge, not a magic formula. First, keep ad delivery and attribution in their own table: spend, impressions, clicks, attributed orders and sales, campaign, advertised ASIN and date. Second, bring in Seller Central sales at the same marketplace and item grain. Third, add the costs the ad console does not own: landed product cost, Amazon fees, promotions, returns allowance and any other variable cost that changes with the order. The result is not a single perfect attribution truth. It is a more honest decision range.

The Amazon Sales and Traffic report specification permits PARENT, CHILD or SKU ASIN granularity and DAY, WEEK or MONTH date granularity. That makes the grain choice explicit. If one file is parent-level and another is child-level, the arithmetic can be correct while the answer is wrong. A child with weak margin can disappear inside a strong parent average. Conversely, a traffic-driving variant may look terrible in isolation while a more profitable sibling receives some downstream orders. Inspect both views before making the spend decision.

Consider a hypothetical 30-day period. A product family has $80,000 in total ordered sales and $12,000 in ad spend, a 15% family TACoS. One advertised child accounts for $4,000 of spend but only $10,000 of its own ordered sales, a 40% direct ratio. The first number says nothing about whether that child should be cut. The child may feed profitable purchases elsewhere in the family; it may also be a costly doorway whose halo is mostly thin-margin items. The purchased-product report and a contribution sheet settle which story is plausible.

Keep the language careful. Ad-attributed revenue is not necessarily incremental revenue. A purchase of another SKU after a click is not automatically profit attributable to the advertised SKU. Total Seller Central sales are not all organic sales. The useful question is whether a proposed action changes contribution after allowing for these uncertainties. If the conclusion reverses when you vary the halo assumption, do not automate the bid. Run a bounded test or seek better evidence.

Our product-level TACoS guide explains the SKU lens in more detail. The Amazon PPC contribution-margin guide covers the economic source. Reporting should join those jobs without replacing either one.

How should an operator review the report each week?

Hold a short decision meeting with a stable sequence. First, check data health: missing days, late conversions, changed field definitions, currency, marketplace and SKU mapping. Second, inspect the biggest movement in contribution, not just the biggest movement in attributed sales. Third, compare that movement with inventory, price, listing status and promotion changes. Fourth, record the recommendation and the person who can approve or stop it.

The meeting should end with a decision log, not a deck. A useful row names the date, affected SKU or campaign, observed signal, competing explanation, action, owner, economic boundary and next review. A bid increase might be capped until sellable cover returns above an agreed threshold. A negative keyword might wait for a listing fix if the query is relevant but the offer has deteriorated. A budget move might be reversed if contribution fails to improve after the attribution window matures.

One seller-origin forum post from April 2026 describes normal Sponsored Products traffic and spend alongside zero attributed sales for a day while total sales remained strong. It is one account's report, not evidence of a widespread outage. The useful lesson is procedural: before rewriting bids from a surprising daily export, preserve the source, check timing and attribution, and distinguish a reporting anomaly from a real offer or demand change.

If you cannot explain why a campaign changed in 2 weeks, the reporting system is missing history, even if every chart rendered. Preserve the raw export and the calculation version. Document exceptions: suppressed listing, lost Buy Box, receiving delay, price promotion or a changed creative asset. Those events change the meaning of the same ad metric. An account manager who did not know about the inventory delay may call the bid change a failure when it was the least damaging option available.

This is the work a senior operating team should own. At ALFI, founder and Chief Strategist Naeela leads the strategy, review and accountability conversation with seasoned senior operators doing the work. The 18-client ceiling protects attention for the connected decision, not a promise of prettier charts. We will not accept a PPC-only mandate and then pretend to own profit while pricing, listings or inventory sit outside the operating scope. If a brand wants only report production, a specialist analyst or internal hire is often the better fit.

When is a new reporting tool worth buying?

Buy a tool when it reliably removes a specific failure. If the team spends 6 hours each week downloading and joining exports, automation may return real capacity. If the same team still cannot name who decides whether an ASIN deserves another dollar, a new dashboard will preserve the ambiguity faster. Test a tool against the decision log before signing.

Give the candidate system 3 representative cases: a healthy hero SKU, a weak child hidden by family averages and a high-ROAS item near stockout. Require it to reproduce raw Amazon totals, expose attribution and grain choices, connect current cost data, retain a reversible audit trail and show where manual judgment remains. A pilot that only demonstrates chart speed has not passed the operating test.

The counterargument is fair: not every business needs a bespoke warehouse or weekly committee. A small catalog with stable margins can run a disciplined spreadsheet and simple scheduled reports. A large team may need a data platform because manual joins are no longer reliable. The choice is not simple versus sophisticated. It is whether the system makes a profitable decision easier to reach and easier to challenge.

The Amazon Marketing Cloud guide is useful if a more advanced measurement tool is the immediate job. This article's job is the reporting decision system that any chosen tool must serve. We would rather turn away a reporting-only engagement than sell automation before the owner and economic rule are clear.

What to do this week

List every scheduled Amazon Ads export and its human consumer. Pick the 3 decisions that have actually moved money or inventory in the last month. Rebuild only the reports needed to make those decisions in unified reporting, then compare them with the old source. Add SKU economics, a named owner and a review date. If the team cannot agree what it would do differently after seeing the report, stop adding charts and settle the operating rule. Brands ready for that connected accountability can explore ALFI's connected Amazon management; teams seeking only report assembly should keep the brief narrower.