Amazon PPC Attribution Window: Why Yesterday's ACoS Is Wrong

Elena sells a $30 stainless garlic press. Every Monday morning she opens the advertising console, sorts her keywords by ACoS and cuts bids on anything above 40%.

One Monday her best exact-match keyword showed Sunday at 83% ACoS: $100 spent, $120 in sales. She cut the bid by 30%.

The following Monday she looked at that same Sunday again. Same $100 in spend. Now it showed $330 in sales and a 30% ACoS, comfortably under her 40% target. Nothing about Sunday had changed. Amazon had simply finished counting it.

She had cut the bid on her most profitable keyword because she read the data four days too early. And she'd been doing it every week.

That's the Amazon PPC attribution window at work. Amazon keeps crediting sales to an ad click for 7 to 14 days after it happens, so recent numbers always look worse than they will end up. This guide covers how the window works for each ad type, how long to wait before you trust a number, why the console never matches your Business Reports, and how to build a routine that doesn't punish your best keywords for being recent.

What Is the Amazon PPC Attribution Window?

Three details matter more than the window length itself.

Sales are credited to the day of the click, not the day of the order. If a shopper clicks your ad on Sunday and buys on Wednesday, that sale appears in Sunday's numbers. This is why past days keep changing. Amazon isn't revising history; it's back-filling it.

Amazon credits the last ad click. If the shopper clicked a Sponsored Products ad on Sunday and a Sponsored Brands ad on Tuesday before buying, the sale goes to the Sponsored Brands click.

The window is fixed. Sellers can't change it in the console. You can only change how you read the data.

Attribution Windows by Amazon Ad Type

Here's what each ad type counts and for how long.

Two practical consequences follow.

First, you can't fairly compare Sponsored Products ACoS with Sponsored Brands ACoS. Sponsored Brands gets twice as long to collect sales and counts the whole brand. A Sponsored Brands campaign at 28% ACoS is not automatically better than a Sponsored Products campaign at 32%.

Second, the longer the window, the longer you wait. A Sponsored Display retargeting campaign needs about two weeks before its numbers settle.

Why Your ACoS Drops Days Later: The Back-Fill in Numbers

Go back to Elena's Sunday. Here's the same day viewed at different points in time. The pattern is typical of a mid-priced product on Sponsored Products: 100 clicks at $1.00, $30 price.

Spend never changed. Only the denominator grew. On Monday morning Sunday looked like a keyword at its limit. A week later it was one of her best days.

Most sales land in the first 24 to 72 hours, so the curve flattens fast. But "most" isn't "all," and the last 10% is often the difference between "above target" and "under target." On top of the attribution lag, spend and clicks can shrink slightly for a few days as Amazon filters invalid clicks, so even the cost side isn't final on day one.

How Long to Wait Before Trusting Amazon PPC Data

The rule we use in our own accounts:

• Sponsored Products: treat the last 3 days as incomplete. Judge performance on periods that ended at least 3 days ago. For final numbers, wait 7 days.
• Sponsored Brands and Sponsored Display: treat the last 7 days as incomplete. Final numbers take 14 days.
• High-priced or considered purchases: add more time. Shoppers compare, save to lists and come back.

That last point catches a lot of sellers. Tom sells a $189 standing-desk converter. On a garlic press, most of the attributed sales show up within a day. On Tom's product, shoppers click on Friday, read reviews over the weekend and order on Tuesday. When he checked his Black Friday to Cyber Monday spend the morning after, his ACoS read 58% against a 35% break-even. He nearly switched off his top campaigns for December.

Ten days later the same four days had settled at 31%. Nothing had been wrong. He had just measured a long-consideration product with a short-consideration habit. If you sell anything over $100, our guide to PPC for high-priced products covers the bidding side of this in detail.

Why the Ad Console Never Matches Business Reports

Sellers often notice that ad sales in the console don't line up with orders in Business Reports. Sometimes ad sales look higher than they could possibly be. The attribution window explains most of it.

So a week with heavy clicks late in the period (say, Black Friday on a Friday) will show ad sales in the console that Business Reports places in the following week. Over a full month the two line up much better, but they never match exactly.

The fix isn't to choose one. Use the console to manage campaigns, and use total sales from Business Reports to judge whether ads are helping the business. That's what TACoS does: total ad spend divided by total sales. If ACoS and TACoS are new territory, read our ACoS vs TACoS breakdown.

Advertised SKU Sales vs Other SKU Sales (Halo Sales)

Inside the attribution window, Amazon also counts purchases of products other than the one you advertised. In the Advertised product report you'll see the split as "7 Day Advertised SKU Sales" and "7 Day Other SKU Sales." The Purchased product report shows exactly which other ASINs shoppers bought.

This is useful and dangerous at the same time.

Useful, because halo sales are real money. If your garlic press ad leads a shopper to buy your herb scissors instead, that sale wouldn't have happened without the click.

Dangerous, because it can make an ad look profitable when the product you're paying to promote isn't selling. A campaign at 30% ACoS can be a 55% ACoS campaign for the advertised product with the rest coming from other products.

Check the split for your top-spending campaigns once a month. If other-SKU sales are more than about a third of the total, ask why shoppers are clicking one product and buying another. Often it's price, reviews or a better variation, and the right move is to advertise the product they actually buy.

How the Attribution Window Breaks Your Bid Decisions

Reading fresh data is a mistake you make once a week. Automating on fresh data makes the same mistake every day.

Hannah runs a home-fitness brand. She set up a rule: if a keyword's ACoS over the last 3 days is above 40%, lower its bid by 15%. It sounded sensible. But the last 3 days are always the least complete, so her rule saw inflated ACoS on nearly every keyword, every day. Bids drifted down. Within three weeks her impressions had fallen by 45%, and her ACoS barely moved, because the "high ACoS" had never been real.

The common versions of this mistake:

1. Short lookback windows on bid rules. Any rule or script that judges ACoS on the last 1 to 3 days is judging incomplete data. Use at least 14 days of data that ends 3 or more days ago.
2. Monday reviews of the weekend. Saturday and Sunday are always the least complete days on a Monday. Review the week that ended on Thursday instead.
3. Comparing periods of different ages. "This week vs last week" compares an incomplete week with a complete one. The current week always loses. Compare two periods that are both at least 3 days old.
4. Judging event spend the morning after. Prime Day, Black Friday and Cyber Monday bring heavy clicks in a short burst. The orders keep coming in for days. Our Q4 PPC strategy guide covers how to plan the budgets; the attribution window decides when you can grade them.
5. Hourly decisions on attributed sales. Dayparting by the hour on sales data is even more exposed to lag than daily decisions. If you schedule bids by hour, base it on several weeks of matured data, not yesterday.
6. Pausing Sponsored Brands too early. With a 14-day window, a Sponsored Brands campaign can look poor for its first week and then settle at a healthy ACoS. Give it the full window before deciding.

A related point: when sales suddenly look like they've dropped, check the date range first. The single most common "sales crash" we see is just the last 72 hours being incomplete. Our guide on why Amazon PPC sales dropped walks through the full diagnosis.

A Lag-Aware Weekly PPC Routine

Here's the routine we'd give any seller who wants to make decisions on numbers that won't change under them.

1. Pick a fixed review day and a fixed data window. If you review on Monday, analyse the 7 days ending the previous Thursday for Sponsored Products. For Sponsored Brands and Display, use the 14 days ending a week ago.
2. Decide on clicks, not days. A keyword hasn't proven anything until it has enough clicks to expect orders. A simple threshold is 3 divided by your conversion rate: at a 10% conversion rate, wait for 30 clicks before calling a keyword a loser.
3. Compare like with like. When you compare periods, both must be equally mature. Week ending 3 days ago against the week before it, not "this week so far" against last week.
4. Log every change with its date. Bid changes affect clicks immediately but sales over the next 7 to 14 days. Without a change log you'll credit or blame the wrong change.
5. Re-check last week's decisions. Each review, look at the period you judged last week. If it has improved since, your data window is too short. If it hasn't changed, you can tighten it.
6. Judge the business on TACoS monthly. Monthly totals smooth out the click-date vs order-date mismatch. That's where you see whether ads are growing total sales.
7. Keep your own archive. Some Amazon reports only go back about 60 days. Download matured monthly data so you can compare this Black Friday with last year's. Our Amazon advertising reports guide lists which reports to save.

If you need a custom view, such as a 30-day path to purchase or how many shoppers saw a Sponsored Display ad before clicking a Sponsored Products ad, Amazon Marketing Cloud lets you run your own attribution analysis on event-level data. For most sellers it's overkill. The routine above handles 95% of the problem.

Black Friday and Cyber Monday: When to Grade Your Spend

This matters most in the next few weeks. Event days concentrate clicks, and many shoppers click during the event and buy a few days later, especially on gifts and higher-priced products.

A simple schedule for the 2026 Black Friday to Cyber Monday period (November 27 to 30):

• Tuesday, December 1: check that budgets didn't run out and nothing was suppressed. Don't judge ACoS yet.
• Friday, December 4: first read on Sponsored Products. Expect it to improve a little more.
• Monday, December 7: final Sponsored Products numbers for the event.
• Monday, December 14: final Sponsored Brands and Sponsored Display numbers.

Any bid or budget cut you make before December 4 is based on a number that's still moving. Read the Sponsored Products documentation and Amazon's ACoS guide if you want Amazon's own definitions behind these metrics.

Frequently Asked Questions

What is the attribution window for Amazon Sponsored Products?

For sellers, Sponsored Products attributes sales for 7 days after an ad click. For vendors it's 14 days. Sponsored Brands and Sponsored Display use 14 days for everyone.

Why does my ACoS change after a few days?

Amazon credits sales to the date of the ad click and keeps adding sales for the full attribution window. As late orders arrive, sales for past days increase while spend stays the same, so ACoS falls.

How long does Amazon PPC data take to update?

Clicks and spend appear within hours, though they can be trimmed slightly for a few days as invalid clicks are removed. Most attributed sales show up within 1 to 3 days. Final numbers take 7 days for Sponsored Products and 14 days for Sponsored Brands and Sponsored Display.

Can I change the Amazon attribution window?

Not in the advertising console. Sellers get a fixed window per ad type. Amazon Marketing Cloud allows custom attribution analysis, but it doesn't change what the console reports.

Does Amazon use last-click attribution?

Yes. When a shopper clicks more than one of your ads before buying, Amazon credits the sale to the last ad clicked.

Conclusion

The Amazon PPC attribution window doesn't change your results. It changes when you can see them. Sponsored Products keeps counting for 7 days, Sponsored Brands and Display for 14, and every sale is filed under the day of the click. That means the most recent data is always the worst-looking data.

So set your review window to end 3 days back for Sponsored Products and a week back for Brands and Display. Make decisions on clicks, not calendar days. Compare periods of the same age. And judge the business on TACoS once the month has settled.

Elena now reviews the week that ended on Thursday. She hasn't cut a profitable keyword by mistake since. The bids she used to slash on Mondays are the ones carrying her Q4.