Purchases
What it means
Purchases is the number of orders Amazon credits to an ad click inside its attribution window. It's a count, not money — a €15 order and a €150 order each add one. Organic orders, repeat buyers and brand-search orders are not in it.
Amazon and Meta both label a column Purchases, and the two are not the same measurement. Each platform counts only the orders it attributes to itself, over its own window, on its own storefront. Read each against its own history and never add them into a single total.
Show the math
Formula and a worked example
Orders credited to ads counts each order Amazon links back to a click on an Amazon ad, within Amazon's attribution window — whose length depends on the campaign type. The order has to be placed on Amazon, so nothing from your own shop is in this figure.
Worked example. In March your ads earn 3,000 clicks and are credited with 60 orders — an order rate of 2%. You spent €1,800, so each order cost €30. Those orders carried €4,200 of credited revenue, which makes the average ad-driven basket €70.
The count fills in late. Amazon credits the order back to the day of the click, so the last week or two of any period keeps rising after it closes — and always reads worse than it will finish.
It answers the question
How many orders did the advertising bring in? It's the denominator under every cost-per-order figure on the page.
Why it matters
Clicks tells you who reached the shop. Purchases tells you who reached the till, and only one of those pays the rent.
It's also the number that decides where budget goes next. A SKU with clicks and no purchases isn't a bidding problem waiting for a higher bid — it's a price, review or stock problem that more spend only makes more expensive.
What good looks like
There's no benchmark for a count like this: 50 orders a month is strong for a two-SKU catalogue and alarming for a hundred. Judge it against your own trend and against the ClicksClicksRaw traffic from ads; only valuable if clicks convert, so read beside CTR. that produced it — an order rate that holds while volume grows means you're scaling something that works. Falling orders on rising clicks is the shape to worry about.
Nor can the band be borrowed from the Meta tab. Amazon requires a click before it credits anything, so this count is built on a narrower event than Meta's and will nearly always be the smaller of the two.
How to improve it
| Lever | What you do | Expect | How long | Watch out for |
|---|---|---|---|---|
| Fast Move budget to what already converts | Shift spend from campaigns with clicks and no orders to those with a proven order rate | More orders on the same total spend | 1–2 weeks | Your winners get crowded and your newer SKUs starve — they never build the velocity that earns organic rank, so the catalogue narrows over a quarter. |
| Fast Fix price and delivery promise | Match the going rate and the delivery speed on SKUs that pull clicks but no orders | Order rate lifts within days on the same traffic | 3–7 days | Margin per order falls immediately, and a price you cut to win the comparison is hard to raise back without losing the rank it bought. |
| Slow Build reviews on your ad-driven SKUs | Follow-up requests and the review programmes Amazon offers on the products carrying your spend | Order rate rises, so the same Clicks produce more orders | 1–2 quarters | Nothing moves for weeks and the effort never appears in Ad Spend, so the gain later looks free when it wasn't. |
| Slow Advertise further down the catalogue | Open campaigns on SKUs and terms you've never bid on | Order volume grows from products that were invisible | 1 quarter | New campaigns spend for weeks before they produce enough orders to judge, so the cost of an order rises across the account while they learn. |
Every lever costs something somewhere. The last column is the one to read twice.
Read it with
Purchases counts the orders. The traffic beside it says what they cost to find.
Growth you bought
More orders because you bought more traffic. The order rate didn't improve, so this is a spending decision rather than a performance one.
Converting better
Fewer visitors, more orders. Tighter targeting or a better listing is doing the work — the cheapest growth on the chart and the easiest to miss.
Paying for the wrong visitors
Traffic is up and orders are down, so spend is climbing for nothing. Check the terms first, then price, stock and reviews on the listing they land on.
Drying up
The whole funnel shrank. A budget cap, a lost Buy Box or a bid that stopped clearing the auction explains this far more often than demand does.
One is volume, the other is price, and multiplying them gives you the spend. Orders up 20% with CPA flat is genuine scale; orders up 20% with CPA up 40% means you paid more for the growth than it brought in — and the order count alone would have called that a good month.
Revenue divided by orders is the basket size of an ad-driven customer. Orders holding steady while credited revenue falls means the campaigns are fine and the mix has shifted onto cheaper SKUs — a margin problem neither number shows alone.
Common misreads
It's your ad-credited order count. Every organic sale, repeat buyer and brand-search order is missing, which is why Purchases can fall in a month your business grew.
Recent days are always incomplete while the attribution window fills in. A week that looks down on Monday often finishes flat, so compare settled periods rather than fresh ones.
The count treats a €15 impulse buy and a €150 bundle as one apiece, and so does CPA. Check the credited revenue behind two identical order counts before you move budget between them.
Also called
Ad-attributed orders · attributed orders · ad-credited orders
See yoursYour Amazon ad-credited orders for the last 30 days, next to CPA and Ads Attr. Sales.
Open Ads Analytics →Purchases
What it means
Purchases is the number of purchases Meta credits to your ads. It's a count and not money — the euros sit in Purchase ValuePurchase ValueTotal purchase value attributed to this format. Pair it with spend to judge which formats return the most revenue beside it. Meta reports several kinds of purchase action, because accounts are tracked differently, and Store Analytics uses the broadest one your account actually returns a figure for.
That last point matters more than it sounds: your count and another store's count may not be built from the same action, so this figure is for comparing your own periods, not for comparing accounts.
Show the math
Formula and a worked example
Orders credited to ads counts each purchase Meta attributes back to someone who engaged with the ad, under the attribution setting on the campaign. Amazon needs a click before it credits anything; Meta's denominator is broader than that, which is the whole reason the two counts differ.
Worked example. In March your video ads spend €4,000 and Meta credits them with 200 purchases worth €14,000. Cost per order is 4,000 ÷ 200 = €20. Average order is 14,000 ÷ 200 = €70. Return is 14,000 ÷ 4,000 = 3.5.
On the Ad Formats table the count covers the whole period you selected. The chart underneath plots delivery only — spend, impressions, clicks — so there is no day-by-day line for this number to read.
It answers the question
How many orders is Meta claiming for the money you spent there? Not how many orders you took, and not how many of them Meta actually caused.
Why it matters
It's the denominator under CPACPAWhat one order costs you to win through ads. and the divisor that turns Purchase Value into an average order, so a wobble here quietly moves two other figures on the same row.
It's also the only column on the Ad Formats table that counts orders rather than money — Purchase Value, ROAS and AOV all price the same purchases a different way. Video, Photo, Carousel and Dynamic buy exposure at different rates, and this is the number that says which of them bought anything worth having.
What good looks like
No band applies, and one borrowed from Amazon would mislead. A count scales with budget and catalogue, so read it against your own trend and against SpendSpendWatch it to control budget and back the winners. — orders growing faster than spend is the shape worth having.
The two tabs are counted on different footings. Amazon credits an order placed on Amazon after a click on an Amazon ad; Meta credits a purchase it attributes to an ad it served, under a setting you choose, for an order placed on your own shop. Meta's is the looser test, so its count is usually the larger one — and that difference is definitional, not performance.
How to improve it
| Lever | What you do | Expect | How long | Watch out for |
|---|---|---|---|---|
| Fast Shift budget onto the format that converts | Read Purchases by format on the Ad Formats table and move spend to the winner | More orders on the same total spend | 1–2 weeks | You concentrate the account on one format, so when it tires you have nothing warmed up behind it and the recovery takes a full creative cycle. |
| Fast Fix the page the ads point at | Work the worst-converting destinations on the Top Landing Pages table rather than the ads pointing at them | Order rate lifts on traffic you're already paying for | 1–2 weeks | A page tuned for cold paid traffic often reads worse to the email and organic visitors landing on the same URL, and that loss never appears in the ad numbers. |
| Slow Improve the purchase signal you send Meta | Make sure orders are reported back reliably, so credited purchases stop being undercounted | Reported Purchases rises toward what actually happened | 3–6 weeks | This changes measurement, not demand. The count can climb while the business stands still, and your history either side of the change is no longer comparable. |
| Slow Build new creative around what already sells | New concepts on the products with a proven order rate, not more variants of a tiring ad | Order volume holds as the current winner fades | 1 quarter | New creative competes with proven creative for the same budget, so the blended cost per order usually worsens for weeks before the replacement is found. |
Every lever costs something somewhere. The last column is the one to read twice.
Read it with
Purchases counts the orders Meta claims. The figures beside it say what they were worth and what they cost.
Divide the value by the count and you have the average ad-driven order. Purchases holding while Purchase Value falls means the campaigns are still working and the mix has drifted onto cheaper products — a margin change that neither column shows on its own.
Spend divided by Purchases is the cost of an order. Orders up 30% on spend up 30% is a budget decision, not an improvement; orders up 30% on flat spend is the campaign genuinely getting better, and the count alone reads identically in both cases.
When orders stall while spend holds, Frequency usually explains it. The same people are being served the same ad more often, and the buyers among them have already bought — a creative and audience problem that the order count only reports after the fact.
Common misreads
Neither. Meta counts the orders it attributes to its own ads under its own attribution setting; Shopify counts every order however it arrived. They are answering different questions and will never reconcile. Use each against its own history.
The two are credited by different platforms, on different storefronts, under different rules and different windows. Adding them produces a figure with no single definition behind it — which makes it impossible to say whether it went up for a good reason.
Attribution credits the ad for orders that a share of those customers would have placed anyway. The count is what Meta claims, not what Meta caused — which is why a change in tracking can move it without a single extra sale.
Also called
Attributed purchases · purchase actions · attributed conversions
See yoursYour Meta purchases by creative format, next to the Purchase Value and Spend behind each one.
Open Ad Formats →