Your Amazon auto campaign is four campaigns: close match, loose match, substitutes, complements
An automatic Sponsored Products campaign looks like one thing: you pick a product, a budget and a bid, and Amazon finds the shoppers. Inside, it runs four separate targeting groups. Each one has its own bid and its own on/off switch, and each one shows its own numbers in the targeting report.
- Close match: shoppers searching for terms closely related to your product.
- Loose match: shoppers searching for terms loosely related to your product.
- Substitutes: shoppers viewing products similar to yours.
- Complements: shoppers viewing products that go with yours.
If you created the campaign with one default bid, all four run on that bid, and you judge the campaign by its total.
The measured numbers below come from real accounts: 11 advertiser accounts in Europe and the US, 130 automatic ad groups, Sponsored Products targeting report from March 28 to September 29, 2026. We left out the last 7 days because ad-attributed sales are not final yet. We compared ratios only inside one account and currency, never summed euros with dollars.
The money does not split evenly
In the euro accounts, substitutes took about half of all automatic spend (49.7%). In the pound accounts it took 58.1%. In the dollar accounts loose match led with 62.1%. Complements got the least everywhere: 7.6%, 3.5% and 1.3%.
Most ad groups also did not sell through every group. Of the 82 ad groups where all four groups got impressions, only 5 had sales from all four.
The biggest spender is usually not the best group
To compare fairly, we took the ad groups where at least two groups had 3 or more purchases each. A group with one or two sales can look excellent or terrible by chance. That left 13 ad groups in 5 accounts.
In 9 of the 13, the group that took the most money was not the one with the lowest ACOS.
Two examples from the data:
- In one ad group substitutes spent 3.4 times as much as loose match, at 39% ACOS. Loose match sold at 24% ACOS.
- In another, substitutes spent 173 euros at 42% ACOS. Close match spent 21 euros and sold at 4% ACOS.
Among the groups with at least 3 purchases, the highest ACOS was 1.55 times the lowest in the median ad group. Across the 13 ad groups this ratio ran from 1.06 to 10.67.
There is no group that always wins
Across the 13 ad groups, the lowest ACOS came from close match 5 times, loose match 4 times and substitutes 4 times. Complements never had the lowest ACOS in this set.
The winner also changed inside the same account. Three accounts had at least two of these ad groups, and in two of them the winning group differed from one ad group to the next. So a rule like "always cut substitutes" or "always push close match" would be wrong for some of your ad groups. Decide per ad group.
What we could not see
Our data has the spend of each group, but not the bid set on it. So we cannot tell how much of the uneven split comes from bids and how much from how many shoppers each group reaches. We also did not test what happens to the other groups when you lower one bid.
Twenty minutes per auto campaign
- In Campaign Manager open the automatic campaign, then the ad group, then the Targeting tab. You will see the four groups as four rows.
- Set the date range to the last 60 days, ending 7 days ago.
- Judge by ACOS only the rows with at least 3 purchases. Judge the rows without sales by money. A worked example: if your target ACOS is 30% and the product sells for 20 euros, a group that spent 6 euros without a sale already lands at or above your target with its first sale.
- Raise the bid a little on the row that sells well under your target. Lower it on the row that sells above it. Pause a row only when its spend without sales is past that limit.
- Do it for each ad group separately, and look again a month later with the new window.
The total of an auto campaign can hide one group that pays for the others. The four rows show you which one it is.