Top-of-search impression share: why your number is low, and why that is normal
Amazon shows a column called top-of-search impression share next to your Sponsored Products campaigns. Most sellers who open it see a number like 0.3% or 2%, and the first reaction is that something is badly wrong. The usual next step is to raise bids or the top-of-search adjustment until the number goes up.
We measured what the number looks like on real accounts, what it does and does not tell you, and what happened on the days when it was higher. The short answer: a low number is the normal case, and a higher one mostly buys you more expensive clicks, not more buyers.
What the metric measures
Amazon defines it as the percentage of top-of-search impressions your campaign received out of the total top-of-search impressions it was eligible to serve on. Top of search means the ad slots at the top of the first page of search results.
Two things follow from this definition:
- The denominator is every top-of-search slot your campaign could have entered, not the ones you were close to winning. An auto or broad campaign can be eligible for a huge number of searches, so a small share is expected.
- It is a share of the top-of-search slots, not a share of your own traffic. A campaign can have a tiny impression share and still get a large part of its clicks from top of search.
What the number looked like on 145 campaigns
The data below comes from the daily Sponsored Products campaign reports of 9 advertisers, from August 19 to October 17, 2026, on Amazon Germany, France, Spain, Italy, the UK and the US. We counted a campaign-day only when the campaign had at least 50 impressions, and a campaign only when it had at least 10 such days: 145 campaigns, 5,236 campaign-days, 2.59 million impressions and 16,353 clicks. For each campaign we took the median of its daily top-of-search impression share.
- The typical campaign (the median of the 145) had a top-of-search impression share of 0.43%.
- 86 of the 145 campaigns were under 1%, and 126 were under 10%.
- Only 14 campaigns were at 25% or above, and 4 at 50% or above.
The targeting decides most of it. We grouped campaigns by where their clicks came from in the search term report:
- Campaigns with at least 90% of their clicks from exact keywords: median 2.74% (33 campaigns).
- Campaigns with at least 90% of their clicks from auto targeting: median 0.20% (39 campaigns).
- Mixed campaigns: median 1.07% (43 campaigns). The remaining 30 had fewer than 20 clicks in the search term report and were not grouped.
So if your auto campaign shows 0.2%, it is in the middle of the pack, not broken.
A small share of slots can still be a large share of your clicks
To see how much of a campaign's own traffic came from top of search, we used the placement report for the same campaigns and days.
- In the typical campaign, top of search was 2.0% of its impressions.
- The same top-of-search slots brought 13.3% of its clicks (both medians across the 145 campaigns).
Top-of-search impressions are rare but get clicked far more often than the rest. That is why the impression share alone says little about how much top of search matters to a campaign. The placement report answers that, as we described in top of search or product pages.
What happened on the days with a higher share
The share moves from day to day inside the same campaign. We took the 77 campaigns with at least 20 qualifying days, split each campaign's days into its lower-share half and its higher-share half, and kept a campaign when both halves had at least 20 clicks. Then we compared each campaign with itself.
- The median share went from 0.53% on the lower days to 2.75% on the higher days.
- Top of search's part of the campaign's clicks roughly doubled, from 8.5% to 16.9% (medians). It rose in 59 of the 77 campaigns.
- Cost per click was higher on the high-share days in 53 of the 77 campaigns, by a median 5%.
- Click-through rate was higher in 50 of the 77, by a median 8%.
- Conversion rate was higher in only 19 of the 48 campaigns that had at least 5 orders across both halves.
- ACOS was lower in 14 of the 34 campaigns that had at least 3 orders in each half.
So a higher share came with more top-of-search clicks and somewhat dearer clicks, but not with more buyers per click, and ACOS improved in fewer than half of the campaigns.
The arithmetic explains part of it. Suppose top-of-search clicks convert 50% better than the others. Moving from 8.5% to 16.9% of clicks then lifts the whole campaign's conversion rate by about 4%. At a 6% conversion rate, that is about one extra order per 400 clicks, too small to show up reliably in a campaign with a few hundred clicks.
What we could not see
These are the same campaigns on different days, not an experiment. A day can have a higher share because you raised a bid, because a competitor paused, or because the search mix changed, and the report does not say which. Amazon does not publish how many top-of-search slots a campaign was eligible for, so we cannot rebuild the denominator. Orders are counted for 7 days after the click, and an order can be for another product of yours than the one in the ad. The 50 impression, 10 day and 20 click limits are our choice, not a measurement. In 8 of the 77 campaigns, the click totals in the placement report and the campaign report did not match exactly.
How to use it in your account
- Do not compare your number to a target like 20% or 50%. Compare it to the same campaign's own history.
- Read it by targeting. Expect well under 1% on auto campaigns and around 1% on mixed ones. Exact campaigns are where a higher share is realistic.
- Before you push the share up, open the placement report for the same campaign, last 30 days, and write down clicks, spend and sales for top of search and for the other placements.
- If top of search has a clearly lower ACOS there, raise the top-of-search adjustment in small steps and watch the share and the ACOS together for two weeks.
- If top of search has a similar or higher ACOS, a low share is not a problem to fix. Leave it.
A low top-of-search impression share means there are many slots you did not take. Whether you should take them is a question for the placement report, not for this column.