Long-tail searches on Amazon: not cheaper clicks, better buyers
A common piece of Amazon advertising advice says long searches are cheaper per click, because fewer sellers bid on them. We checked it on real accounts. The cheaper click mostly did not show up. What did show up was a better buyer: in most accounts, long searches sold at a lower ACOS.
The measured numbers below come from the Sponsored Products search term report of 12 advertiser accounts in Europe, the UK and the US, day by day from March 28 to October 5, 2026. Not every account ran for the whole period. We stopped on October 5 so that the ad-attributed sales of the last days are close to final. That is 43,093 daily rows and 61,782 clicks. Every comparison below is made inside one account and one currency, because a click in pounds and a click in dollars cannot be averaged together. Below, "account" means one account in one currency.
We counted the words in what the buyer typed. Rows where the search is an ASIN (a product page, code starting with B0) have no words, so we left them out of the comparison. They were 38.5% of all clicks. We compared short searches (one or two words) with long ones (four or more). Three-word searches sit in between, and we kept them out to make the two groups clearly different.
Where long searches come from
Among the text searches, 24.8% of clicks came from searches of four or more words.
Few of them came through exact match. Of the clicks on four or more word searches, 8.9% came through an exact keyword, 37.1% through broad match and 48.8% through automatic campaigns. For one and two word searches, exact match brought 42.5% of the clicks. Long searches are mostly found by broad and automatic campaigns, not by keyword lists.
The click was not cheaper
10 accounts had at least 100 clicks in both groups. In each, we divided the average cost of a click on long searches by the same number on short searches.
- Long searches were cheaper per click in 5 accounts, and equal or more expensive in the other 5.
- The median ratio was 0.99. The lowest was 0.60, the highest 1.22.
- In 4 of the 10 accounts the difference was under 5%.
So within one account, a long search usually costs about the same per click as a short one. If you plan your budget on cheap long-tail clicks, the numbers do not support it.
The buyer was better
The conversion rate (orders divided by clicks) was higher on long searches in 7 of the 10 accounts. In one of the 7 the higher rate rested on 2 orders in each group, which is too few to count.
ACOS tells the same story. 8 accounts had at least 5 orders in both groups:
- In 6, long searches sold at a clearly lower ACOS: for example 17.5% against 26.3%, 37.0% against 73.2%, and 52.7% against 75.8%.
- In 1, the two were practically the same: 38.1% against 38.2%.
- In 1, long searches were worse: 53.4% against 37.2%. In that account short searches converted unusually well, 12.8% of clicks became orders.
This is not a test. Short and long searches in one account can lead to different products at different prices, and that alone moves ACOS. What the numbers show is that in most accounts the advantage of a long search is the buyer behind it, not the price of the click. And that it is not a rule: one account in eight went the other way.
What we could not see
The search term report only lists searches that got a click, so we do not know how often long searches showed the ad without a click. We did not split by product, so part of the ACOS difference may come from price. The 100-click and 5-order cut-offs are our choice, not a measurement. Counting words by spaces treats "12mm" and "12 mm" differently. And the report counts purchases for 7 days after the click.
Check it in your account in 20 minutes
- Download the Sponsored Products search term report for the last 60 days, ending at least a week ago.
- Delete the rows where the search is an ASIN (starts with B0).
- Add a word count column. In Excel or Google Sheets: =LEN(TRIM(A2))-LEN(SUBSTITUTE(TRIM(A2)," ",""))+1, where A2 is the search.
- Make a pivot table by word count with clicks, spend, orders and sales. Add cost per click, conversion rate and ACOS.
- If long searches convert better in your account, harvest the ones with at least two orders as exact keywords, and add them as negative exact in the campaign they came from. Set their bid from what a click is worth to you (price times conversion rate times target ACOS), not from the hope that they are cheap.
- If they do not convert better, do not build campaigns around long tail. Keep finding them with broad and automatic campaigns, and judge each search by its own numbers.
A long search usually does not save you money on the click. When it pays off, it pays off at the checkout.