Amazon Search Trends: How to Tell a Real Rise From a One-Week Spike

Every week Amazon re-ranks every search term shoppers type. Watching which terms climb is the closest thing a seller has to a live feed of demand. It is also easy to misread: a term can jump because of a sale, a news story or one viral video, and be back where it was a week later.

This post is about reading that movement. It uses the weekly search-term data from Amazon Brand Analytics (ABA), the same data behind our weekly rising search terms ranking. How to read a single ABA row — click share and entry cost — is covered in reading Amazon Brand Analytics search terms.

What "rising" means in this data

ABA publishes a rank, not a count. A term ranked 9,000 this week was searched more than every term ranked below it. So a rising term is a term whose rank number got smaller.

The growth rate compares the rank now with an earlier rank:

growth = (earlier rank − current rank) ÷ earlier rank

A term that went from rank 20,000 to 10,000 has a growth of 0.5, or +50%. Two consequences follow, and both change how you read a list of risers.

The rate tops out just under 100%. A rank cannot go below 1, so even a term that leaps from 500,000 to 5,000 scores 0.99. Very different jumps look the same at the top of a list. Look at the ranks themselves, not only the percentage.

Falls are not capped. A term that slides from 1,000 to 15,000 scores −14. Averaging growth rates across terms therefore says little; compare terms one at a time.

Read three windows, not one

Each weekly row carries the rank one, four and twelve weeks ago, and a growth rate against each. Put the three side by side and most terms fall into one of four shapes:

The table is a reading aid, not a rule. A term with a strong 12-week rise and a flat last week is often a better candidate than one with a dramatic single week, because the long window has already survived a few weeks of noise.

Separate brand terms from product terms

Many risers are brand names. When a beauty brand runs a sale, its name can climb tens of thousands of places in a week. That is demand for the brand, and only the brand can serve it.

Product terms — a type of item rather than a maker — are the ones any seller can answer. Two checks help:

• Who takes the clicks. ABA lists the top clicked products for each term. If the leading brands are small and varied, a new listing has room; if one brand takes most of the clicks, the term is already owned.
• Whether the term is even a product. Book titles, films and songs show up in search data too. Drop terms whose top clicks are media.

Size the term before you get excited

A rising rank says nothing about how big the term is. Climbing from rank 400,000 to 200,000 is a large percentage on a term few people search. The weekly data carries an estimated search volume next to the rank; use it to decide whether a term can carry a listing on its own or is one keyword among several.

This is also why our ranking samples terms across six volume bands between 10,000 and 100,000 weekly searches instead of taking the biggest risers overall: otherwise the list fills with a handful of large brand terms.

Here is the latest issue. The table follows each new issue, so read it with the shapes above rather than as a buy list:

Confirm a riser in two more places

Its own history. The ABA trends endpoint returns a term's rank as a series. Set timeGranularity to W for weekly points or M for monthly ones; the monthly series shows whether the term rose at the same time last year.

curl --request POST \
  --url https://ecommercedataapi.com/v1/amazon/aba/research/trends \
  --header "Content-Type: application/json" \
  --header "X-API-Key: ${ECOMMERCE_DATA_API_KEY}" \
  --data '{
  "marketplace": "US",
  "keyword": "heated vest",
  "timeGranularity": "M"
}'

Outside Amazon. A trend that also rises on Google is less likely to be an on-site promotion. The Google Trends endpoint returns the same kind of series for web search.

Build your own weekly watchlist

The ABA weekly endpoint filters on rank growth directly. This asks for terms between 10,000 and 30,000 estimated weekly searches whose rank improved at least 30% in the last week:

curl --request POST \
  --url https://ecommercedataapi.com/v1/amazon/aba/research/weekly \
  --header "Content-Type: application/json" \
  --header "X-API-Key: ${ECOMMERCE_DATA_API_KEY}" \
  --data '{
  "marketplace": "US",
  "minSearches": 10000,
  "maxSearches": 30000,
  "minRankGrowthRate": 30,
  "order": { "field": "searches", "desc": true },
  "page": 1,
  "size": 40
}'

Mind the units. minRankGrowthRate takes a percentage — 30 means 30% — while the growth rates in the response are ratios, so 30% comes back as 0.3. A page returns at most 40 terms and can only be sorted by volume and a few related fields, not by growth, so narrow the volume band to see more of the fast risers. Add includeKeywords or departments to keep the list to your category, run it on the same day each week, and keep every week: the series is worth more than any one pull.

For the holiday quarter specifically, the same data read over twelve weeks is in what to stock for Q4.

FAQ

Is a search term that rose 90% this week a good product idea? Not on that alone. Check the 4- and 12-week growth, whether it is a brand term, who holds the clicks and how many searches sit behind it.

Why do so many risers show almost the same percentage? Because the growth rate cannot exceed 100%. Large jumps all land just under it; compare the ranks.

How often does this data update? Weekly. ABA weeks end on Saturday, and a new week appears a few days after it closes.

Can an AI agent run this watchlist for me? Yes. The endpoints are available through the remote MCP server, so an agent can pull the week's risers and their trend series in one session.