Amazon search volume: where do the numbers come from?

It counts searches, not shoppers and not sales. One shopper who searches “yoga mat” three times adds three. The number belongs to one Amazon store and one period: a US figure says nothing about amazon.de, and a December figure says little about June for a seasonal product.

Tools also count different things under the same name. Some show two numbers: exact volume for the phrase itself, and broad volume, which also counts longer searches that contain it. Jungle Scout’s Keyword Scout, for one, shows “exact and broad keyword search volumes.” Before you compare two numbers, check that they measure the same thing.

Amazon shares search data only inside Seller Central. Most of it is in Brand Analytics, a free set of reports for brand owners enrolled in Brand Registry, which needs a pending or registered trademark.

• Search Query Performance gives “the number of times customers search for a query”, by week, month or quarter. It is a real count, but only for queries relevant to your own branded products.
• Search Terms ranks every search term in a department by how often it was searched, over a day, week, month or quarter. That issearch frequency rank: a place in a list, not a count. It also lists the three most-clicked products for each term, with their click share and conversion share.
• Product Opportunity Explorer, under Growth in Seller Central, groups search terms into niches and shows their search volume and growth. Amazon says it refreshes niche metrics at the start of each week.

Outside Seller Central, Amazon publishes no search counts.

Outside tools don’t receive Amazon’s counts, so each one builds an estimate from its own data and its own model. Few say how. Keyword Tool is one that does: it calls its Amazon numbers “modeled estimates based on clickstream signals”, meaning data on what a sample of people search and click, scaled up to everyone.

That makes every tool’s number a guess with a method behind it. A good guess still helps: it tells you which of two keywords more people search. It can’t tell you exactly how many.

Because they start from different data and count in different ways. The usual reasons:

• Different samples. A model scaled up from one panel of shoppers won’t match one scaled up from another.
• Exact or broad. One tool counts only “yoga mat”; another adds “yoga mat for kids” and every longer search.
• Different periods. Last month, the average of 12 months, or a 30-day window all give different numbers for a seasonal keyword.
• Variants. Tools differ on whether plurals and common misspellings count toward the same keyword.
• Freshness. A number refreshed last week and one refreshed last quarter describe different months.

A gap between tools doesn’t mean one is broken. It means you are comparing two estimates.

• Compare within one tool. Its numbers are made the same way, so their order is more useful than any single value.
• Check the phrase is real. If Amazon suggests it as you type, real buyers search it. Amazon’s suggestions are the free evidence; see how Amazon autocomplete keywords work.
• Look at the trend. A 12-month trend shows the peak month, so you can update the listing before it.
• Put relevance first. A smaller phrase that describes your product exactly beats a bigger one that doesn’t.
• Trust Amazon’s numbers where you have them. If your brand is registered, check Search Query Performance for your own queries.

Shopperwords is a Chrome and Firefox extension for Amazon’s search box. The free version shows the keywords buyers type before and after yours, from Amazon’s own suggestions, with no numbers and no account.

Pro adds estimated monthly searches to every keyword, with a 12-month trend, peak month, competition and cost per click, and sorts the Before and After columns by volume. The numbers are estimates, not Amazon’s counts, and the extension labels them that way: use them to compare keywords, not as a promise of traffic. See pricing for the 7-day free trial.