Yandex.Wordstat is a free service from Yandex that shows how many times a month people enter a given query into search. It's the main tool for building a keyword core for the Russian-language market: without Wordstat, there's no way to know what words people search for you with.
The service address: wordstat.yandex.ru. The basic functionality is free, but there are limits on volume and operators — more on those below.
| Query | What it shows | Example |
|---|---|---|
| cake | all variations with the word | 142,000 |
| "cake" | only this word in any form | 12,400 |
| "!cake" | exact word form | 2,100 |
| cake -kids | a negative word excludes | 110,000 |
Where to start: registration
You simply can't open Wordstat and start searching — the service requires authorization. You need either a Yandex account or an account in Yandex.Direct (Yandex's advertising platform). A regular Yandex ID will do — the one you use for mail or Dzen.
If you already have Yandex Mail — just log in with it. If not — registering at passport.yandex.ru takes a minute.
After signing in, Wordstat becomes available.
The main mode: "Keyword selection"
Wordstat's main screen is an input field and two columns of results:
Left column: query expansions
This is where queries containing the entered word land. If you typed bicycle, the left column will show:
buy bicyclekids bicyclefolding bicycle- ... and hundreds of other variations.
This is the main material for a keyword core — expansions in which users specify exactly what they want.
Right column: similar queries
Here are queries that the same people searched for as your main one. For example, for bicycle: scooter, electric scooter, skateboard. These are adjacent topics, sometimes useful for expanding the core beyond the main word.
Important: the right column is actually less useful than it seems. It's session statistics, not semantic ones, and irrelevant products often end up there.
The numbers next to queries
This is the total frequency — how many times over the last month this query was entered into Yandex. If it says, for example, bicycle — 1,800,000, that means that many people entered the word "bicycle" in any context: those who wanted to buy, those who searched "bicycle photo," and those who mistyped.
To get a more precise number, you need operators.
Query operators: precise frequency targeting
This is the most important part of Wordstat. Without operators the numbers lie.
Quotes "query" — exact phrase
If you enter "buy bicycle" — Wordstat shows the frequency of exactly this phrase, without any other words added. No "buy kids bicycle" or "cheap buy bicycle" — only the exact match.
This frequency is usually 5-20 times lower than the total.
Exclamation mark !word — exact word form
If you enter !buy !bicycle — Wordstat shows the frequency of queries with exactly these word forms: "buy" (not "buying," "purchase," "bought"), "bicycle" (not "bicycle's," "bicycles").
This is the harshest filter — usually another 2-3 times lower than the exact-quote frequency.
Plus +word — force-include a stop word
By default Wordstat ignores "stop words" (prepositions, conjunctions, particles): in, on, and, how, what. If these words matter — put a plus. The query +how to fix a bicycle guarantees "how" is counted.
Minus -word — exclude queries with the word
The query bicycle -kids -used will remove results containing "kids" or "used." Useful for cleaning out obviously irrelevant topics.
Parentheses and | — grouping and "or"
The query (buy|purchase) bicycle will find both "buy bicycle" and "purchase bicycle" variants in one result. Useful for synonyms.
Combining
Operators stack: "!buy +with +delivery bicycle -used" — exact word forms, forced stop words, exclusion of "used."
Regions and devices
On the left in Wordstat's header are the toggles:
- "All regions" vs. a specific region. If you have a local business (say, a dental clinic in Tula), you need to look specifically at Tula: nationwide frequency is misleading.
- "All" vs. "Desktops," "Mobile," "Mobile only" (tablets). The modern pattern: for many niches 70-80% of queries come from mobile. If you're doing a mobile-first design — filter for mobile.
Query history: seasonality and trends
The "Query history" button (a tab at the top) shows a frequency graph for the last 2 years. You can see:
- Seasonality — air conditioners peak in summer, New Year goods — in December.
- Trends — whether demand for your topic is rising or falling.
- Sharp changes — for example, the spike in "masks" queries in spring 2020.
This tab often gives insights impossible to see in the main results.
Limits and captcha
Wordstat doesn't like active users:
- After 30-50 queries in a row it starts showing a captcha.
- With really intense use — it temporarily blocks the account.
- Scripts and parsers get caught regularly.
For one-off work this is fine. For systematic work with large cores there are three paths:
- Paid parsers — MOAB Tools, XMLRiver, Bukvarix API. They knock on Wordstat through proxies and captcha services, and you get ready data.
- Yandex.Direct API — the official way. Free, but you need Yandex's approval (an application through the Direct dashboard, reviewed in 7-14 days).
- Cloud services — like Semantic Core, Topvisor, Rush Analytics — they already know how to pull data from Wordstat and immediately assemble it into a proper keyword-core dictionary, without captchas and limits.
Suggestions that complement Wordstat
Wordstat isn't the only source of frequency. A good core is built from several:
- Search autosuggestions — what Yandex/Google show as you type. Often the queries there are more "alive" than in Wordstat.
- Search Console — the real queries your site is already found for. The best source, if the site has existed for any length of time.
- Metrica → the "Search phrases" report — the same thing.
- Bukvarix — an independent database of 2+ billion Russian-language phrases, frequency not in real time, but broader in coverage.
Common mistakes when working with Wordstat
- Looking only at total frequency and being surprised that a page with this high-frequency phrase brings no traffic. The exact frequency is 10-30 times lower — that's the real potential.
- Not filtering the region — gathering for Moscow what a site from Novosibirsk needs. Seasonality and demand can differ greatly.
- Skipping autosuggestions — gathering only Wordstat, losing 20-40% of "live" queries that don't appear in Wordstat.
- Not using operators — gathering "total" frequencies and then being unable to understand why the core is so bloated.
How we can help
If working with Wordstat by hand gets old, Semantic Core gathers the data for you in the cloud: you connect your site, we pull the expansions, filter by intent, cluster them — and every week we show how your positions for these queries have changed in Yandex and Google.
Free tools to try:
- Query clusterer — group already-collected keywords into topics.
- Intent classifier — figure out which of them are commercial and which are informational.