Russian keyword research: intent beats volume

You collected ten thousand phrases. The export looks impressive, the projected reach looks enormous, and the inquiries are not there.

The substitution happened at the very beginning: someone took the size of the keyword set for the size of the demand. They are different quantities. A thousand people who are searching for your service right now and can pay for it are worth more than ten thousand who typed a word from your category for an unrelated reason.

Intent is the answer to one question: what does this person want to do when they type this. Buy now. Compare before buying. Understand the topic. Find it free. Get a job. The same word carries different intents behind it, and most of a wasted budget goes exactly where the word is yours and the intent is somebody else’s.

For a company advertising into Russia this matters more than it does at home, because you are usually reading the queries through a translation layer. The intent signal in Russian sits in places an English-speaking marketer does not look.

Why “more keywords is better” actively hurts

There is a durable myth that maximum traffic comes from collecting as many keywords as possible, especially long and precise ones. In practice it runs the other way.

Betting everything on exact match for the sake of cleanliness brings only a small share of the available traffic, because real queries are endlessly more varied than any list you can invent in advance. People do not type the way keyword planners assume.

The opposite extreme hurts too. Take broad two-word keys and launch them, and you open a gate for a flood of junk sitting inside them. In English “buy sofa” will pull “buy sofa cheap”, “buy sofa pallets DIY”, “sofa sales job”. In Russian the same two-word key pulls more junk than its English equivalent, and the reason is grammatical. Russian builds meaning through case endings rather than word order, so a two-word phrase matches a much wider set of reformulations, inversions and insertions than the same phrase would in English.

The working middle is this. Start with keys of three or more words, where the intent is already legible: “buy corner sofa inexpensively”, “sofa eurobook with delivery”. Traffic on those is cleaner because the person has already narrowed down what they want. Connect broad two-word keys later and deliberately, once the narrow ones have taken the volume and it is still not enough. And connect them with eyes open, knowing you will be cleaning the query stream behind them.

The point is not to chase the length of the list. It is to collect the formulations behind which there is a readiness to buy, and to cut off the ones behind which there is anything else.

Match types protect you less than you think

Many advertisers are certain that phrase or exact match reliably limits impressions to queries containing all the important words. It does not, and the misunderstanding costs money.

Advertising systems expand queries on their own: they substitute synonyms, add near-meaning formulations, and sometimes serve an ad on a query that has dropped a key word from your phrase entirely. Both Yandex Direct and Google Ads do this. You set a phrase with an important qualifier, and the impression happened on a query where the qualifier is absent and the meaning has changed.

Yandex adds a second mechanism on top: autotargeting, which serves your ads on queries that appear in none of your keywords at all. It lives in the ad group at the group’s bid unless you give it its own, which means a carefully worked keyword set and a broad automatic pick-up compete for the same money at the same price.

Two practical consequences follow.

First: fix what genuinely matters. Yandex Direct has operators that force the system to keep the words you need. Quotation marks fix the number of words. Square brackets fix word order. An exclamation mark before a word fixes its exact grammatical form, which in Russian is the operator that actually does the work, because a word can appear in a dozen inflected forms. Dosage is the principle: the fewer words you lock, the more traffic arrives through synonyms of the unlocked ones, and the more of it is accidental. Usually it makes sense to lock the main product word and its defining characteristic, and to leave commercial and geographic add-ons free, where synonyms are useful. Locking the entire phrase is more often harmful: you cut off the living reorderings and insertions people actually write.

Second, and more important than any operator: read the search terms report against what you specified. The real queries that triggered your ads almost never match your keyword list. That report is where you see where the system took your budget, and it is where negative keywords come from. Without that reconciliation you are advertising blind, trusting a label that guarantees nothing.

Group by meaning, not by spelling

Collecting phrases is half the job. Then they have to be laid out into groups, and this is where campaign structure most often breaks.

The typical error is grouping by a formal feature: by the presence of a word, or alphabetically. Queries with different intents land in one group, and a single ad tries to speak to all of them at once. It ends up speaking to none.

The right logic: a group is one intent, not one word. “Buy”, “price”, “order” is hot commercial intent, the person is ready to act. “How to choose”, “which is better”, “reviews” is comparison before purchase, the person is still thinking. “What is”, “DIY”, “free” is not your customer right now at all. These layers cannot be mixed in one group, because each needs its own message.

There is a Russian-language complication worth knowing. The word “цена” (price) behaves like a commercial signal in most categories, while “стоимость” (cost) skews more informational, and “сколько стоит” (how much does it cost) sits between them and shifts by category. A translated keyword list treats all three as the same word. They are not the same intent, and grouping them together blunts every ad you write for that group.

When the keyword set is laid out by intent, everything downstream gets simpler. Each group gets an ad that answers exactly that intent. To the hot one you show price and a fast order path. To the hesitant one you show an argument and remove a fear. And you allocate budget consciously: more to hot, less or nothing to cold. The campaign stops being a dump and becomes a map of demand.

Where Russian demand hides outside Wordstat

Wordstat is the habitual tool and a narrow one. Relying on it alone means seeing demand through a keyhole. Two honest observations about its limits.

The right-hand column of similar phrases is nearly useless for finding synonyms. It was designed for exactly that, for alternative names and professional slang, and in practice it returns noise and duplicates of what you already know. As a source of junk for the negative list it occasionally helps. As a source of synonyms, almost never.

Wordstat shows monthly demand with no built-in seasonal perspective. Demand for a great many goods and services lives by the calendar: tyres, heating, gifts, travel. A figure for one month misleads if you are not holding the annual wave in your head. Plan a budget on a single month and you will eventually miss a season by a quarter.

So keyword research is assembled from several sources rather than one. Search suggestions and related queries, the planners inside advertising systems, third-party clustering and synonym tools that find semantic matches better than the built-in column does.

And one separate, rich seam: your own search terms from your own statistics. People have already come to you through formulations that are not in your keyword list. Collecting those and adding them continuously is the most honest source of live demand there is. It is also the one source that is automatically in the right language, written by the actual buyers, with no translation layer between you and them. For a foreign company this single report is worth more than any amount of upfront research, and it only exists after you have spent some money.

What to do with your keyword set this week

A short route worth walking:

• Rebuild your priorities. Separate hot commercial queries from comparison and from informational. Make sure the money goes to hot first.
• Check the grouping. Open the ad groups and ask honestly whether each holds one intent. If “buy” and “how to choose” are mixed in one group, it is time to split it.
• Open the search terms report. Compare real queries against your keywords. Everything foreign in meaning goes into work: part into negatives, part into new groups.
• Account for the season. If your demand comes in waves, look at the annual dynamic rather than the last month, and build that into the plan.
• Widen the sources. Add synonyms and professional slang from third-party tools and from your own query statistics.

A keyword set is not a one-off preparation before launch. It is a map of who is looking for you and why, and it gets refined every month as you read what people actually type.

Whoever collects keywords by volume pays for other people’s intentions. Whoever collects by intent pays for their own customers.

Frequently asked questions

Sources

• Yandex Direct help, keyword match operators: yandex.com/support/direct/en/keywords/symbols-and-operators
• Yandex Wordstat, the demand research tool: wordstat.yandex.com
• Russian version of this article on belokrylovo.ru: Семантика по намерению, а не по массе ключей