Yandex autotargeting and competitor brands: what it reads, what it risks

Autotargeting picks an audience not from your keywords but from the meaning of your ad. It reads the headlines and the text as a vector of intent and looks for people whose intentions resemble that vector.

A simple consequence that almost nobody holds in mind: the quality of your copy directly determines the quality of the audience it finds. Weak text, blurred vector, random people.

That is where work with autotargeting starts. Not “switch it on and hope”, but deliberately feeding the algorithm something it can understand you by.

How autotargeting reads your headlines

The old trick for preparing headlines for autotargeting was to drag one suffix down the whole keyword spreadsheet. Append “buy” or the city name to every phrase in the column.

The output is junk. “Buy sneakers buy”, “sushi Moscow Moscow”, “window repair prices price”. Headlines like that are useless to a human and to the algorithm that takes its intent signal from them.

Something else works. Instead of an identical suffix, insert real frequent phrases from your keyword set into the headlines.

The logic goes like this. Keywords group into clusters of meaning: in Russian, four different ways of saying “PVC windows” are one cluster of object plus material. Short category phrases are weak in a headline on their own, because there is no additional intent in them. So for that same cluster you look in neighbouring groups of your keyword set for fuller phrases with a geographic addition and hot words in natural order: “buy PVC windows in the city with installation price”.

The result is headlines filled with real, frequent, syntactically alive phrases of the same meaning, with no doubled words and no stumps. The algorithm gets a dense signal of the intent vector instead of a row of identical suffixes. The occasional awkward headline caused by odd word order in the source keyword gets fixed by hand. That is an exception, not the rule.

For an English-speaking team this is the step most easily outsourced to a translator and most easily ruined by one. The phrases have to come from what Russians actually type, not from a grammatical rendering of your English headline.

Where to get live phrasing

Headlines from the keyword set answer the question of phrase structure. What remains is content: which claims to put in the ad at all. There are two working sources, and they complement rather than replace each other.

The client’s own site. Sometimes this really is fastest. If the business has a substantial site, the keyword set is small and the deadline is short, it is simpler to write out the key claims straight from the pages, agree them with the client and assemble ads from them. Building automation for that is unnecessary; direct reading beats it on speed.

The client’s site has a built-in limitation, a survivorship problem. The site contains only what the owner decided to put there. Everything else that could have been a strong argument is invisible to you unless you are an expert in the niche.

A bank of claims from competitors’ ads. This is the more valuable source exactly where your own expertise is thin, which for a foreign company entering Russia is most of the time. Collecting ads across an industry, from hundreds to well over a thousand domains, gives you ready formulations about advantages, already refined by professionals in that market. Those claims get adapted to the client and travel further than search ads: into site copy and a social content plan.

A large array is not worked through by eye. A summary table by frequency shows, for each domain, the number of advertised URLs, unique headlines and claims, and the regions of display. Sorting separates active market players from random sites that happened to appear once. A relevant competitor is distinguished from an irrelevant one by industry words in the title, heading and description of the home page. One collection detail: do not scrape competitor activity at the weekend or at night, or the picture of activity skews.

Competitor brands: fine in keywords, not in the text

Advertising on competitors’ brands, where you appear on a query containing someone else’s name, is common and tempting. It carries two different risks, and confusing them is expensive.

The first risk is legal. Mentioning someone else’s brand directly in the ad text, in the headline, the description, or through automatic insertion of the keyword phrase, gives the brand owner grounds for a claim. Using their name as a keyword, with the ad shown without that name in the text, is noticeably safer. The line is simple: appear on the brand, do not write the brand. (This is how the practice is usually handled; for your own case, check with a lawyer who works with Russian advertising law.)

The second risk is metric, and it is more treacherous. Someone typing a specific brand expects that brand’s site. Landing on yours, part of them leave silently, and part leave an inquiry that falls apart on the first call, when the conversation with sales reveals they are in the wrong place.

Because of that, intermediate metrics lie. Click-through, cost per click, the number of inquiries, impression share can all look decent while the real value is zero. The only indicator worth following on that keyword set is the cost of a qualified lead, an inquiry that reached genuine interest in your product. Without a qualification step in the sales team you cannot separate it.

If your Russian leads are handled by a distributor rather than by you, agree the qualification rule with them before launching brand campaigns. Otherwise you will be reading a report that says the campaign works.

Technical control over how the ad is shown

Even strong copy can be shown differently from how you intended.

In responsive search ads there is an unpleasant surprise: the second headline is sometimes moved into the start of the description and turns into ordinary non-clickable text. It does not happen every time, so the same ad shows one headline in some impressions and two in others.

The fix is pinning a headline to a specific position. If at least one headline is pinned to the second position, that is where it appears. There is a trade-off: pinning several headlines at once can drop the built-in ad strength rating from excellent to average, because the system is built around free combination of all variants. Unpin them and the rating usually comes back.

The dilemma here is honest. In niches where you need to occupy as much text space as possible and to filter out the wrong audience in the headline itself through product specifics, a guaranteed display of the right text matters more than a formal rating. The system’s score is not revenue, it is a hint about revenue. Sometimes it is worth sacrificing so the person sees exactly what they should.

What to do with autotargeting this week

1. Remove identical suffixes from headlines. Replace them with frequent, living phrases from neighbouring meaning groups in your keyword set, with no doubled words.
2. Decide on your source of claims. Substantial client site and a small keyword set, read it directly. No expertise in the niche, build a bank of claims from competitors’ ads.
3. Appear on competitor brands, but keep their names out of your text. Judge those campaigns only by cost per qualified lead.
4. In responsive ads, pin the headline that must be visible and decide consciously what to do about the strength rating.

Autotargeting is not an independent force to be tamed. It is a mirror of your copy. Fill the ad with precise, living formulations and the algorithm will find you people with precise intent. Leave junk in the headlines and it will honestly bring you a junk audience at your own expense.

If autotargeting is driving irrelevant traffic and you cannot tell whether the cause is the settings or the copy, that is what I look at first in the free review.

Frequently asked questions

Sources

• Yandex Direct help, autotargeting: yandex.com/support/direct/en/keywords/autotargeting
• Yandex Direct help, ad text requirements and moderation: yandex.com/support/direct/en/moderation
• Russian version of this article: Автотаргетинг: польза, риски и чужие бренды