Closing Related Keywords

In Yandex.Direct currently, Related Keywords and autotargeting perform similar tasks — they automate work with keywords and help quickly target the desired audience. Therefore we are closing Related Keywords. If you use Related Keywords, please check your campaigns in advance and, if necessary, enable autotargeting to avoid losing additional ad impressions and conversions.

Even for experienced professionals selecting keywords can take up to half the time they spend setting up campaigns. Autotargeting allows you to save resources at the start of your campaigns and work out keywords on the fly. It makes it possible to run ads with few or no keywords — you only need to write ad texts and set a bid per click. The system indexes ad texts and product pages and shows ads in response to relevant search queries and for the appropriate target audience in ad networks.

Here is one example. The Pulkovo Logistics Company tested autotargeting for its target audience in the B2B segment, which includes both logistics managers, business managers and private customers. As a result, the conversion rate for the ad campaign with autotargeting enabled was 14.9% — three times higher than in the campaign with keywords, and CPL was 53% lower*.

We are constantly working to improve autotargeting's performance, and not only by providing accumulated statistics for further machine learning. Special teams of assessors regularly check whether the ads match the queries that trigger impressions. This helps reduce the percentage of errors and minimize cases when the system misinterprets the meaning of the query. Learn more about how autotargeting works and how to enable it in the Yandex.Direct Help guide.

New names for match types

There will be another change: in the Search queries report synonyms will change to Semantic match, and for impressions triggered by a more literal correspondence between the query and keyword —  Word-for-word match. This update in Report Wizard will make statistics more transparent and allow you to better assess the system's performance.

What is a Semantic match?

In response to a user's query, Yandex Search shows a complete list of relevant responses that correspond to the query literally (repeating the same words) but also in terms of meaning. All of Yandex Search works according to this logic, and search ads are no exception. The comparison of keywords’ and queries’ meaning is handled by the Yandex search model, which is based on neural networks.

Ads can be served for search queries that match the meaning of your keyword. Let’s say, for example, that a user searches for "heater installation price." This user will also see results for the keyword "heater installation cost."

In the report, you can view the selected synonyms, or rather semantic matches, and add irrelevant ones to your negative keyword list. The number of unique or just very rare queries is significant, meaning that working through them manually is too time-consuming (and by extension, pointless). Using a robot, however, these queries can be worked through in real time (and this should be done). Semantic matching allows you to avoid missing such target queries and now gives an average of 15% additional clicks.

In general, automatic technologies for working with keywords help you respond to real demand in real time and work more accurately with your target audience. We will keep you updated about the next steps.

If you have questions, please write us — we will be glad to help!

* The full version of the case is available in Russian on sostav.ru