In Yandex. Direct statistics, “normalization of word forms” is the rake on which you can fill a lot of bumps without noticing yourself, getting a drop in CTR and an increase in bids in an advertising campaign.

Implicit normalization

Unfortunately, few people know what Yandex is.Yandex. Direct can “confuse” seemingly different words with each other.

The first example of such normalization that I personally managed to encounter is “сахар“. As you can see, statistics are glued together from several logically different topics:

  • сахар — food product
  • сахара – the Sahara Desert
  • сахаров – surname “Sakharov”

It’s the same word for Yandex. Direct. (There is also a large layer of queries for analysis, but the problem here is not in word forms, this is treated with the usual use of negative keywords)

This is how the normalization mechanism works. When words can represent different parts of speech — they stick together.

In such cases, you should force the correct word form to be specified using the “!“operator.

Unwanted word forms

Different parts of speech (adjectives. verbs, etc.) can change the meaning of the phrase depending on their word form. As a result, due to normalization, we may not see the share of the audience that puts a different meaning in the keyword we are interested in.

An example is an online store selling Iphone 6. The most logical solution seems to be, among others, creating a campaign with selling queries-queries containing the words “price”, “buy”, etc. Usually, such campaigns set an increased CPC, assuming a higher CTR and higher conversion rate.

However , when using the keyword купить iphone 6, they are also shown купил iphone 6 for the and queries купила iphone 6. This is how male and female gender plus elapsed time reduce CTR for the main phrase, which as a result harms the budget as a whole.

So in such phrases, it is advisable to use constructs like: купить iphone 6 -!купил -!купила

Of course, in this example, the share of such requests is not high, but:

  1. Remember – “a penny saves the ruble”. By saving on small things, you can generally win very significantly.
  2. There may be topics where the share of such requests is significantly higher, or their impact is significantly greater. For example, the difference in the forecast of the cost of traffic when placed in a special placement by keywords взять кредит -!взял -!взяла -!взяли is ~250,000 rubles.


  1. Always collect the most complete semantic core possible and consider it for gluing together
  2. Always think about what your keywords will turn into when you change the word form