
How to an online store to stop the "drain" of the budget in contextual advertising

Ideally, contextual advertising leads to the store a "hot" audience. Users are already charged for the purchase, while they both come and go. A conversion above 3% is considered an occasion to open champagne) As a rule, the effectiveness is much lower: from 0.5 to 1.5%. Why is only 1 out of 100 shop visitors making a purchase and how to stop the "drain" of the advertising budget?
In this article, we will show 3 critical errors of contextual advertising, which almost all owners of online stores repeat from time to time. We will also analyze the case of a clothing store, where we managed to increase the conversion of orders from 1.3 to 3.2% due to the personalization of offers in product categories.
Initial data
Immediately emphasize that the "rake" of the advertising campaign listed in the first part of the article is characteristic of any niche. In our case, this is an online store of women's clothing with a simple structure: 15 categories of goods, 2,484 items. The Yandex.Direct Campaign is set up for 3,687 key requests distributed under 597 unique ads.
Mistake No. 1. Irrelevant landing page
As a landing page - either a catalog or a separate category. For example, in the “Dresses” category there are 240 items. Under them, 85 ads have been compiled, all lead to the corresponding page of the catalog. Despite the fact that buyers have much more selection criteria: “Dress is cheap”, “Dress with polka dots”, “Summer dress”, etc. It is understood that after the transition, visitors do manual sorting by six parameters (size, price, color, fabric, season, destination) or find a specific product through the search bar.

Most stores operate according to this scheme, despite the fact that making fine tuning is not difficult. Especially in color, season and purpose, excluding unnecessary action for users. The fewer steps on the way to the product, the higher the conversion.
It is enough to spend several hours to make a selection by parameters and to register the addresses of landing pages in accordance with the announcements. As a result, the traffic from the request “Buy a summer sundress” should be directed to a collection of summer sundresses, and not to the general category.
Mistake number 2. Ads for inappropriate requests
These are offers that absolutely do not match the user's needs.
For example, for the query “Dolls with clothes” show:

and even more so absurd for the query “Buy puppy cardigan” show an ad:

which leads to the next page:

Obviously, a person who needs a dog is unlikely to be happy with women's outfits. This is where the budget is wasted. At the same time, advertisers stubbornly continue to use homonyms (the same words with different meanings).
The information on the login page should be as close as possible to the request and ad text
Our customers had 49 inappropriate requests with a total cost of 4,450 rubles per month. It would seem a trifle, but with the cost of one client with Yandex.Direct 745 rubles, these are 6 buyers, each of which can become a permanent one. Read
how to reanimate a conversion for idle requests in this article .
Mistake No. 3. Traffic to the general page of the catalog
For a number of requests, traffic goes to the page of the catalog, where the visitor wants to be oriented. For example, for “Women's clothing is inexpensive” we see:

Moreover, the store has a section “Sale”, where sending users by this request would be much more effective. Another relevance related error. More precisely, with its absence.
Personalization of product offers using adaptive content
This is a way to increase conversion to orders without complex scripts and behavioral analytics.
We took the five most marginal categories: dresses, skirts, trousers, blouses, suits. In addition, they have the most contextual situations: customers choose the product by price (“Blouses are inexpensive”), style (“Plaid skirt”), color (“Yellow dress”), purpose (“Business skirt”), season ( "Summer women's pants") and size ("Women's suits of large sizes").
Combined requests ("Summer Dress with Polka Dots") were missed so as not to complicate the process. The texts of the ads were left as is, the changes affected only the content of the landing pages.
Firstly, we combined the requests into 140 groups related to customer needs and uploaded the Yagla service substitution table . For example, a user searches for a product by color:

In the initial version, he gets to a page in the “Skirts” category, where he has to do the subsequent sorting by color:

After setting up the spoofing, the landing page acquired the following form:

Traffic was sent to the page with the selection using the “yellow color” parameter ( similarly to other categories). But the most important thing is the replaceable heading “Fashionable yellow skirts with home delivery”, which contains an attractive condition - home delivery.
An important point:mechanical substitution of headers for queries does not give a strong effect. The decisive factor is the personalization of a trade offer with a favorable purchase condition (free delivery, fitting, discount, etc.).
More detailed fatal errors of adaptive content are analyzed in sensational material .
Another example, a group of requests for summer dresses (the selection parameter is the season):

A month after testing the substitution in five categories, the conversion to order increased from 1.3 to 3.2%. For comparison, the average conversion of online stores in the niche “Clothing, shoes, accessories” is 1.6% (data from the Enter center). Yes, the next step is that store owners plan to set up substitutions on the most margin goods cards. Follow the results in upcoming articles.
With you was Alexander Alimov, head of Yagla.ru
See you in a week!