Why you should pay attention to YAN
Contextual advertising in the search for Yandex, as a rule, is more effective than placement in the advertising network. This is due to the fact that the mechanism for displaying on thematic sites is slightly different: the advertising system analyzes the content of the pages and shows your ad in those of them where the key phrase assigned to the ad is most relevant to the content of the site.
Therefore, it can be argued that the user is interested in the subject of the resource, but not at all in the fact that he currently needs your product or service. From here and feature of work with keywords. If the search encourages the use of low-frequency semantics, designed for a ready-to-buy audience, then for thematic sites, higher-frequency key phrases of limited length (2-3 words) are recommended.
')
What we found
Testing commodity campaigns on YAN, a couple of years ago we received disappointing results. Although the CPC was at a reasonable level - 5.17 rubles, the share of advertising expenses (DDR) in the amount of 18.6% was depressing. Moreover, on search, the average DDR is 5%. Therefore, we decided to wait with the placement of offers on the network.
However, time is running, and Yandex is changing. A couple of months ago, we decided to repeat the experiment with our client's product campaigns - the federal chain of electronics and home appliances stores Tekhnosila. And they were pleasantly surprised by the results. It turned out that everything that sells well on a search is also well implemented through the YAN. According to the results of the recent placement of CPC amounted to 6.20 rubles, and the DDR decreased to 10.5%. What is especially important is that now the Yandex network gives a good return on budgets of hundreds of thousands of rubles. Earlier, investing more than 20-30 thousand rubles in YANA was meaningless.
At the same time, you need to understand that creating campaigns for a federal-scale store is a very difficult job, so we decided to share our experience.
What problems arise when generating large commodity campaigns for YAN?
Underwater rocks
Firstly, these are the same difficulties that await us when generating contextual RK on the search. First you need to collect information about all the product offers of the client and break them in some way by campaign. Then create thousands of contextual ads and upload them to the advertising system.
It is clear that it is impossible to cope manually with large amounts of data without an army of account managers. For example, the company "Technosila" - more than 17,000 SKU, information on which is updated dynamically. But even if your store or catalog contains a dynamic database of a couple of hundreds of goods / services, keeping the context with your hands turns into a difficult task, time-consuming. Without automation can not do.
Another aspect specific to YAN, is the importance of adding images to the ads, moreover, of a specific product advertised. Our experience shows that the use of individual product images helps to increase the CTR of ads in some cases by more than a hundred percent.
You can quickly add images to all ads using a data feed (in YML, GMC or another format). In the feed, each offer contains a link to the product image. These pictures are substituted into ads, “filled” into the context system through the API, adjusted to the required dimensions, and a highly relevant announcement with an attractive picture, text and necessary keys is obtained.
By the way, the presence of pictures in ads for thematic sites allows you to run shows in Vkontakte. And sometimes it turns out that posting through the YAN in the VC is much more efficient than posting on the social network through the native service of targeted advertising. For example, we had the experience of advancing antivirus with the target action of the “Installation”. The cost of a click when placing through a targeted advertising system was 12 rubles, and the cost of conversion into installations was 300 rubles. At the same time, the results of the campaign on the YAN (on the Vkontakte site) were pleasantly surprised: CPC reached 14 rubles, but the conversion cost dropped to 45 rubles. Thus, placement in VK through YAN turned out to be 6 times more efficient!
Choosing a generator
As we have already written, our client, Technosila, has more than 17 thousand offers, information on which is updated every 4 hours. For each product offer, you must specify the current price and discount based on the feed data. Moreover, these products are advertised both in the search context and on thematic sites.
For such a serious amount of advertising objects requires an appropriate technical solution. We stopped at the automation system contextual Garpun . The defining factors were the ability of Harpoon to “digest” feeds into tens of thousands of offers, as well as the ability to insert images into ads for YAN, pulling them up with links from a data source.
The advantage of Garpun is that it can be used to construct individual additional links for each ad. For example, if we advertise an Indesit refrigerator of a specific model, and the user is generally interested in Indesit refrigerators, then it would be nice to generate a corresponding pre-link leading to a customer site section on refrigerators of this brand. It is also useful to offer the client to look at the whole category of refrigerators without being tied to a specific brand. All these things can be done automatically through the Garpun system. Such additional links help to highlight the ad among all the others, increase its clickability and prevent the user from switching to an unnecessary section (which often happens when specifying additional links at the campaign level).
How it all works in RealWeb
Since, as we have seen above, working with large customers and large product databases is associated with a number of specific actions, Realweb has a special department for the automation of contextual advertising. The launch of the campaign takes place in several stages. The first is of course the negotiations and the conclusion of a contract on their results. Next we get a data feed with product offer information. Usually we recommend a YML file with an extended offer description. But sometimes it happens that even a large client does not have a data feed, so we create it ourselves.
Next, we study the client feed and, if necessary, give recommendations on how to refine it if the advertiser has technical capabilities. During our work with contextual advertising, we realized that half of the success of product campaigns is a high-quality data feed. For example, it happens that the feed contains the substitution of Cyrillic Latin letters, as a result of which incorrect key phrases are generated. Therefore, it is always necessary to conduct a comprehensive check of the data source.
Next, we outline the stages of launch and plan, based on the goals of the client. To do this, we determine the most marginal categories of goods, the average bill and draw up a plan for income. Based on these and many other parameters, we define priorities in the launch of advertising by categories. It is also very important to properly work out the structure of the account, especially in cases where you need to advertise tens of thousands of products. The structure of the account depends on the structure of the feed, the business characteristics of the advertiser and the types of semantics (broad requests, requests by categories, requests of the category “category + vendor”, requests related to a specific product). Proper account structure is essential for convenient campaign management and analysis of their results. As a rule, the partition goes into top-level categories and types of semantics.
After that, we start a campaign: set the feed in the Garpun system, define objects to uniquely declare ads, write text templates and construct templates for key phrases. In cases where the feed is not very high quality, before creating templates of key phrases, the feed tags are cleaned using the Spear filters. After that, it remains to set the campaign settings and load it into the context system.
Our recommendations
Finally, we want to give some advice to those who lead large commodity campaigns.
It remains only to guess what causes a significant increase in the efficiency of allocations on the YAN. We did not find any official reports of Yandex about this, but there is a feeling that the context system is actively working on targeting algorithms based on the history of search queries.
In any case, thanks to the successful experience, we have begun to more actively advertise clients on the partner network of Yandex sites, which we also want.
Therefore, it can be argued that the user is interested in the subject of the resource, but not at all in the fact that he currently needs your product or service. From here and feature of work with keywords. If the search encourages the use of low-frequency semantics, designed for a ready-to-buy audience, then for thematic sites, higher-frequency key phrases of limited length (2-3 words) are recommended.
')
What we found
Testing commodity campaigns on YAN, a couple of years ago we received disappointing results. Although the CPC was at a reasonable level - 5.17 rubles, the share of advertising expenses (DDR) in the amount of 18.6% was depressing. Moreover, on search, the average DDR is 5%. Therefore, we decided to wait with the placement of offers on the network.
However, time is running, and Yandex is changing. A couple of months ago, we decided to repeat the experiment with our client's product campaigns - the federal chain of electronics and home appliances stores Tekhnosila. And they were pleasantly surprised by the results. It turned out that everything that sells well on a search is also well implemented through the YAN. According to the results of the recent placement of CPC amounted to 6.20 rubles, and the DDR decreased to 10.5%. What is especially important is that now the Yandex network gives a good return on budgets of hundreds of thousands of rubles. Earlier, investing more than 20-30 thousand rubles in YANA was meaningless.
At the same time, you need to understand that creating campaigns for a federal-scale store is a very difficult job, so we decided to share our experience.
What problems arise when generating large commodity campaigns for YAN?
Underwater rocks
Firstly, these are the same difficulties that await us when generating contextual RK on the search. First you need to collect information about all the product offers of the client and break them in some way by campaign. Then create thousands of contextual ads and upload them to the advertising system.
It is clear that it is impossible to cope manually with large amounts of data without an army of account managers. For example, the company "Technosila" - more than 17,000 SKU, information on which is updated dynamically. But even if your store or catalog contains a dynamic database of a couple of hundreds of goods / services, keeping the context with your hands turns into a difficult task, time-consuming. Without automation can not do.
Another aspect specific to YAN, is the importance of adding images to the ads, moreover, of a specific product advertised. Our experience shows that the use of individual product images helps to increase the CTR of ads in some cases by more than a hundred percent.
You can quickly add images to all ads using a data feed (in YML, GMC or another format). In the feed, each offer contains a link to the product image. These pictures are substituted into ads, “filled” into the context system through the API, adjusted to the required dimensions, and a highly relevant announcement with an attractive picture, text and necessary keys is obtained.
By the way, the presence of pictures in ads for thematic sites allows you to run shows in Vkontakte. And sometimes it turns out that posting through the YAN in the VC is much more efficient than posting on the social network through the native service of targeted advertising. For example, we had the experience of advancing antivirus with the target action of the “Installation”. The cost of a click when placing through a targeted advertising system was 12 rubles, and the cost of conversion into installations was 300 rubles. At the same time, the results of the campaign on the YAN (on the Vkontakte site) were pleasantly surprised: CPC reached 14 rubles, but the conversion cost dropped to 45 rubles. Thus, placement in VK through YAN turned out to be 6 times more efficient!
Choosing a generator
As we have already written, our client, Technosila, has more than 17 thousand offers, information on which is updated every 4 hours. For each product offer, you must specify the current price and discount based on the feed data. Moreover, these products are advertised both in the search context and on thematic sites.
For such a serious amount of advertising objects requires an appropriate technical solution. We stopped at the automation system contextual Garpun . The defining factors were the ability of Harpoon to “digest” feeds into tens of thousands of offers, as well as the ability to insert images into ads for YAN, pulling them up with links from a data source.
The advantage of Garpun is that it can be used to construct individual additional links for each ad. For example, if we advertise an Indesit refrigerator of a specific model, and the user is generally interested in Indesit refrigerators, then it would be nice to generate a corresponding pre-link leading to a customer site section on refrigerators of this brand. It is also useful to offer the client to look at the whole category of refrigerators without being tied to a specific brand. All these things can be done automatically through the Garpun system. Such additional links help to highlight the ad among all the others, increase its clickability and prevent the user from switching to an unnecessary section (which often happens when specifying additional links at the campaign level).
How it all works in RealWeb
Since, as we have seen above, working with large customers and large product databases is associated with a number of specific actions, Realweb has a special department for the automation of contextual advertising. The launch of the campaign takes place in several stages. The first is of course the negotiations and the conclusion of a contract on their results. Next we get a data feed with product offer information. Usually we recommend a YML file with an extended offer description. But sometimes it happens that even a large client does not have a data feed, so we create it ourselves.
Next, we study the client feed and, if necessary, give recommendations on how to refine it if the advertiser has technical capabilities. During our work with contextual advertising, we realized that half of the success of product campaigns is a high-quality data feed. For example, it happens that the feed contains the substitution of Cyrillic Latin letters, as a result of which incorrect key phrases are generated. Therefore, it is always necessary to conduct a comprehensive check of the data source.
Next, we outline the stages of launch and plan, based on the goals of the client. To do this, we determine the most marginal categories of goods, the average bill and draw up a plan for income. Based on these and many other parameters, we define priorities in the launch of advertising by categories. It is also very important to properly work out the structure of the account, especially in cases where you need to advertise tens of thousands of products. The structure of the account depends on the structure of the feed, the business characteristics of the advertiser and the types of semantics (broad requests, requests by categories, requests of the category “category + vendor”, requests related to a specific product). Proper account structure is essential for convenient campaign management and analysis of their results. As a rule, the partition goes into top-level categories and types of semantics.
After that, we start a campaign: set the feed in the Garpun system, define objects to uniquely declare ads, write text templates and construct templates for key phrases. In cases where the feed is not very high quality, before creating templates of key phrases, the feed tags are cleaned using the Spear filters. After that, it remains to set the campaign settings and load it into the context system.
Our recommendations
Finally, we want to give some advice to those who lead large commodity campaigns.
- Conduct test placement on YAN, even if previously such campaigns did not bring you the expected result. Now the Yandex network is able to bring a serious amount of orders at an affordable price.
- Be sure to add specific offers and relevant additional links to ads.
- Take the time to create a data feed if you don’t have one yet. In the future, it will significantly reduce the cost of resources for creating and updating campaigns.
- Do not skimp on the system of automation of contextual advertising. For quite reasonable money you will save yourself a lot of time and be able to do more important things.
- Conducting contextual campaigns is a rather difficult task with a lot of nuances. If you are not confident in your abilities - contact the advertising agency. Professional context management services are quite affordable even for medium-sized companies. In this case, you will be able to negotiate with the KPI agency in advance - at the moment it is quite standard practice in any decent Internet agency.
It remains only to guess what causes a significant increase in the efficiency of allocations on the YAN. We did not find any official reports of Yandex about this, but there is a feeling that the context system is actively working on targeting algorithms based on the history of search queries.
In any case, thanks to the successful experience, we have begun to more actively advertise clients on the partner network of Yandex sites, which we also want.
Source: https://habr.com/ru/post/293360/
All Articles