20 times measure: complete checklist semantic core test

Formation of each semantic core is held in 4 stages: selection of the direction of advance, the selection query, clustering and distribution of pages. Product Owner service Rookee Catherine Hindikaynen examine each of them through the prism of common mistakes that website owners.

Semantic kernel (CR) - a list of targeted queries your site. Simply put, these are the questions to which your site can give full, detailed response to the user. It seems to be a truism, but a good half of all errors, which will be discussed in this article are related just to this.

First stage: selection of the direction of advance

Error 1: not taken into account the margins of the goods / services of the company.

The ultimate goal of each advertising channel - it's not the position, and the traffic and profit for the company. The sooner we start to make a profit, the better. Therefore, it is important to prioritize areas in terms of margins.

Error 2: The lack of study sections of the forthcoming increase in seasonal demand.  

At the stage of starting to be understood that the selection of requests, preparation of technical specifications, coordination, optimization of pages, the calculation of improvements to the site and indexing of all these changes will take 1.5-2 months. Sections that are promoted by certain requests will enjoy the greatest demand in about 2 months. Here is their and should be promoted in the first place. It would be nice to understand clearly which sections will be most in demand after 2 months, 3, 4 and so on. However, some of the requests are obvious peak demand (for example, "buy a down jacket"), and some - absolutely not obvious ( "buy diapers" with peak demand in October).

To determine the popularity of requests by month, you can use the internal validation script or service Yandeks.Vordstat statistics.

Mistake 3: ignoring important infopovodov, which entail an increase in demand.

On the one hand, the demand can be formed by seasonality, with another - infopovodami. Let us consider a very simple example. the new iPhone release date in Russia is scheduled for September, respectively, during this period, and even before the demand for partition with iPhones will increase. In fact, infopovodom can be anything. And one more important thing - they can be your competitors' ads on television, on the Internet, any replicated action. In any case, all forms demand, and you can get additional traffic, use it for their own purposes.

The second stage: the selection of queries

Errors 1 and 2: the absence of requests promoted to the main page and the card catalog.

This characteristic error for beginners. Sometimes they forget that the requests should be promoted to the main page. But more often website owners ignore requests for the selection of the commodity page, referring to the fact that they have low demand. This is fundamentally wrong.

By the way, that the trade request is not popular. Notice how the search engine on request "to buy a washing machine LG» (requests highlighted in yellow rectangle):

This product queries on models, which also must be considered.

Error 3: The selection of requests for non-competitive catalogs of goods or services.

The site, which is moving on-demand "pencils for lips", having in the catalog are only two suitable products, will never be able to take a leading position in the search. Even if he has a beautiful design, it brilliantly optimized for search engines, anyway, he will give in terms of range. In this case it is better to spend time on other areas, moving into the top more competitive.

Error 4: The query is, and landing pages is not.

This is a fundamental difference organic dispensing with contextual advertising. For example, the query "buy red refrigerator" in the content block can lead to any landing page, including the homepage, or the general catalog. How appropriate - a different matter, but it is possible. It is quite another thing - an organic issue. If we introduce danyj query into a search engine, you will see only those pages that fully meet the user's query - they are just a red refrigerator. The conclusion is simple: You can not create a page - do not try to trim and optimize certain queries.

Error 5: the selection of common phrases for narrowly focused sites.

For example, the company sells spare parts. Do I have to move it on request "parts for machines", "spare parts for loaders"? The answer is - no. After all, for «parts for forklifts," with only parts of the engine, the site will not be able to fully respond to the user's request in terms of the range.

Mistake 6: Missing information requests in the semantic core.

All business owners want to move on to commercial demands, because for them it is important to sell. And rightly so: in commercial queries absolute priority. But the fact that requests for information have a big enough demand. How do you know how big?

With the help here is a straightforward design, as in the example with a request "bedding":

In Yandeks.Vordstat you can see how often ask about linens. If the number of requests is quite large, it is necessary to include them in the CR, and set up a special information section. It may be a page of questions and answers, the company blog, and so on. The algorithm is simple: Drive traffic to these requests, giving an answer to user questions, and then using the relink to those or other sales pages, you can convert this traffic.

Error 7: The position is already in the top - optimization is not necessary.

The search engines are evolving, new factors and ranking algorithms. Site page of the top-3, a month or two can not compete with the other resources that are constantly evolving.

Error 8: do not consider competitors' requests.

Even being on the crest of success, you need to constantly monitor and analyze the issue of tracking events on competitors landing pages.

Error 9 and 10: there are only frequency in the semantic nucleus (popular) and competitive demands. There are no low frequency and mikrochastotnye requests.

On the one hand, the frequency needs to bring more potential traffic. On the other, in high-demand to grant more advertisements than the low-frequency (low demand for Yandeks.Vordstat). In addition, there is always the big brands to compete with that is very difficult. This does not mean you have to completely abandon such requests, but to concentrate only on them can not.

The advantage of the low-frequency and mikrochastotnyh queries that are easier to bring to the top. They are less competitive and more conversion than the high-frequency queries these queries high purified inquiry (query impressions without phrases in which it is included) and the highest CTR. Why? Because they are better articulate the needs of users.

Error 11: do not count the requests that are already bearing traffic.

Going on a query, you can lose already gained traffic to other keywords. To avoid this, you can view Yandeks.Metrike, for whatever phrases pages of the site are already visible in the search, and add them to the semantic core.

The third stage: Clustering

To remember that this clustering, you need to understand two concepts.

Cluster - a group of inquiries, united by a common intent.

Intent - intent of the user upon request.

Practically, this means that all requests are divided into several clusters, subsequently matched landing pages. At this stage, there are two serious enough errors.

Error 1: clustering occurs based on logic, not statistics. One brief advice - do not believe Me, believe the analytics services of Yandex and Google, they're unpredictable. View issue, you can use a free service coolakov.ru.

Error 2: The lack of compatibility testing clusters.

Let us consider the following example. From the point of view of logic, two clusters of data respond to different needs:

However, on these requests from competitors to issue the same page.

Most likely, at the request of "illuminated letters order" users are much more likely to visit a page with volume light letters and surf there is information that made some kind of conversion action. On the basis of the search engines we came to the conclusion that according to inquiries need to show the same page.

Step 4: The distribution of the pages

We picked up the requests for priority topics and broke these queries into meaningful groups. Now it is necessary to allocate these groups through your site.

Error 1: The request is progressing only on the following page.

If you have two matching pages under one request, and it is impossible to bring the site to a single page, try to redirect the request to another page. Perhaps so you succeed.

Error 2: the same page promoted in Yandex and Google.

Not everyone understands why extradition Yandex and Google may be different pages. Here it is necessary to look back and remember how to set up the search engines. Initially Google was an information retrieval system, Yandex - commercial. Google is much easier to find and otranzhirovat page with a sufficiently large text, Yandex loves and is able to rank the sales pages. Our recommendation: in Google pay more attention to information pages in Yandex - commercial.

Error 3: the same page promoted in the desktop and mobile issuance.

Search engines, in addition to the adaptation for mobile devices offer another option - the formation and creation of adaptive pages, such as Google AMP and Yandex-turbo. The bottom line is that in these pages there is no extra scripts and images, due to this they are loaded very quickly.

So, we looked at all the stages of forming a semantic core. Chosen priorities, picked for their requests rasklasterizovali and accurately distributed them through your site. And here we come to the last error.

Error Error: very common SEO-experts are limited to one-time selection of the semantic core.

This is a wrong tactic. First, talk about it themselves search engines, such as Yandex. Every day 100 million Yandex set of queries that are rare or even unique.

The main question - how much time you need to update and verify the semantic core. Our recommendation - at least once a quarter. Why? Firstly, you need to add relevant queries. And secondly, some of your queries over time can become non-earmarked, as the issue has changed.

It is bad when you're wrong, but even worse, if it happens when you're warned in advance. Check your semantic core of all the mistakes that we have discussed today, and take some time to correct them.

Do it quickly will this checklist:

  • We take into account the margins of products / services.
  • We focus on seasonal demand.
  • Select queries to the home page.
  • Select requests for card products.
  • Select questions for sections with a sufficient range.
  • Select queries to the relevant landing pages.
  • Select target requests for narrowly focused sites.
  • We consider and select queries.
  • We monitor the positions of the top requests.
  • We select high and mid requests.
  • Select low frequency and mikrochastotnye requests.
  • Analyze competitors requests.
  • We take into account requests that generate traffic.
  • Klasteriziruem with a focus on search results.
  • When you use clustering services.
  • Change landing page for unsuccessful requests.
  • We analyze the issue in Yandex and Google.
  • Use AMP and Turbo mobile issuance.
  • Quarterly update the semantic core.

Subscribe to SearchEngines newsletter

preview The largest SEO-myth version of Bill Slavskaya

The largest SEO-myth version of Bill Slavskaya

Based on materials from the article Director of SEO-Research Go Fish Digital and expert Google Patents Slavskaya Bill (Bill Slawski...
preview The new format of the stored copy of the page in Yandex

The new format of the stored copy of the page in Yandex

In October 2019 the year Yandex sly, without any announcements and advertisements, replacing the stored copy of the page format...
preview How to calculate the CTR for a website based on data from Search Console

How to calculate the CTR for a website based on data from Search Console

Author: Luca Bares (Luca Bares) - Senior SEO-analyst, Wayfair...
preview Reference attributes nofollow, sponsored, ugc and their impact on SEO

Reference attributes nofollow, sponsored, ugc and their impact on SEO

Author: Cyrus Shepard (Cyrus Shepard) - expert technical SEO, a former employee of Moz, founder Zyppy.com...
preview How to make a website prominent food delivery service in the search Yandex: a local issue

How to make a website prominent food delivery service in the search Yandex: a local issue

As site food delivery service to survive in a competitive environment. A Beginner's Guide Author: Elena Turanova, content marketing referr.ru...
preview "Trick or Treat?" - how to please Google assessors

"Trick or Treat?" - how to please Google assessors

September 5, 2019 came the next version of Google Guidelines for assessors . Well, September 24th started the next ranking algorithm update - September 2019 the Update Core...