What semantic SEO tasks should you tackle before ordering a website?

SEO First, Website Second: How to Design Semantics [Webinar Synopsis]

The methodology can be adapted for any type of project, be it e-commerce or informational. Relevant for sites that have not yet launched, and for young projects. For sites that are several years old, it can be used to update semantics.

For those who are more comfortable with the text, we made a summary of the webinar.

What to do before ordering a website

A common mistake among the owners of young sites is the lack of semantics. Often, projects on the Internet appear according to this model: first, an entrepreneur orders and receives a website, then he finds a SEO, and only then they decide how the site needs to be redone for SEO.

Before you start building a website, you need the first SEO steps:

  1. Collecting semantics.
  2. Clustering requests.
  3. Designing pages – what and where should be on the page. You can design together with the designer and the business owner.

When the structure and blocks that will be placed on the landing pages are outlined, you can start layout. The end result is a website ready for more aggressive promotion, link building and other further work.

If in commercial sites the owner without SEO knowledge can, on a whim, break down products into categories close to clusters that will result from working with semantics, then in information and other projects there is often a mess in the structure and the absence of a semantic core.

How to work with semantics before creating a website

The clustered semantic core should be the basis of the future site so that you do not have to redo the entire finished resource for it. Let’s figure out the main points of how to work with requests before starting the project.

Quantity is not equal to quality

For some information projects, you can always collect semantics, trying to cover everything that is on the topic, but never start working with the site. Therefore, you need to decide at the start how much semantics you need.

You don’t need to turn a site from a good niche project into a low quality Wikipedia, but with all the information on related topics.

A user who is looking for information on frying pans on the site most likely does not want to read about the origin of Teflon, but wants to choose a better frying pan. You can find a million keys that describe something that is on the site, but they will touch other user intents that we are not interested in. Without knowing the intent of your user, you can leave your topic in a wider field where there may be no traffic. The task of the webmaster is to reveal only the topic of the site, and not all related topics.

Does an online store need informational intents?

If you come across information requests when parsing keys, you can save them to use in content and then convert information traffic into sales. Many top stores are mastering informational queries and even launching contextual advertising on articles in order to then convert readers into buyers.

Commercial queries can be separated from informational queries by clearing semantics. Informational ones are usually distinguished by a set of words such as “how to do”, “how to choose” and so on. It is also worth checking the results and see what the search engines offer – commercial sites or articles. You should not try to get into the top with an article if there are only stores in the search results.

An example from Alexander’s work on a jewelry website that shows how important it is to watch SERPs: bounce rates grew for one product, although the site was in the top. It turned out that the top page for the query “penelope black diamond” got a page of a site with a ring that has such a name. But people more often looked for an actress from the genre 18+ who had such a pseudonym. The site went to the top for this key, but did not get into the intent. The situation was resolved by renaming the product.

It is necessary to clearly understand what queries can be covered on the site, which will help the user in choosing and provide the necessary information leading to the purchase. Otherwise, there is a risk of creating a lot of extra pages for one sentence and wasting resources there.

Does it make sense to collect semantics for a huge site?

For example, if the site consists of 10 million pages, it will be difficult to work with semantics, however, it may make sense. For example, in the case of a car parts website, someone will search for a spare part by car make, and someone by serial number, and it may turn out that the serial number is not listed on the site.

Alexander shared an example of a gas equipment website. Users search for some products by SKU. Not all articles are googled, but there is a certain set of articles that are usually searched for, with a low frequency – 30, 40, 50 queries. While the article was on the page, it didn’t even make it to the top 100. We put internal links with an anchor with the SKU – nothing happened. As soon as the article in brackets was added to Title, the pages took the top 1 and started collecting this low-frequency traffic.

When working with large sites, you should first analyze which pages are in the search, how much traffic they give, how many of the pages can be attributed to zombies – pages without traffic and impressions that have not been visited by search bots for more than six months.

Site structure and page weight

The task of the webmasters at SEO agency London is to reveal certain topics on the site. To do this, you need to form the site structure and design the page weight, distribute the link weight by priority. Ideally, if the webmaster can build such a model before the site is built, in order to highlight the semantics.

You can design the weight in special programs, Alexander uses WebSite Auditor. For a design example, a model with a small number of pages:


The site has a root and a main menu – in the diagram, it consists of two yellow circles.

Absolutely all pages of the site have links to the main menu, and the main menu links to two trees below it: T – products, red branch of the tree, C – product properties, blue branch.

Each product has a brand; in the diagram, these are orange circles on the right. In a classic menu that has links to all pages, those circles would be larger – brands would get more link power and higher priority pages would receive less.

The webmaster must understand what information or service he wants to provide to the client in the first place.

The client of this site is not looking for information about the brand, he is interested in specific products or their properties, so the brands were not included in the main menu, but kept in the structure in case someone is interested.

The main product and property sections have normal frequency and little competition, while brands are highly competitive. Even if the site manages to outbid the brand and go out on request with its name, brand sites, aggregators and other giants will still be in the top. This is not targeted traffic, the user who entered the name was most likely looking for a branded site, so they will leave your resource as soon as they find that the site is not the right one.

Usually e-commerce advises the SILO structure, where there are trees or page verticals that do not link to each other. Conventionally, if the site has a section with irons and refrigerators, there is no point in linking them. In information niches, this is not always realizable – in the diagram, for example, the blue and red trees have links to each other, this is done for the sake of usability and disclosure of the user intent, so that the client can conveniently get the maximum information that interests him.

Semantic page design

Designing a page is not about the technical part, but about where the titles should be, what they should be, and so on.

When collecting semantics and clustering, the webmaster works with a bunch of words with different frequencies and tails. It is usually impossible to use each key due to spamming. Let’s figure out how to design a page semantically.

For the title in the Title tag, we use the most frequent key. Next, we decide on subheadings, content and descriptions, for them you need to select the maximum frequency keys used the maximum number of times.

How to design a page semantically

This method allows you to capture the maximum semantics and give the page maximum significance.

  1. Break all semantics into separate words, not whole keys.
  2. Count which word how many times is included in the list of keys, and calculate its visibility, that is, significance.
  3. Use the most significant in Title and H1, in subheadings, less significant in content and descriptions. There are some low-frequency queries that could not be used in the text and meta tags, you can use them for anchors in internal linking.

How to count the number of mentions of words in the list of search queries:

1. Copy the keys to the adjacent Excel column.

2. Break this column by words: the function is under Data – Text by Columns – Delimited – Space as delimiter

There will be several columns with words.

3. Transfer all received words to a new page in one common column and remove duplicates using the Data – Remove duplicates function.

4. Transfer terms to the first sheet. Now the first column contains all the semantics, the second contains the frequency, then several columns with terms, and in the far right there is a list of terms without duplicates.

5. Let’s calculate the weight of the term – the frequency, taking into account the fact that it is present in several keys. We use the SUMIF formula, which goes through all the split keys, looking for a term and looking at the frequency from the first column. The formula looks like this:

Repeat for all terms.

The more frequent the term, the more meaning it needs to be given and used in the Title tag, content, H1 heading, and so on. The lower the frequency, the less meaning the term, we send it to the text and descriptions. Unimportant terms can be used in the comments on the page. If there is nowhere to insert low frequency terms, they can be used to dilute the inner anchor sheet.

6. Go through the list and highlight ideas that suggest terms: for example, that on a product card you need to place a title in Latin and Cyrillic, the inscription “Buy” and “Features”, “Review” or “Properties”, specify colors, add reviews and something else.

Video tutorial by Igor Shulezhko on the semantic design of landing pages

An example from experience

Alexander shared an example that working on semantics gives a good result.

He worked on a website for a beauty clinic in New York, Google categorizes such sites as YMYL and is especially strict with them. Initially, the site was made in WordPress, on a template with template images, there was no semantics, and all images were inserted as background.

The site was rearranged, the headings were written, the semantics were collected, the pages were optimized, the template images were inserted correctly – they carried out pre-project SEO, but with a ready-made project. We started building links using the crowd.

A month later, half of the semantics entered the top 50, 5% into the top 10, some important keys like “laser clinic” and “laser clinic + district” entered the top 10.


If you are going to order a website for your business, first study your competitors, collect and cluster semantics, work on designing a website, and then start developing it with ready-made recommendations from SEO specialists.