Yandex vs Google SEO: Report on the algorithm differences
Russian search market still has duopoly, which makes some difficulties in promoting the website in organic results.
Yandex and Google consider various ranking factors in different ways.
Therefore, in the long term to occupy high positions in both search engines, you need to look for compromises, and this requires more advanced approach to SEO activities.
In April, analysts of “Ashmanov and partners” agency presented an annual study of the main ranking factors for websites from the e-commerce category. I guess this category is the most interesting for most readers of my blog. Below I give the main conclusions from this report.
In recent years there is a trend in ranking of commercial websites in Google and especially in Yandex — algorithms prioritize large websites over smaller ones.
So, we have a vicious circle: those who are already big get even more traffic (for example, local e-commerce leaders: Ozon, Wildberries, Avito), and the rest should enjoy the crumbs of organic traffic.
So, banal but real advice: if you want to have a high position in Yandex and Google on a wide range of commercial key phrases, you should strive for leadership in terms of overall traffic. The more visitors you have in general, the better search engines will rank you.
In Yandex, the correlation of site traffic with the rankings can reach a record level of 0.3-0.35.
In Google, which denies the use of traffic factor in the ranking, the correlation with the position is not as strong (0.1).
If you start to send newsletters, to promote in social networks or advertising in search engines and achieve a serious increase in traffic, it can also pay off because of the growth of organic traffic from search engines.
To get into the top-10 on Yandex the large assortment in the online store is essential. In the top-3 search results websites with a wide range of products occupy 90-98% of positions. About 45-60% of them are pages with a huge assortment.
It is important that the website represents a large selection of brands. If you have a monobrand store, you have virtually no chance. The share of such websites in the top-30 of Yandex is about 5% (in Google — about three times more). And about the same proportion of sites offering only several brands.
The transition from medium to wide range, and from wide to huge can be very important for SEO. Of course, it is much easier to say than to get done. But if you don’t do it, other SEO activities can be ineffective.
It may seem that if you type a request for a product while being in Moscow, what’s difference whether this store works also in St. Petersburg and in other cities? However, the availability of a store in St. Petersburg for Yandex is one of the strong ranking factors (for Google — a little weaker).
In the top-3 of Yandex results in Moscow there are 67-75% of websites that also work in St. Petersburg, while in the top-30 — about half of that.
Again, it’s about size: a store that operates only in one location isn’t big enough, which generally means it’s also not reliable enough.
Other factors related to regions, addresses, and phone numbers work in a similar way: just a few regions and regions; a few addresses, and a lot of addresses and phones. All this, most likely, are indicators that the store has a serious offline projection. Which means you can trust it.
Category, product page or homepage?
Approximately 80% of the researched Yandex results are category pages, and about 15% are homepages. And only the remaining few percent of SERP positions get the product pages.
Of course, the type of pages you prefer depends on your request, but Yandex often displays category pages rather than specific products, even when the request contains a specific model. And only by supplementing the name (or number), model, color or memory size, you can get a page with a particular product.
Below is an example of search results page for “xiaomi redmi note 5” in Moscow on Yandex (left) and on Google (right):
As you can see, Yandex is dominated by pages listing the products of this model, and Google has mainly pages of one product. The titles of pages found in Google show the amount of memory; in Yandex it doesn’t, because on the found pages there are models of various sizes. Marketplaces and Yandex.Market are present both on the left and on the right, but these are different pages: in Yandex it is a universal landing page of this model, in Google — subpage of individual modifications.
The category page should contain:
- Optimized product names. They have a strong impact on the text relevance of the page.
- Brief technical information about the products. Also in order to improve the text relevance.
- Product photos. The number of photos on the website correlates with the position in Yandex and Google, and in Yandex also correlates with getting into the top-30.
- Buttons or links to add product to the cart.
- Filtering and sorting functions.
- Adding products for comparison.
The size of the category page should be large enough. Instead of showing a small number of products (for example, up to 10 pieces), it is better to show users and search engines as much as possible on one page.
Currently, inbound links are considered to be one of the strongest ranking factor in Yandex, and in case of key phrases related to e-Commerce inbound links correlate with a position stronger than in the overall sample. The number of total backlinks providing to the domain is much more important than the number of backlinks of a specific page of this domain.
In order for your store to withstand competition, you need to make people talk about it. The rule is simple: the more people talk about your store, the more links they post.
Pages that are at the top of search results for phrases related to products have more backlinks, than pages in the whole researched sample. For example, the website from the top-3 in Yandex has more than 3200 incoming domains, and in Google — 2850. In the top-30 Yandexa it has more than 400 ilinking domains, in Google it’s about 1000 (according to MegaIndex tool).
Not only the amount of links and its link juice is important, but also the whole link profile in general. It’s better to have less links providing only to the homepage, this way your backlink profile is more diverse.
The anchor factor works for online stores as well as in the general sample of queries, but even stronger.
In Google, the text-link factor is stronger than in Yandex.
For Yandex, backlinks to a domain in general (and their anchors) are much more important than backlinks to a specific page. The ranking concerns the domain, but not the single page.
For Google recently, backlinks to the domain have become almost as important as links to subpages.
Links should not contain exact key phrases. The broader is a key phrase, the stronger this factor is. The use of individual words from the key phrase in anchors has a greater correlation with the position in Yandex and Google than the exact queries in the anchor.
Correlations are even greater when you consider not only the words from the key phrase, but also their synonyms and words marked in the snippets in Yandex and Google.
Content and “Baden-Baden” filter
Online stores has suffered from Yandex’s struggle against poor texts the most. In their case the ranking formula has strongly changed and it became very different from the ranking in Google, which creates serious problems for promotion in both search engines.
In addition to product feeds, there were always SEO texts on the category page. They were made more for search engines than for humans, and its relevance was mathematically calculated. The required number of instances of each of the key phrases and the words contained in them was calculated in advance to create an appropriate task for the copywriter.
For a long time everyone was happy, and Google is quite satisfied even nowadays. But in the spring of 2017, Yandex has introduced the “Baden-Baden” filter, overlapping penalties for low-quality content (more about Yandex filters).
And now we need to understand how to cope with the ranking of online stores in Yandex and Google.
In both search engines:
- The further away from the exact query, the more important this factor becomes. For example, the presence of a key phrase in the text in its exact form for Google has a very weak correlation with the position (0.04), and the occurrence of individual words of the key phrase is quite strong (0.18). However, in the case of product key phrases, this principle works a little less.
- In addition to the words which the key phrase contains, textual relevance may consider the synonyms of these words or words highlighted in the snippets as well
- In addition to appearing in texts, the ranking may be affected by keyword occurrence in <title>, headings, links, description, and other areas of the document.
- However, the importance of areas that are traditionally perceived by SEOs as significant (headings
<b>, <strong>, and even
<title>) seems to be a bit exaggerated.
- But the appearance of keywords in the
<li>elements and internal links (and in their alt attributes) are essential.
- The page should contain a lot of words from the query and other words that are important for the ranking. On the average page from the Yandex top-30 there are so many words from a query that they would be enough for more than another 40 key phrases, and if we take synonyms into account — more than 46. For Yandexa, however, the number of words from query on the page should be estimated not counting the SEO text (see below).
- To make all this happen, the content must be large — the size of the text correlates with the position in both search engines, and in Yandex Is strongly associated with getting into the top-30.
If your site is not in the top of Yandex, perhaps this is because Yandex does not consider it relevant. If the position on Google is low — maybe the reason is that its pages are missing words from the key phrase.
Text factors are strongly associated with getting into the top-30, but correlate with the position much weaker.
It seems that Yandex evaluates the text relevance of the page in the early stages of the ranking, with a quick pre-selection of candidates. And in the final evaluation the text factors, even if taken into account, have a lower value.
For online stores, this is pretty logical approach. What does it matter how many times the product name is on the page? The most important thing is that it is really on sale, it’s not fake, will be delivered on time, has a guarantee, and that the price is reasonable.
Google text parameters are typically correlated with the position. This means that Google takes into account the text in the evaluation in much the same way as it takes into account other factors — for example, backlinks or commercial factors.
But there is one important exception. The exact form of the key phrase in the page title
<h1> is statistically related to getting into the top-30. That is, for Google, unlike Yandex, it is important that the query is in the title of the page in the exact form, without changes.
After Yandex coped with backlink spam, the search team decided to apply their experience to combat text spam — primarily it was about large blocks with optimized content on commercial websites.
To do this, it was necessary first of all to learn how to clearly distinguish the texts of different types — information blocks, menus, product tabs, specifications, etc. What is quite acceptable on the product card (for example, the repetition of words from the query), in the information block can become a marker of poor-quality text.
The next step after extracting SEO-texts was their exclusion from the ranking forumla — and maybe even adding them to the negative factors (the less such texts, the better ranking).
In the autumn of 2018, there was not visible correlation between SEO texts and the position in rankings, but there was a connection with getting into the top-30: there were much less words from the key phrases on the pages from the Yandex top rankings. Now there are quite strong correlations with the position: the closer to the top-1, the less words from queries are in the text. And for commercial queries, such inverse correlations are even stronger than for the general sample of key phrases.
It seems that even if a site is not filtered by “Baden-Baden”, over optimized SEO texts are not only useless, but also harmful for ranking in Yandex.
As for Google, SEO text has always been just text on the page. It didn’t matter where the robot met the keyword — in SEO text or, for example, on a product tab. Now, however, it looks like it’s starting to change. There is a weak correlation between position and key phrases in SEO texts. Much weaker than in the case of the whole text on the page.
Delete SEO texts or them leave for Google?
SEO texts interfere with ranking in Yandex and are practically useless for users. Why not just delete them?
The problem is that they can negatively affect your ranking in Google, because the relevant words from the query will be less, and in addition, the page loses even this minimum of useful information that contains this optimized text.
The cheapest solution is to close SEO text with tags
<noindex></noindex> (or comment
<!–noindex–>), without waiting until the site gets filtered by “Baden-Baden”.
In this case Yandex bot simply ceases to notice the text, and Google will index the text as if nothing happened, because it ignores this tag.
More difficult, but also a better solution requires a comprehensive approach:
- find out what information from SEO text is really useful for users;
- divide useful content into separate blocks, which do not have to be different on different pages — you can use standard content wherever it is appropriate;
- these blocks may be used as excerpts with links leading to pages where the desired topic is explained more widely, such as describing the purchase process, giving advice on product selection, their useage, etc.;
- the rest of the SEO text should be removed;
- recompense reducing the relevance of text to Google by increasing the number of products on a category page and increasing the complexity of its structure — for example, by adding more advanced product names and one-line descriptions.
Although Yandex and Google are developing in about the same direction, sometimes some nuances may differ significantly. In many aspects, the simultaneous optimization in both search engines requires additional actions, analysis and constant search for compromises.
A full report on the ranking factors for 2019 can be found here (in Russian).