As we all recognize, Google isn’t always the simplest search engine used by customers worldwide and faces opposition in select nations from the likes of Yandex and Baidu.
There is but another search engine competing with Google – one which, for the maximum component, loads of search engine marketing experts in no way encounter because it’s confined to the Czech market, Seznam.
Seznam was founded in 1996 by Ivo Lukačovič and reached the top round in 2010. It was anticipated that the then populace of the Czech Republic (10. Five million) would be journeying the hunt engine month-to-month, if not extra often. However, since 2010, Google has gained more market dominance and is now the dominant player.
Unlike in other markets, Seznam and its numerous products hold a full market share (more or less 30% of the hunting market and processing 15 million daily queries).
Optimizing for the engine nevertheless calls for attention while getting into Czechia.
For any search engine competing against Google, apprehending how you approach the conflict may be useful (as an SEO) on your very own important thinking and idea procedures.
To gain insight into Seznam’s technique, I’ve been lucky to ask Tomáš Pergler, Seznam’s Search Division Director, a few questions about Seznam’s method to fashionable day seek and how their search engine processes modern-day JavaScript-powered websites.
Many external resources attribute Seznam to eleven market proportion of the Czech search engine market, down from approximately 30% in January 2010. Where does Seznam region itself in the marketplace in terms of balance? Dusan Janovsky has formerly quoted 25%; is that this nevertheless correct?
Tomáš Pergler (TP): It is difficult to mention, as Google does not publish its numbers inside the Czech Republic.
We guess that we have around 30% proportion on desktops but lose on cell devices. It could be very hard for us to compete with Google as its programs are preinstalled on all Android gadgets.
At the beginning of this year, advertising and marketing employer Evisions published its case observation about how its clients use Google and Seznam. It is best in Czech. However, I am certain you’ll be capable of recognizing the graphs.
When we (SEO professionals) communicate about optimizing a website, we tend to focus on 3 core regions: inbound links, technical, and content. Without gifting away the secrets and techniques to Seznam’s algorithm, may you enlighten us on how Seznam attributes those elements and establishes its ranking system?
TP: At a few stages of abstraction, we smash the trouble of relevance and ranking down into 3 essential dimensions: Accuracy, Usefulness, and Quality.
Accuracy offers what the consumer question is set, what the report or internet site is about, and how properly they correspond.
Usefulness is how beneficial the given end result might be for maximum customers. Some outcomes may be accurate but useless – out-of-stock merchandise, outdated news, lacking body textual content, etc.
Quality deals with usability, navigation, facts, shape, and credibility. Some effects can be accurate and beneficial but low best.
We classify each document or question record in all three dimensions independently.
Accuracy and usability are question-record-based, so they must be evaluated in query time.
Quality may be evaluated properly after downloading the document because it’s a file-based type simplest.
Each classifier utilizes specific features:
Link graph capabilities are critical for usefulness and excellent.
Body text size is one of the functions essential for use. In maximum cases, empty body text is useless.
HTTPS has a few important features.
User feedback capabilities are sturdy for predicting first-class.
Dictionary-based features are sturdy for usefulness and satisfaction.
Results visible within the SERP are purported to be accurate and beneficial already, so we rank them in particular according to quality.
Backlinks have ordinary importance for us. Each size makes use of them in a one-of-a-kind manner. When predicting accuracy, we want to determine what the web page or page is about.
Backlinks keep records about how the web page is remembered by users. The documents from anchor text on a particularly visited web page are precious because numerous people use them for navigation.
That’s why we need one-way links for navigational queries. However, extra backlinks don’t suggest more accuracy.
Usefulness is an exceptional undertaking completely. The link graph capabilities are robust here, which is why hyperlinks work in search engine optimization. But it should now not be pretty much extra backlinks = higher scores.
Historically, we’ve used a few mechanisms to expect which hyperlinks are right or natural and which are awful. However, it’s a completely complex task, and there are probably better ways.
Direct and indirect links from rather credible websites can undoubtedly affect the quality score of the goal. Highly credible websites don’t often promote backlinks; hence, they accept them as true, propagating through the hyperlink graph.
When dealing with accuracy, the content material itself is the maximum without difficulty to have a source of records. It essentially tells us what the web page is about.
A web page title and an advanced body textual content extraction are vital. We use BM25-like scoring capabilities, and we are aware of Czech.
How capable is Seznam in phrases of processing JavaScript websites?
TP: We have been experimenting with webpage rendering for a while now – several years ago, every snippet in the SERP contained a web page thumbnail (it became only a “layout” characteristic).
Nowadays, we use webpage rendering for better knowledge of the content and layout of the website.
It’s nevertheless achieved on a low extent of crawls, generally for news articles. From a rendered web page, we can extract its major textual content, fundamental image, or article release date as it should be.
In the future, we need to increase and balance the ratio of rendered crawling site visitors to the intake of the crawl budget of the crawler.
Now, an average rendered crawl of 1 website consumes tens of GET requests compared to at least one GET request for not rendered move slowly.
We use the cutting-edge solid chrome model (presently seventy-three. X) for web page rendering. This means that our crawler gets the same effects as an actual user.
How big of an element is mobile usability in Seznam’s rating determinations?
TP: Our first-rate classifier currently doesn’t distinguish between computer and cellular. However, we’re getting ready for a brand new nice rating.
Our raters will charge the exceptional without delay on their cell phones. We’ll probably locate some new features based on this information. We plan to incorporate them into our ranking version until the stop of Q3/2019.
How often does Seznam replace its rating criteria/algorithms?
TP: In the last 12 months, we completely rebuilt our foremost relevance version. Once. Then, there were some minor tweaks throughout the year.
Our group has grown up, and we need to move faster this year.
How does Seznam cope with an international internet site (i.e., one with more than one language variation, Czech protected)?
TP: The search engine crawler SeznamBot focuses on the pages our users can search for. That method digs the internet for pages in the Czech language within the first vicinity.
SeznamBot additionally crawls other pages to permit the search engine to reply to “global” queries – for example, navigation to international websites, international businesses, programming, films, social, etc. So, it additionally crawls the global net.
If a few websites serve content in a couple of languages on equal URLs, the crawler gets only one language model – preferably Czech.
Suppose the language variations are on hand via special URLs. In that case, SeznamBot might also slowly move several language versions of the pages – for example, the same content material at Wikipedia in Czech, Slovak, and English.
Does SeznamBot face any issues with crawling non-Czech websites?
TP: Unfortunately, we experience severe entry problems while crawling the international web. An increasingly wide variety of websites tends to dam all visitors except for GoogleBot.
For instance, we’ve communicated with ProjectHoneyPot.Org lately to whitelist our crawler’s IPs because SeznamBot is a wellknown search engine crawler, and blocking off it doesn’t bring any advantage.
It might be extraordinary if this article could assist in encouraging webmasters to allow SeznamBot on their websites so it can bring some visits from customers in the Czech Republic.
Or at the least to reveal to them that having a technical touch within the feedback phase of the robots.Txt record on their net is beneficial.
Moving far away from looking for a moment, how famous are Seznam’s other belongings (such as Novinky, Sreality, Mapy)?
TP: Seznam. Cz has a reach of ninety-five percent of the Czech Internet populace. More than 3.5 million people go to our domestic web page every day, and within two years, we managed to get nearly one million visitors a day on SeznamZpravy.Cz (our very own news carrier).
Some highlights from our wider merchandise encompass:
Email.Cz – The email service facilitates 71.5 million emails daily.
Firmy.Cz–Firmy is an established catalog of agencies with contact records and opinions, listing more than 670,000 corporations.
Kupi.Cz – An online cut price/coupon-code directory with over 10,000 discounts and 300 advertising flyers daily.
May. Cz – A maps issuer, at some point during the tourist season, will attain up to 1 million customers every day.
Novinky.Cz – The most famous online Czech information website, with 57% of the web information reading market proportion.
Sport.Cz – The most visited sports website in the Czech Republic, with over 1,600 articles and 160 online streams daily.
Stream. Cz is an online TV streaming provider with more than 35 million views consistently monthly.