In a fascinating ‘fireside chat’ type interview, Ilya Segalovich was interviewed about Yandex’ development.
In a frank exchange, Segalovich revealed that the name was originally an abbreviation for ‘Yet another Index’, and the motivation for building it was that there was no alternative – in the early 19990s the Internet didn’t exist in the state it is now, and the developers didn’t see huge market for it. Their first product was used by 100 people – Segalovich was both engineer and customer support.
There was no shopping index in Russia – the founders tried to make “something beautiful” to help.
Perceptions of Yandex differ inside Russia and outside – they are number one for news, shopping, payments and search in Russia, and don’t have a plan to change perceptions of the engine: in a disarmingly candid way, Segalovich implies that it is what it is – different to each user. As I listened I felt like I was taken back to the heady days of the Internet boom, when the web was a shiny, exciting new place.
It’s a breath of fresh air: Segalovich notes that it’s getting harder to compete, so they have to work harder now to stay ahead. Yandex was a big winner on the basis of its Russian language. As other competitors move in, it’s having to innovate. Whilst it’s increasingly using geotagging, it’s still trying hard not to dilute or spoil the results. Guessing user intent is a sophisticated process. Yandex was two years ahead of Google – it was trying simply to improve the user experience.
Yandex resisted personalisation for a long time – ‘in your face’ personalisation can alienate users. They start with regular search results (one size fits all), but tailor the results according to user habits such as a preference for free, or willingness to download. Measuring personalised relevance is hard – people can switch off personal results in Yandex, but few ever do.
Yandex has followed a Facebook initiative and allowed people to go and collect all of their data in a single zip file. Segalovich is calling for the other search engines to allow people to port this information across search engines – perhaps at a browser level. The searcher should own the data.
Yandex has been acquisitive, although it calls itself and SME by comparison to other monolithic search engines. For example, machine translation has become an important area of investment, and Yandex is also looking at speech translation.
Yandex recently launched in Turkey and already boasts around 3% penetration there – people who are hugely loyal. Yandex Panorama is, suggests Segalovich, a very useful tool for transport – like Streetmaps but more advanced. This has been a valued feature in Turkey. Adapting the engine when searches reach so few people, combined with the complexity of language means the engine has faced some interesting challenges, about which Segalovich is disarmingly honest.
From here, Yandex will expand country by country as they make the business case for it. They know that both Google and Baidu will be watching them closely. However, China has both language and government limitations, so we are unlikely to see and Yandex expansion there soon.
Ranking on Yandex is based on machine learning. There are no penalties for not having local content, Segalovich notes, but having addresses on the pages relevant to the search query really helps.
Yandex consistently wins on search content, and is trying hard to create training programs for newcomers. This openness and vision of a search is resonant of Firefox – a more open, quality driven service. Segalovich believes that Google isn’t very open on the search front. He noted that Yandex regularly wins, above Google, in search comparisons.
He notes that Google’s changes (Panda) are probably for the best for Google, but that Yandex will continue to serve quality content, no matter where it’s been found.
But it’s mobile that’s possibly the space to watch. Mobile, said Segalovich in the same breath as acquisitions, is important, but smartphone penetration in Russia is still low. Yandex is, however, already working on mobile, with an Android app already launched. He thinks, however, the learning from mobile will be interesting because of the user interaction with the screen – the signals may be very informative to the search engines.
We watch with interest.