Outside Google: where else people look for you, and what changes

Let us start with a number, and then immediately spoil it a little.

According to StatCounter, in August 2026 Google held 91.1% worldwide. Then Bing at 4.5%, Yahoo at 1.23%, Yandex at 0.99%, DuckDuckGo at 0.7% and Baidu at 0.62%.

These figures circulate everywhere. Almost never with the month beside them, and almost never with an account of how they were measured. So let us do that, because it changes how you read them.

Where those percentages come from, and why it matters

StatCounter does not run a poll and has no access to the engines’ own data. It has tracking code installed on more than a million websites, from which it records more than three billion page views a month, and it sees where each visitor arrived from.

So it is a sample. A large one, but a sample, and a sample with a particular shape: it measures the sites that installed that code.

To be fair, it says so itself. It states plainly that “no artificial weightings are used” and that this is “a conscious and deliberate decision”. It also states that for countries with a small sample “it is possible that there may be some bias in the data”, and that figures are subject to quality assurance and revision for forty five days after publication.

Hold on to that, because we will need it shortly.

The map, with a date on it

August 2026, by country:

Two things stand out immediately. In Russia Google comes second, by a wide margin. In China it is practically absent, at 1.41%.

And a third you might not have expected: in South Korea, Naver is almost level with Google. No longer first, as often written in pieces that have not looked at a number since 2020, but forty three points is not something you ignore.

And now the part where we have to be honest

Look at the China row again. Bing at 18.92%, and further down the same list Yandex at 11.41%. In China. Then Japan, with Bing at 28.29%.

Those numbers need explaining, at best. In one case, the sample of sites carrying a western analytics script inside China does not look like the profile of the average Chinese user. In the other, a high figure has more to do with software defaults than with what people prefer.

I could have left them out of the table and never shown them to you. I am leaving them in deliberately, because they are the best possible demonstration of what StatCounter itself said two sections ago: in countries with a small or peculiar sample, the percentages need reading with caution.

The useful way to read this table is not the second decimal place. It is the order of magnitude: there are markets where Google is not the door.

Bing is not a destination, it is infrastructure

Bing’s 4.5% looks like nothing. It is nothing, if you read it as a destination.

Except that DuckDuckGo writes on its own help page that “we have more traditional links and images in our search results too, which we largely source from Bing”. Which means work done for Bing does not pay into one share, it pays into several places at once.

That changes the arithmetic. You are not comparing 91.1% against 4.5%. You are comparing the effort of one job against the sum of the places that collect it.

One protocol that notifies them all together

There is a piece of this work that is cheap, done once, and then runs by itself: IndexNow.

It is an open notification protocol. When a page changes or is added, the site tells the engines rather than waiting to be discovered. According to its documentation, the participants today are Amazon, Bing, Naver, Seznam.cz, Yandex and Yep, and in its own words: “Each endpoint sends your submission directly to its respective search engine, and your submission will be shared across all IndexNow-enabled search engines.”

One notification, six engines. We do it on every deploy of this site, with a single command that runs afterwards.

Google does not participate. There, the sitemap and Search Console do the job, as always.

Let us talk about JavaScript, because that is where it is decided

And now the most useful conclusion in the whole article, the one I would keep if I kept one.

Google is the only engine that genuinely copes well with pages that build their content inside the browser. The rest, some less and some more, struggle.

Here is what that looks like in practice. Yandex runs a separate robot for page rendering, YandexRenderResourcesBot/1.0, which its documentation describes as the one that “loads resources for page rendering with JavaScript”. Separate, meaning the main robot does not do it on every page whenever it feels like it.

The practical result is harsh, and I have seen it: a site that looks perfectly fine to Google can look empty to another engine. Not wrong. Empty.

The fix is neither exotic nor expensive: have the server send finished HTML. That is all. This site does exactly that, so every crawler sees precisely the same thing regardless of what it can execute, and no engine needs any special work of its own. The same holds for the answer engines’ crawlers, which are a separate conversation and the same argument.

What we do not do, said plainly

Baidu. This is not SEO work, it is infrastructure and licensing work. As Cloudflare’s documentation puts it, “all websites with their own domain name that operate inside China must obtain a license”, and “this applies whether the site is hosted on a server in Mainland China or delivered to visitors in China through a CDN”. Without a licence number, providers “are instructed to shut down any website (often without notice)”. Anyone selling you Baidu SEO without having solved that first is selling you air.

Naver. A different ecosystem, in a different language, with a different logic. We do not work it, and I am not going to describe how it works on the strength of having read two articles.

So who does this actually concern

Let us be honest: most Greek clients, hardly at all. If your audience is in Attica, Google is the job and the rest is curiosity.

It changes if you run a property with guests from Czechia or Korea, if you export, or if you are in a sector with customers outside the West.

And here is the good news, which I saved for last: almost all of the work is the same work. Finished HTML from the server, clean structure, a sitemap, notification via IndexNow, correct hreflang if you have a second language. There is no “SEO for Yandex” as a separate project with a separate invoice. There is a site that does not require the engine to be clever in order to understand it.

Would you like to see whether your site looks the same to an engine that does not run JavaScript? It is a few minutes of checking and the result surprises people more often than you would think. Tell me.