I launched a Next.js blog in 21 languages as a personal challenge and an SEO experiment. After the first week, Google had produced 439 impressions and 5 clicks, and the site had also started receiving impressions in Yandex.
I was excited by those numbers. But looking back, the more useful lesson is not that “multilingual SEO worked in a week.” The data was much narrower than that. It showed that some of my pages were being discovered and shown in search. It did not prove that the multilingual setup itself caused growth, that every language was performing well, or that the pages had established stable rankings.
What I actually launched
The project was a personal blog built with Next.js and published in 21 languages. Each article had 21 localized versions. I chose Next.js because I wanted a practical foundation for internationalized routing and for pages that could be rendered on the server or generated statically.
That framework choice helped me build the site, but it is important not to turn it into an SEO claim. SSR, SSG, or an i18n-capable routing setup do not make Google index a page simply because the page uses Next.js. They are implementation tools. The search engine still has to discover, crawl, process, index, and rank the resulting URLs.
The first week: 439 impressions and 5 clicks
The clearest early result was 439 Google impressions and 5 clicks. That works out to a click-through rate of roughly 1.1%, so the click volume was still small.
What interested me more was that visibility had started across a multilingual site so soon after launch. With 21 localized versions of an article, I had many more possible search entry points than I would have had with one language. But “more possible entry points” is the careful way to describe it. It is not the same as saying that 21 languages multiply rankings or traffic by 21.
This was my first important correction in how I read the data: impressions are not traffic, and early impressions are not proof of a durable SEO strategy.
The indexing numbers came later
At a later check, after several weeks, I was looking at 471 pages. My articles had 21 language versions each. Of those 471 pages, 436 were indexed and 35 were still not indexed. In other words, roughly 93% of the pages in that snapshot were indexed.
That was encouraging, but it also taught me to separate another pair of concepts: indexing is not ranking. A page being in Google’s index does not mean it will rank for useful queries, receive impressions consistently, or generate clicks. Conversely, a page that is not indexed yet is not evidence that the site has been “penalized.” It is simply a different state that needs to be investigated separately.
My sitemap workflow
For this project, sitemap.xml was generated automatically, and I submitted it manually in Google Search Console.
I originally treated that step as part of “getting pages indexed.” A more precise way to think about it is that a sitemap helps search engines discover URLs and gives me a place to monitor sitemap processing. Submission is not an indexing switch. Google can know about a URL without indexing it, and an indexed URL can still receive almost no search traffic.
That distinction matters even more on a multilingual site because a single article produces many URLs. At that point, I do not want one vague metric called “SEO performance.” I want to know which URLs were discovered, which were indexed, which received impressions, and which actually earned clicks.
What the 21-language experiment taught me
- Multilingual publishing increases surface area, not guaranteed performance. More localized pages create more opportunities to match searches, but every locale still has to earn visibility.
- Framework features are not ranking factors by themselves. Next.js made the implementation convenient for me; it did not remove the need to evaluate the pages as search documents.
- Index coverage and search performance belong in separate reports. “436 indexed” and “5 clicks” describe completely different parts of the system.
- Early data is useful, but easy to overinterpret. One week was enough to confirm that Google was already showing some pages, not enough to declare the experiment a long-term SEO success.
- Scale multiplies mistakes too. When one article becomes 21 localized pages, a structural, translation, metadata, or linking problem can also repeat 21 times. That makes validation more important, not less.
How I would evaluate a multilingual launch now
If I were judging the same kind of launch today, I would keep four questions separate:
- Can search engines discover the URLs? Check sitemaps, internal links, crawlability, and response status.
- Which URLs are actually indexed? Track coverage without confusing it with ranking.
- Which locales receive impressions and clicks? Compare languages instead of treating the whole site as one number.
- Are the localized pages genuinely useful? Check the translation, titles, metadata, language targeting, and the page itself rather than assuming that producing more URLs creates value.
For multilingual sites, I would also verify the language-version signals such as hreflang and make sure each locale has a stable, crawlable URL. That is implementation hygiene, not a guarantee of rankings.
The result I actually trust
The most defensible version of this story is less dramatic than my original framing, but more useful. In the first week, my 21-language Next.js blog received 439 Google impressions and 5 clicks, while Yandex also began showing the site. Several weeks later, 436 of 471 pages were indexed and 35 were still pending.
Those numbers show that the site was being discovered and that most of the measured pages eventually entered Google’s index. They do not prove that multilingual SEO caused rapid growth, that Next.js accelerated indexing, or that every language version performed well.
That became the real lesson for me: multilingual publishing can create a much larger search surface very quickly. Measuring whether that surface turns into durable rankings and useful traffic is a separate job.
Blog URL: https://blog.jsvar.com
YouTube video: https://youtu.be/Zk8thnf0kk4