How to Read Search Console by Language

Alejandro Rioja
Alejandro Rioja
6 min read
TL;DR

Search Console has no language report, so a multilingual site gets one blended number that hides where the traffic really comes from. Filter the Pages report with a regex on the language folder, compare each locale for clicks, CTR and position, then look at one post across all of its translations. On my site, the Russian version of one post earned 1,362 clicks in 28 days, more than twice the English original's 590, from about 40% of its impressions.

Building with AI Agents

Learn how to leverage AI agents for marketing, sales and product. Every week I share what I learned with 5,400+ readers.

Search Console has no language report. If your site publishes in more than one language, the headline numbers average them all together, and the average is usually wrong about every single one of them.

My site runs in 13 languages. For 18 October to 14 November 2026, Search Console shows 7,918 clicks and 1,130,728 impressions across the whole property, a 0.7% CTR at an average position of 11.8. That one line says almost nothing. Split it by language and the picture changes completely. Here is the routine.

Step 1: filter the Pages report by language folder

Language versions usually live in folders: /ru/, /es/, /fr/. That makes them filterable.

  1. Open Performance → Search results and set the date range to the last 28 days.
  2. Click + New → Page, change the match type from “URLs containing” to Custom (regex), and pick “Matches regex”.
  3. Enter the folder, for example alejandrorioja.com/ru/, and apply.
  4. Write down clicks, impressions, CTR and average position.
  5. Repeat for each language.

English is the one that takes a trick, because it has no folder. Use “Doesn’t match regex” with every other language in one pattern:

code
alejandrorioja.com/(ar|de|es|fr|hi|it|ja|ko|nl|pt|ru|zh)/

Swap in your own domain and your own language codes. If your site uses subdomains or country domains instead of folders, filter on the hostname the same way.

Step 2: compare the locales on three numbers

Clicks tell you where the traffic is. CTR and position tell you why. Here is mine for the 28 days ending 14 November 2026, from Search Console:

LocaleClicksImpressionsCTRAvg. position
English (no folder)2,891980,0070.29%11.5
Russian1,93864,5213.0%8.6
Chinese80718,0274.48%7.8
Italian55544,7481.24%7.8
Spanish53770,0750.77%13
French34541,3060.84%11.1

Read it like this:

  • English has the impressions and the weakest CTR. About 593,000 of those impressions come from a single page, my SEO tools list, which shows at position 10 and earned 47 clicks. One page that ranks without being clicked drags the whole English row down.
  • Russian and Chinese are small and efficient. Russian shows up 64,521 times and gets 1,938 clicks. That is a smaller audience that clicks.
  • Spanish is the leak. It has more impressions than Russian and a quarter of the clicks, at an average position of 13. That’s page two, and it’s where the next hour of work should go.

I’d run this table before deciding which language to translate next, or which one to stop paying for.

Step 3: take one post and line up all its translations

Locale totals hide a lot, because each one is a different mix of pages. The cleaner comparison is the same post in every language. Filter the Pages report on the post’s slug instead of the folder, then add the Page dimension to see one row per language version.

My post on email services that don’t require phone verification gives this, same 28 days:

VersionClicksImpressionsCTRAvg. position
English59052,7631.12%8.7
Russian1,36221,0756.46%5.9
Italian2915,2625.53%5.6
French1286,0132.13%7.5
Spanish642,8302.26%8.3

The Russian translation took 1,362 clicks from about 40% of the English version’s impressions, and it did it two to three positions higher. The English original is the one I wrote first and the one I’d have bet on.

I can’t tell you from this table why Russian wins. What I can see is the query side: the Russian version’s top query, “почта без номера телефона”, has 213 clicks at position 5.2, and a close variant with “2026” in it has 166 clicks at position 2. Searchers in that language type a different set of phrases, and the translated page happens to match them. A translation that matches the local phrasing beats a faithful copy of the original, which is why I treat each language as its own page to optimize rather than a mirror of English. I cover the same split from the AI-citation side in GEO for multilingual sites.

Step 4: sort each locale into one of three actions

For every language, put it in one bucket:

  1. Position under 10, CTR above 3%. Leave it alone and copy what it does. Look at its title and description and ask whether the weaker languages could use the same pattern.
  2. Position under 10, CTR under 1%. The page ranks and people skip it. Rewrite the title and description in that language; don’t translate the English one word for word. Check that the title still fits the roughly 60-character display budget, because the same idea runs longer in French, Spanish and German.
  3. Position over 10. The page is on page two. Use the same process as in how to fix pages stuck on page two: pick the pages with real impressions first, change one thing, and wait 28 days.

Don’t change a locale just because its line looks low. A language with 18,000 impressions and a 4.5% CTR is doing its job.

Step 5: don’t read this during the rollout

Google’s November 2026 core update began on 10 November and was still rolling out when I pulled these numbers. Rankings move while it runs, and my Spanish average position slid from about 12 at the start of the window to 20.1 on its last day. That could be the update or ordinary noise, and I won’t know until it finishes. Save the table now as your baseline, then follow the routine in how to read Search Console during a core update before you touch anything.

Use Search Console for this, not GA4. My GA4 property reports 33,128 sessions for the same period, 69% of them Direct, with an average engagement time of 11 seconds. GA4’s own traffic-quality check marks that mixed, so I don’t trust it to say which language is performing.

The prompt I paste into Claude

Export the Pages report once per language and attach the files:

I’m attaching Search Console Pages exports for one site, one file per language folder, covering the same 28 days. For each language give me total clicks, impressions, CTR and impression-weighted average position. Then find every post that appears in more than one language, and list the ones where a translation earns more clicks than the original. For each language, list the five pages with the most impressions and a CTR below 1%. Output three tables and nothing else. Do not guess at causes.

Before you act on the answer, spot-check two or three rows against Search Console itself. Models misread CSV columns more often than you’d expect, and one bad row sends you to rewrite the wrong page.

If you’d rather ask for these numbers directly, SiteWide MCP is the connector I use to pull Search Console and GA4 into Claude, so I don’t need to export anything. The split above is a thirty-minute job either way.

Advertisement
Keep reading

Related posts

Keep reading

Get the AI playbook in your inbox

Learn how to leverage AI agents for marketing, sales and product. Every week I share what I learned with 5,400+ readers.

↵ ↵ to see all results esc esc to close