SEO case study: From 6 to 122 Google clicks per week in ten weeks

In the week from 3 to 9 August 2026, bettersorted.de received 6 clicks from Google Search at 718 impressions. In the week from 12 to 18 October 2026, it was 122 clicks at 16,701 impressions, roughly twenty times as much. Four things made the difference: fixing technical errors, writing content based on real search queries, offering every page in seven languages, and evaluating Search Console data every week.

This case study presents the numbers honestly, including the setbacks. The starting point was small, and a new website grows faster at first than an established one. However, the sequence of measures can be transferred to almost any SME website.

Figures at a glance

The source is the performance data from Google Search Console. A click counts when someone clicks the website in the search results, an impression, when it appears in the results. The average position was around 17 in mid-October, so often on page two. That means the website is now shown frequently, but there is still a lot of room for improvement.

Phase 1: Fix technical errors

After the relaunch to a new system, the website was fast, but faulty for Google. The key findings:

• Incorrect language references:The references to other language versions (hreflang) pointed to non-existent addresses on translated pages. A crawler reported around 1,860 error pages.
• Truncated titles:For 63 posts, the page titles were cut off in the middle of a word; around 270 titles were too short or too long.
• Hidden answers:FAQ answers were only generated when expanded. As a result, search engines saw only a fraction of the answers.
• Outdated robots.txt and sitemap:Old files were obscuring the correct ones.

None of these errors were visible on the website. We found them using Search Console, a crawler, and spot checks in the source code. As you can see for yourself, the article explains Check your website: 12 points.

Phase 2: Content based on real search queries

Instead of guessing topics, we analyzed the search queries from Search Console: For which terms is the website already being shown, but in a weak position? This led to articles that answer exactly one question, for example about costs, data protection, or funding. Each article follows the same rules: answer in the first paragraph, clear subheadings, a table for comparisons, frequently asked questions at the end, and links to the relevant service.

Just as important was what we did not do: target two pages at the same keyword. When the funding page for BAFA consulting competed with posts for the same terms, its ranking dropped. We restructured the pages and linked them internally in a clean way.

Phase 3: Seven languages

Each page is available in German, English, Turkish, Swedish, Italian, Spanish, and French, with translated addresses. Part of the growth comes from Italy, Spain, and France, where posts about open-source tools such as Paperless-ngx or Ollama perform well. Machine translation alone is not enough here: technical terms and brand names are often translated incorrectly and need to be checked. More on this in the article Multilingual website.

Phase 4: Evaluate every week

Since September, we have been evaluating Search Console data weekly: Which pages are shown often but clicked rarely? On those pages, titles and descriptions are revised. Which search queries are appearing newly? These lead to new posts. This cycle of data, drafting, human review, and measurement is described in the article Website with AI content engine.

What did not work

• At the end of August, impressions briefly rose to over 1,000 per day and then fell back to around 300 in September. Google initially tests new pages; only the relevant ones remain in the long term.
• Many clicks come via informational queries (“What is …?”), which rarely lead to inquiries. Pages with purchase intent rank more slowly.
• Individual days fluctuate significantly. Weekly and monthly comparisons are meaningful, not daily values.

What you can take away from this for your website

1. First, the technical side: check indexing, titles, language versions, and loading time.
2. Then the data: set up and read Search Console, see Search Console for beginners.
3. Then content that answers real questions, each topic only once.
4. And then stay on it: measure weekly and make improvements.

We offer exactly this process as a service: AI visibility from bettersorted connects website, Search Console and AI so that your website is found better over time. The free SEO and GEO check.

Frequently asked questions

How long does it take for SEO measures to take effect?

Technical fixes often show results after a few days to weeks, once Google has recrawled the pages. New content usually needs several weeks to months before it ranks consistently. At bettersorted.de, clear effects were measurable after ten weeks.

Are 122 clicks per week a lot?

For a specialized B2B website, that is a solid start, but not a large number. What matters is not the number of clicks, but how many of them turn into inquiries. That is why we also measure inquiries and appointment bookings.

Can these results be transferred to any website?

The sequence can: technology, data, content, ongoing evaluation. How quickly it happens depends on the industry, competition, and starting point. A new website usually grows faster in percentage terms than an established one.

Were the articles written with AI?

The drafts are created with AI support based on search data and research. Each article is reviewed by a person before publication, and facts are backed up with sources. Google evaluates content based on its quality, not on how it was created.

As of October 2026. Figures from the Google Search Console of bettersorted.de.