Why did we need a separate specialist page?
webpot.hu covers all of our services, from website development to building automation. A site that broad struggles to become an authority in one narrow field, and the owner of a Magento 2 webshop is looking for a specialist, not a general agency. That's why Magento got its own domain, devoted entirely to it: development, hosting, migration and a server tuned for Magento. webpot.hu points to it as an umbrella, while the deeper technical material lives on magento2.hu.
Content structure
One question, one page
Every service got its own page: development, B2B webshop, maintenance, hosting and VPS, migration, a free assessment and webshop creation. One page answers one search intent, so two of our own pages never compete for the same term. Search engines can also see clearly what each page is for.
Knowledge base in three topics
The 16 articles fall into three groups:
- Magento development: custom modules, APIs, multi-source inventory (MSI) and POS integration.
- Performance and server: speed optimisation, VPS, Redis, Elasticsearch and OpenSearch, Varnish.
- Hosting and security: migration, version upgrades, PHP 8.3, security, troubleshooting and moving to a VPS.
The articles link to the service pages, and the service pages link back to the articles. This way, visitors travel from their question to the solution, and search engines see a connected topic, not a scatter of separate pages.
A glossary for machine understanding
The two main service pages (development, and hosting and VPS) explain technical terms with a short definition. This is also passed on as DefinedTermSet structured data. That way, search engines and language models see not just the text, but also which concepts the page covers and how it uses them.
Technical foundations
FAQs and structured data on every page
Every service page and every article got a question-and-answer block marked up as FAQPage. It's worth knowing that Google has not shown FAQ rich results for commercial pages in search listings since 2023. So the benefit of the markup isn't a flashy search snippet, but that machine processing gets clear question-and-answer pairs, which an AI search engine can pick up verbatim.
Fast delivery without plugins
WebPotCMS serves pages with several caching layers (APCu, Redis, Varnish) and automatic AVIF and WebP image conversion. Since there's no template or plugin stack behind it, a page loads only the code it actually needs.
llms.txt in Hungarian and English
An llms.txt file at the site root sums up for language models who the provider is and where the most important pages are, with a separate language version on the Hungarian and the English domain. There's no reliable measurement today of how much this actually matters. It's cheap to make, so we did it, but we're not relying on it.
The results, measured
Search Console data is available from 22 January 2026. The last two 28-day periods combined, from all countries:
| Period | Clicks | Impressions | Average position |
|---|
| 28 July – 24 August 2026 | 7 | 93 | 7.7 |
| 25 August – 21 September 2026 | 9 | 346 | 13.8 |
The drop in average position here isn't a bad sign: in the second period the page appeared for far more and broader searches (for example, for the bare word “magento” and for searches from abroad), and these pull the average down. Filtered to Hungarian searches in the same period, the average position is 7.9, and the main terms stand as follows:
| Search query (Hungary, last 28 days) | Impressions | Average position |
|---|
| magento 2 | 20 | 2.5 |
| magento fejlesztés (development) | 2 | 2 |
| magento webáruház (webshop) | 3 | 3.3 |
| magento webshop | 8 | 7.9 |
| magento | 74 | 16.5 |
AI search engines: yes for one question, no for three
On 19 September 2026, we measured with four fixed questions whether the company appears in the web search layer, the results list that AI answers mostly draw their sources from. For the question “Melyik magyar cég foglalkozik Magento 2 teljesítményoptimalizálással?” (“Which Hungarian company specialises in Magento 2 performance optimisation?”), magento2.hu was among the results, cited third in the summary. It did not appear for the other three questions (Magento 2 developer; Magento with VPS combined with technical SEO; custom webshop specialist). Without a web search, the company doesn't feature in the models' trained knowledge. This is a single measurement at a single point in time, so it's only indicative; we repeat it monthly.
What we haven't achieved yet
- Click numbers are small: 18 over 90 days. Impressions are growing, but clicks haven't followed yet.
- Most of the articles get barely any impressions; the migration, Redis and OpenSearch articles, for example, sit around position 50 or worse.
- For the broad “magento” term, we're around position 16, on the second page.
- Of the four AI questions, we appear for one, exactly the narrowest one.
The next steps follow from this: strengthening internal links between the articles, gathering external professional mentions, and a monthly measurement, changing one thing at a time so we know what's moving the numbers.
What can you take to your own site?
- One question, one page. If two of your pages compete for the same term, both end up weaker.
- The answer belongs in the first paragraph. Both search engines and AI search engines prefer to work from the top of the page.
- Structured data, but no miracle cure. It helps machine understanding, but it doesn't bring rankings on its own.
- Measure before and after. Change one thing at a time, or you won't know what worked.
- Consistent business details everywhere. The same name, address and services on your website, in your Google Business Profile and in directories.