Why we send smaller images to mobile screens
A September step in CMS4: choosing the right image size and reserving its space before it loads.
Read articleP+P website launch
Notes after the November launch of the P+P website: why content, old URLs and visitors’ habits belong in a migration as much as the new design.
Search engines read what visitors never see: links in the code, hidden elements, page addresses. So when we migrated the P+P website, we checked that invisible layer too – every link has to lead to a page that exists.
A check like that is as much part of a migration as the new design. On 1 November 2026 we moved ppsro.cz from WordPress to TwoDo CMS4. The brief was the same pages, structure, content and look – on a clean technical foundation.
Every deviation from the original site is a deliberate decision, and it is written down.
The source of the migration was an export of the old website, not just what was visible in a browser. From it came a list of pages in both languages: services, certifications, ecology, careers, contacts and privacy. For each one we established where its content came from, which documents and contacts belonged to it and where old links pointed.
Certifications such as BRCGS, GLOBALG.A.P., KEZ and KLASA moved across with their scanned documents and descriptions for screen readers. Whistleblower information existed only as a PDF in the uploads folder; it now has its own page in both languages, with the PDF to download.
Links in the code: we checked the ones visitors never see as well, and only links to existing pages made it into the new site.
Language prefixes: English links go straight to the target page, with no extra redirect.
Clean URLs: we checked page addresses for invisible characters such as zero-width spaces. Where an address changed, the old one redirects to the new one.
Titles in the page’s language: Czech pages and tiles carry Czech titles, such as ‘Dodavatel služeb’ (Service Provider).
Alternative text: images have real descriptions in both languages.
No obsolete code: the tag for the retired Universal Analytics service was not carried over.
People keep old links in bookmarks, emails and search results. All 97 URLs of the old website therefore redirect permanently (301) to the matching new page. The www variant and the temporary CMS address lead to the main domain, which is also named by the canonical URLs, language alternates, sitemap and robots.txt.
The crawler rules also record a new decision: search engines and AI search may crawl the site, while crawling to train AI models is disallowed.
It is not enough for a new page to respond. It has to hold the right information, load its images, work on a phone and offer a sensible route to an enquiry. On launch day we checked all 48 pages in both languages and both error pages. An automated accessibility check of the Czech and English homepages, in light and dark mode, found no issues.
The logos and icons have exact dimensions so nothing jumps while loading, and the site carries structured data about the company. The site settings, including the cookie consent texts, are kept as source files in the repository. Changes can be compared with that source and restored from it, not retyped from memory.
List every page, document and old URL, including those reached only through links from other websites.
For each page, decide whether to keep it, clarify it or redirect it.
Decide what the new site must improve, and write down every deviation from the original content.
Check each language version on its own, not just the language switch.
After launch, check redirects, images, the phone view and the route to contact on the real domain.
Good modernisation builds on the value that already exists. We are using the same approach to bring the TwoDo websites together.