There is a kind of damage in a website relaunch you cannot see, cannot measure, and only notice weeks later. It happens on switch-over day and settles in quietly.
A site that has been running for years has built something you do not carry over by copying content: its addresses. Every subpage has one, other people have linked to them, search engines hold them in the index. Those addresses are the actual property. Rebuild with new addresses and the old ones run into nothing. No error message, no alarm. Just a curve that bends downward three weeks later.
For Impresol Media Solutions we rebuilt a site that had grown over years. A publisher and an advertising agency in one, with content to match.
Every Old Address, Mapped By Hand#
The most laborious part of the project was a list.
That list assigns every address on the old site its equivalent on the new one. Where there is no direct equivalent it points to the next best page, not to the front page. The difference matters: a visitor looking for an article who lands on the front page is lost. They do not keep searching, they go back.
For a publisher this list gets particularly long, because a magazine does not have one address but many. Every single inside page of every issue is an address of its own on the web. The list ran long.
That is dull work. It is also the work that decides whether a relaunch becomes a step forward or a step back. Since then we do not switch over any grown site without that list standing and checked first.
Magazines You Leaf Through In The Browser#
A publisher making travel magazines has a presentation problem. The product is a printed issue. Offering it as a PDF download means sending the visitor out of the site and hoping they come back.
So the magazines sit on the site as a page-turning view. You click and turn the page like with a real issue, no download, no third-party program.
That is more than convenience. An issue running in the browser has an address for every page. That makes the content findable, linkable and readable for search engines. A PDF behind a download button is practically invisible to the web.
A Catalog Is Not An Image Grid#
Publications and maps sit in a catalog, and every entry has a detail page of its own. That is the decision that costs more than the alternative, and the only one that holds.
The alternative would be a grid of thumbnails with a pop-up window. Looks tidy, is quickly built, and has a catch: whatever sits inside a pop-up does not exist as a page of its own for search engines. You cannot link it, share it or find it.
A page per title means the opposite: every publication can be found on its own terms, with its own title, description and address. For a publisher with many titles that multiplies the surface on which the site is findable at all.
The Rest, Briefly#
On top come an area for the house's own apps, a blog, a wall of partners and clients, and an AI chat for the questions that come before an enquiry. The site exists in German, English and Spanish.
And it carries the full machine layer every site of ours gets: structured data, a plain-text file for language models, clean language markup. For a publisher that pays off twice, because a catalog of titles and issues is exactly the kind of information a model likes to read in structured form.
What I Tell Everyone Before A Relaunch#
If an existing site has been running for years and has visitors, the relaunch is not the risk. The switch-over is.
Three things have to be in place beforehand: the complete list of old addresses, a destination for each of them, and a look at the numbers from the period before, so there is something to compare against afterwards. It is unspectacular, it costs time, and it rescues years of work.
The nicer new appearance is worth nothing if half the people who knew the way get lost on the way in.
