Websites rarely go bad overnight. They age quietly: one more plugin, an image nobody compressed, a subpage last touched four years ago. At some point the enquiries thin out, and nobody can say exactly why.
The five signs below can all be checked objectively. Each comes with a self-test that takes about ten minutes and needs nothing but your phone, a browser, and some honesty about your own site.
1. It is unusable on a phone
For locally focused businesses, the projects we look after usually see 60 to 80 percent of traffic arriving from a phone. The site still gets tested almost exclusively on a large office monitor, with a mouse and a fast connection.
The self-test: Use a real device, not a browser window dragged narrow. Turn off Wi-Fi and stay on mobile data. Operate the site one-handed, standing up, and complete three tasks: find the scope of a service or its price, fill in and submit the contact form, call the phone number.
Where it usually breaks down:
- Tap targets smaller than 44 by 44 pixels. Menu items glued together are impossible to hit reliably with a thumb.
- Form fields with the wrong input type. A phone field has to bring up the numeric keypad, an email field the keyboard with the at sign. Autofill attributes remove half the typing.
- A phone number as plain text instead of a tappable link. On mobile this is one of the most common drop-off points there is.
- Font sizes under 16 pixels in input fields, which makes the operating system zoom in on tap and throws the layout off.
- Horizontal scrolling. If the page slides sideways, some element is wider than the screen.
Responsive does not mean everything gets smaller. It means the order still works: the most important thing sits at the top, navigation stays reachable with one hand, and nothing disappears just because space got tight.
2. It is too slow
Lab data and field data are not the same thing. A testing tool measures from a server under simulated conditions; that is lab data, and it is good for finding faults. What your visitors actually experience sits in the field data collected from real devices, which you can see in Google Search Console under Core Web Vitals. When the two disagree, field data wins.
The self-test: Mobile data instead of Wi-Fi, a private window for an empty cache, and a stopwatch. Measure how long it takes until you see something readable, not until everything has finished loading.
| Metric | Realistic target | What it describes |
|---|---|---|
| First visible content | under 1 second | The moment the page stops being blank |
| Largest Contentful Paint | under 2.5 seconds | When the largest element in view is in place |
| Interaction to Next Paint | under 200 milliseconds | How quickly the page responds to a tap |
| Cumulative Layout Shift | under 0.1 | How much content still jumps while loading |
The causes are almost always the same four:
- Uncompressed images. A 4 MB hero image uploaded straight from the camera. In a modern format at the right dimensions it would be 150 KB with identical effect.
- Too many plugins. Each one ships its own CSS and JavaScript, including on pages where its feature never appears.
- Embedded third-party scripts. A chat widget, two tracking pixels, a map, fonts from an external server: five connections to other people's hosts before your own text shows up.
- No caching. Every request rebuilds the page from scratch even though nothing has changed in weeks.
There is no credible cross-industry number for what slowness costs in bounce rate. The mechanism is not in dispute, though: someone who taps an ad on a phone and stares at a white screen for three seconds goes back to the search results. You will not find that loss in any report, because it happens before anything gets measured.
3. It does not explain what you offer or what happens next
The five-second test: Show your homepage to someone who does not know your business, for exactly five seconds. Then cover the screen and ask three questions: What does this company do? For whom? What would the next step be? Two of the three answers have to land.
The pages that fail hardest are homepages built as a picture puzzle: a full-screen slider, a mood photo, and a line like "Welcome to our site" on top. One sentence naming the service, the audience and the city beats any slideshow.
A website does not have to be liked. It has to make clear within five seconds who it helps and with what. Everything else is fittings.
Two things are almost always missing:
- A clear next step. One primary action per page, named after what actually happens: "Book a 15-minute call" beats "Learn more". Nine buttons of equal weight work about as well as none.
- Trust elements. Named references instead of anonymous logos, a real contact person with a face and a direct line, prices or at least a range such as "projects typically start at 8,000 euros". Hiding the price entirely loses you exactly the enquiries that would have fit the budget.
4. It is legally and technically out of date
The self-test: Open the Impressum, skim the privacy policy, then open your browser's developer tools and check which cookies are already set before you touch the cookie banner.
Common findings:
- An incomplete Impressum. Since May 2024 the mandatory details sit in section 5 of the German Digital Services Act (DDG), which replaced the Telemedia Act. Legal form, authorised representatives or the VAT identification number are frequently missing.
- A privacy policy that does not match the site. It lists services switched off long ago and says nothing about the ones actually running.
- A cookie banner that blocks nothing. If analytics or marketing scripts fire before consent, the banner is scenery. The benchmark is section 25 of the TDDDG.
- No HTTPS, or mixed content, meaning individual images and scripts still loaded unencrypted.
- An outdated CMS. A system last updated two years ago has publicly documented vulnerabilities. That is not residual risk, it is an invitation.
- Backups nobody has ever restored. A backup without a tested restore is an assumption.
On top of that, the German Accessibility Reinforcement Act (Barrierefreiheitsstärkungsgesetz) has applied since 28 June 2025. Among others it covers providers of certain consumer-facing electronic commerce services, meaning websites and apps where consumers can enter into contracts or book appointments. The technical benchmark is essentially EN 301 549, which points to WCAG 2.1 level AA; for services there is an exemption for micro-enterprises with fewer than ten employees and no more than two million euros in annual turnover. None of this is legal advice, and a lawyer should settle the question if there is any doubt. On the technical side you can check three things today: whether the site can be operated entirely by keyboard, whether contrast levels are sufficient, and whether every form field has a properly linked label.
5. Nobody can find it
"We are not on Google" means two completely different things in practice: either the site is not in the index at all, or it is findable but only for terms nobody types.
The self-test in three steps:
- Check indexing. Search for site:yourdomain.de. The number of results should roughly match the number of real pages you have. Ten results for eighty subpages means there is a technical problem.
- Search your brand name. For your own company name you have to rank first. If you do not, something fundamental is broken.
- Search in customer language. In a private window, type what a customer would type, district included. It replaces no proper analysis, but within a minute it shows whether you are in the running at all.
The usual causes: duplicate content, because the same page is reachable with and without www, with and without a trailing slash, or with parameters attached. Missing structure, because four services share one page instead of each getting its own page with its own title. No local anchor, because the Google Business Profile is unclaimed, or because the name, address and phone number are written differently in old directory listings than on the site itself. Fixing that NAP consistency costs an afternoon and often moves local visibility more than any rewrite.
Relaunch or renovation?
Not every diagnosis calls for a rebuild. Rule of thumb: as long as the design and the technical base still hold, targeted repair is cheaper.
| Area | Renovation is enough | Rebuild is cheaper |
|---|---|---|
| Design | Still looks current, a few pages are messy | Layout from another era, no system behind it |
| Tech and CMS | Current version, a manageable set of extensions | No longer updatable, an accumulated plugin stack |
| Structure | Navigation works, a few pages are missing | The architecture no longer reflects the offering |
| Content | Copy is usable and current | Copy describes a business that no longer exists |
| Brand | Unchanged | Name, positioning or audience has changed |
| Effort | Two to five days, clearly scoped | From roughly four weeks, but maintainable again |
The most expensive route is the third one: renovating for years what should have been rebuilt long ago. When you hear "the system unfortunately cannot do that" for the fourth request in a row, the decision has already been made for you.
What a relaunch must not break
A relaunch is not a design project with some technology attached. It is a move, and everything that worked has to come along.
- A redirect plan. Before launch, pull a complete list of the old URLs and point each one via a 301 to its content equivalent. No blanket redirect to the homepage, no chains three hops long, and the rules stay in place permanently.
- Existing rankings. Pages that bring in visitors from search do not get quietly merged away. If they do, the substance moves with them in full, not just the heading.
- Content that works. Look at the analytics before writing the concept: which entry pages actually produce enquiries? Those pages are the benchmark, not personal taste.
- Tracking continuity. Same tool, same goal definitions, same consent logic, plus an annotation on the launch date. Otherwise you end up comparing two measurements that have nothing to do with each other.
- Timing. Not two weeks before your strongest season, not on a Friday afternoon, and never during the company holidays. Plan two to four weeks of close monitoring afterwards, with someone reachable to fix what surfaces.
If more than two points apply
A single sign is an afternoon of work. Three or more usually means the foundation no longer carries the site, and every further repair costs more than a clean break. The only thing that matters is that the decision rests on measured facts, not on someone finding the site boring.
If you are unsure which of the five points apply to your site, we will review it once at no cost and send you an honest assessment with a priority list. What you do with it, and who you build with, stays entirely your call.