2026 is almost done, and instead of the usual "we grew and everything was wonderful" routine, we're doing it differently this time. We're going to show you what genuinely worked at bytebrise and where we got it wrong. Because behind an agency there are no perfect processes, just people trying things and throwing some of them back out. A few of those decisions cost us real money. Others moved clients in Fuerstenfeldbruck, Olching and across the western Munich area meaningfully forward. Both belong in the picture.
What went really well in 2026?
The short answer: clearly scoped packages tailored to each industry. The long answer is coming, because there's a flop behind it that pushed us there in the first place.
The format that landed best is one we call "website plus care" internally. Not a one-off project you hand over and then never touch again. Instead, a site we keep looking after afterwards: updates, small text changes, a check on load times every few months. Sounds mundane. But for a trades business in Maisach or a medical practice in Germering, that's exactly the point where most agencies bail. The feedback on it was so clear that we made the model our default in 2026.
Local SEO also turned out to be the thing that delivers the most direct benefit to clients. Set up a Google Business profile properly, real opening hours, photos of the actual shop instead of stock images, actively collect a handful of reviews. It's not rocket science. One restaurant we look after used to sit on page three of Google for "restaurant Puchheim" and now ranks near the top of the local results. You don't see shifts like that overnight, but over six months the difference shows up in the reservation book.
And where did we miscalculate?
Now the uncomfortable part. Our biggest miscalculation in 2026 came down to project scope. We wrote several quotes where we pitched the effort too low because we'd "built something similar before." That comes back to bite you. Similar isn't the same.
A concrete example. We'd quoted a fixed price for a company website and completely underestimated how many rounds of revisions it would take, because three people on the client's side wanted a say and couldn't agree with each other. Two planned feedback loops turned into six. We paid for that with our own time. What we took away from it: we now name and cap the number of revision rounds in the quote. Not out of pettiness, but so both sides know where they stand.
The thing with flat-rate pricing
The classic among our failed attempts: early in 2026 we wanted to make life easier for ourselves and for clients by offering fixed flat rates across all industries. One site, one price, whether it's a bakery or a law firm. Sounded tidy. In practice it was nonsense.
Why? Because a restaurant and a tax firm have almost nothing in common when it comes to a website. The restaurant needs a menu that's quick to update, table reservations, good photos of the food, fast mobile load times because people glance at their phone on the go. The law firm needs trust, clean legal text, a clear structure covering services and areas of expertise, often a client contact form with data protection that actually holds up (German GDPR rules apply here). The same flat rate means one client pays for things they don't need, while the other is missing the essentials.
We scrapped the flat-rate model after a few months. Since then we've had packages built around the specific industry. That's more work in the consultation stage, honestly. But the quotes fit, and we no longer have to renegotiate afterwards because half of it is missing.
Which tools didn't prove their worth?
Not every tool that looks good survives the day-to-day. We threw two things back out in 2026.
- An AI chatbot for the client-facing website. The idea was nice: a little assistant that answers visitor questions. In reality, it sometimes gave flat-out wrong answers about opening hours or prices because it was making them up. For a service business, a wrong answer is worse than no answer. We only use something like that now when the answers come from vetted content, not generated freely.
- An all-in-one project management tool. We tested a new system meant to bundle quotes, tasks, time tracking and invoices. After a good four months we were back to a leaner combination, because the big tool ate more time in data upkeep than it saved elsewhere. Sometimes less really is less work.
What we kept: the foundation. Vanilla HTML, CSS and JavaScript instead of heavyweight builders. It sounds old-fashioned, but it's what keeps our sites loading fast and lets us keep every line in our own hands. On load times it makes a measurable difference, especially on mobile.
What does that mean for you next year?
We're carrying a few things over from the learning-curve moments of 2026. First: we quote projects more openly and name the revision rounds. Second: the industry-specific packages stay and keep getting sharper, because they simply fit better than a one-size-fits-all approach. Third: we've grown more cautious with tools and test them on a real project before turning them loose on everyone.
If you're thinking about tackling your site in 2027, here's maybe the most honest piece of advice from this year: ask an agency what went wrong for them once. Anyone who has no answer to that has either done very little or isn't telling you everything. We'll keep building websites for businesses in the western Munich area, and looking after them afterwards too. And we'll tell you upfront what's realistic and what isn't.
Curious where your current site stands? We'll take a look in a free 30-minute check. Honest, no sales pressure, with concrete points you could act on even without us. Just get in touch through our contact form and we'll find a time.