Disclosure: I own both Paulshof Guesthouse and OwnerPilot. Everything below is what I learned rebuilding my own site.
In 2020 I opened a guesthouse to save my house. Covid had done what Covid did, and the most honest path forward was to turn the place I lived in into a place other people could stay. I called it Paulshof Guesthouse. I put it on Airbnb. I learned, very fast, every pain point a small hospitality operator knows: the dependency on OTA platforms, the commission that disappears off every booking, the way the algorithm decides whether you exist this week or not, the silence when it decides you don't.
I started a small app to nudge guests towards booking with me directly instead of going through a platform. That worked well enough that I added a loyalty layer to it. Icon Group saw it and asked the question that bent the next five years of my life: what about websites? I said yes before I knew what I was saying yes to, and that turned into GHPro — guesthouse websites with the booking and loyalty piece built in instead of bolted on.
A while later I was sitting at Gilroys in Rivonia, watching a bar full of regulars who clearly wanted to be loyal to a place that had no way to recognise them, and I built PubPro for the same reason — restaurants and pubs deserved the same infrastructure guesthouses got. After that came the Command Centre, then a Demand Manager, then an intelligence layer. None of them were a roadmap. Each one was a thing I had to build because the thing in front of it didn't work without it.
Then a few clients who weren't pubs or guesthouses started asking if I'd do their sites too — a wellness business, a firm, a few professional services operators — and that became OwnerPilot, the thing you're reading this on now.
I'm telling you all of that because what comes next only makes sense if you know where I'm standing when I say it. I'm not a marketing agency. I run a guesthouse and I build the tools the guesthouse needed. The agency exists because the tools turned out to be useful to other owners. Every opinion below is held by a person who has to make rent on the strength of whether the website actually works.
What I thought I knew, and what stopped being true
Paulshof did fine on the old internet. At its peak the site was doing over three thousand visitors a month. I'd done the Google digital marketing courses — keyword research, on-page SEO, the whole nine — and the playbook worked. You picked your terms, you put them in the right places, you waited, you ranked.
Then sometime in the last eighteen months it just… stopped working. Not dramatically. Not all at once. Just a slow leak. The keyword-era playbook turned into a polite suggestion that Google was mostly ignoring. The algorithms moved to something that read intent and context and topical authority and signals I couldn't name. By the time I finished a course explaining what was new, what was new had changed.
And then the AI engines arrived — ChatGPT and Perplexity and Gemini and the AI summaries inside Google itself — and suddenly there was a second search layer running on top of the one I'd been trying to catch up with. Guests were asking ChatGPT for B&Bs near Sandton with parking. They were asking Perplexity for guesthouses with good breakfast. The traditional search advice didn't cover any of it. The people writing the new advice mostly didn't know either.
The honest position to be in, as the owner of a small hospitality business in 2026, is this: the rules of being findable change weekly, the people who claim to know them often don't, and the courses you take to catch up are out of date by the time you finish them. That's not a complaint. It's just the weather.
The diagnosis I didn't want to do on my own site
I audited Paulshof properly this year and the result was worse than I'd guessed. The site was technically alive but functionally invisible: zero pages indexed by Google. The meta description — the thing that's supposed to tell search engines what the business is — literally said "Professional property management and guest engagement platform." It described the SaaS that powered the site, not the guesthouse. Any AI engine reading the page was being told, in plain English, that Paulshof Guesthouse is a software product. No JSON-LD structured data anywhere. No FAQ schema. No way for a crawler to know what kind of business it was looking at.
The site rendered fine in a browser. To a human, it looked like a guesthouse website. To Google, ChatGPT, Perplexity, and Bing's AI layer, it looked like nothing at all.
That was a humbling thing to find on your own property. It was also a useful one. Because if my site — built by a person whose job is building these — could drift into invisibility without me noticing, that's true of almost every small business site I've ever met.
Why a website can't be a delivered thing anymore
This is the bit I want to sit on, because I think it's the single most important shift I've internalised in the last two years.
The traditional model of a small-business website is: you commission it, an agency builds it, they hand it over, you pay them, it goes live, and from then on it's your asset — until you decide to rebuild it three or four years later and the cycle repeats. The site is treated like a piece of fixed infrastructure. A building. Once built, done.
That model is broken. Not because anyone's doing a bad job of building sites — most agencies build perfectly competent sites — but because the environment the site lives in now moves faster than the rebuild cycle. The algorithms shift quarterly. AI engines change how they crawl monthly. Schema standards evolve. Cloudflare changes its defaults. Browser behaviour, mobile ranking signals, voice search, AI answer engines, the rise of zero-click search — all of it shifts under a site that's not being watched.
A website built six months ago and not touched since is silently losing customers right now. Not because it was built badly. Because the world around it moved and nobody told the site.
What a small business actually needs isn't a website. It's adaptive infrastructure that responds to a moving target so the owner doesn't have to chase it themselves. Three things, every week: someone reading the analytics with enough sensitivity to spot a drift before it becomes a trend, someone watching the crawler and AI-engine behaviour to know when discovery is breaking, and an interface flexible enough to change in response without a rebuild.
The owner shouldn't have to do any of that. The owner should run the business. The owner should just see results.
That is the actual thing I'm trying to build OwnerPilot to be.
Rebuilding Paulshof in public
When I rebuilt the Paulshof site this year I tried to do it the way I think every small site should be done now — not as a launch event, but as the first turn of a wheel that keeps turning. Here's what that looked like in practice.
First: the basics nobody's selling because they're not glamorous. Tenant-specific head metadata, so the page knows what business it represents. A proper LodgingBusiness JSON-LD block with the address, the rooms, the price band, the geo coordinates. An Organization entity so search engines know we exist. An FAQ schema, because a third of the questions guests ask are the same six questions and structured answers are how voice search and AI assistants pick up the answer. A unique meta description per page rather than the global default that was bleeding onto everything. None of this is revolutionary. It's just the new baseline. Most small sites don't have any of it.
Then: the layer most sites stop before getting to. Making sure the page that an AI crawler sees — before any JavaScript runs — actually contains the content and the structured data. This took longer than I expected, and went through one full diagnose-then-fix cycle to catch a duplicate description tag that was beating my per-page overrides in the served HTML. The crawler was seeing a generic platform description on every page even after I'd written unique ones for each. Fixing that took observation, not just building. I would not have caught it if I hadn't gone looking.
Then: turning on Cloudflare's AI Crawl Control, which is one of the genuinely useful things to come out of the AI-search-bot era. It shows you, by name, which AI crawlers are actually hitting your site and what they're doing. In the first 24 hours of being properly visible, Paulshof was crawled by Google's AI crawler eight times, by ChatGPT-User twice, by Anthropic's ClaudeBot twice, by Microsoft's BingBot twice. The sitemap was the single most-crawled path. That's discovery happening at the AI-engine layer — the layer that traditional SEO tools can't see and that 90% of small-business site owners don't know exists.
Then: setting up the boring but essential measurement infrastructure so that next week, and the week after, and three months from now, I can tell whether what I just shipped is still working. GA4 wired in. Search Console verified. Bing Webmaster Tools imported. The Cloudflare AI dashboard checked. A weekly read scheduled in my own calendar.
Then: the off-domain work — Google Business Profile rebuilt as a service-area business rather than a physical-location one, listings updated, the website URL corrected wherever it appeared on Booking.com and the rest. The site exists in an ecosystem; the ecosystem has to point at it consistently.
The result, today, is a Paulshof Guesthouse website that's findable by Google, visible to AI crawlers, sharing correctly on WhatsApp and Slack previews, tracked by analytics, supported by a Google Business Profile that doesn't conflict with anything else at the same address. None of that is finished. All of it needs to be watched.
What I'm building toward, and what isn't there yet
I should be honest about this because the whole point of the piece is that founders who sell you fairy tales aren't worth listening to.
The vision is that an owner — running a guesthouse, or a pub, or a restaurant, or a wellness business — can have a site that adapts on their behalf. The analytical sensitivity is the platform's, not the owner's. The weekly reads are automatic. The drift detection is built in. When something shifts in how AI engines are crawling, or how Google's ranking signals are weighted, or how the local pack is structured — the platform notices and the site adjusts. The owner sees results. The owner does not need to do a course.
That's the goal. Where I actually am right now is partway there. The observability layer exists in pieces — Cloudflare AI Crawl Control, GA4, Search Console, the tenant-aware SEO infrastructure. The adaptive layer exists for the technical foundations — head metadata, schema, per-page descriptions, footer structure, internal linking. The analytical-sensitivity layer is forming but isn't packaged yet. The weekly-read automation isn't built. The "owner just sees results" promise is honest as a direction but not yet honest as a delivered product.
I'm telling you this because the cost of overclaiming is much higher than the cost of being plain. Anyone who tells you their platform already does all of that today either hasn't built it or doesn't understand the size of the problem.
What I can tell you with confidence is the cadence is right. Building once and walking away doesn't work anymore. Observing weekly, adjusting often, treating the site as a living product rather than a delivered asset — that's the only model that survives the next few years. Whether the tool you use to do that is mine or someone else's matters less than whether you're doing it at all.
The invitation, rather than the pitch
If you own a small business and your website was built more than a year ago and nobody's been reading it weekly since, it's almost certainly leaking customers in ways you can't see. The fix isn't necessarily a rebuild. It might be as small as the things I described doing to Paulshof — head metadata, schema, AI-crawler observability, a Google Business Profile that's actually correct.
If you want to know what your own site looks like to the AI engines that are deciding whether to recommend you, that's a conversation I'm happy to have. Not a quote. Just an honest read of where you stand. I'll tell you what I see.
The bigger thing I'm building toward — adaptive infrastructure, weekly reads on your behalf, results without the courses — isn't fully shipped yet. I'd rather you know that than be sold it. But the direction is the direction, and it's where every small-business website needs to go.
I'll keep rebuilding Paulshof in public as I work it out. You're welcome to watch.
— Billy