We ran our AI SEO agent on our own site before charging anyone

Before we let rankr near a paying customer, we pointed it at a site we own: lullaflex, a mattress brand we built from nothing. Dogfooding, basically. The logic was simple. If our own AI SEO agent couldn't move a brand-new site we controlled end to end, we had no business charging anyone else for it. This is the honest log of what happened: what worked, what didn't, and the one decision that changed how we think about the whole thing.

Why a mattress site, of all things

We didn't want a friendly demo. We wanted a boring, crowded, unforgiving niche, the kind most Singapore SMEs actually operate in. Mattresses fit perfectly: established brands, real ad budgets, and nobody waiting for a newcomer. A brand-new domain with zero authority, no shortcuts, no home-field advantage. If the playbook works there, it means something. If it only works on easy mode, it's a toy.

What the agent actually did

rankr ran the loop it's built for: find the keywords and gaps worth targeting, draft the content, publish to the site, with us approving each step. On the technical side we got obsessive, because that part is fixable and most sites get it wrong. Clean title tags, one clear H1 per page, LocalBusiness schema, a real XML sitemap, Open Graph tags, the plumbing that decides whether Google and AI engines can even read you. We got it clean enough that when we later built a free technical SEO checker, lullaflex scores about as high as the tool goes.

The honest part: perfect setup, still page five

Here's what all that technical polish did not do. It did not make us rank.

The site is new. It sits around page five for the terms we want, a few thousand impressions a month, and almost no clicks yet. We're not going to dress that up. And that gap, flawless on-page setup but still not ranking, is the entire lesson.

Technical SEO is table stakes. It makes you eligible. It does not make you authoritative, and authority is what ranks you. Authority comes from other sites linking to you and mentioning you, and a brand-new domain has none of that. There is no setting in any tool that fixes it overnight. It's the slow, unglamorous part, and pretending otherwise is how SEO gets a bad name.

The decision that changed our approach

A builder friend made a point that stuck: stop chasing rankings first, build something people actually use, and the traffic and links come to you.

So we did. lullaflex now has a free bedroom layout planner, you plan your room before you spend on furniture. It's genuinely useful, which is the whole point. A useful tool gets shared, bookmarked, and linked to in a way a blog post about mattresses never will. That's the difference between asking for attention and earning it, and for a new domain with no authority, earning it is the only path that compounds.

What we'd tell any Singapore SME

Four things we'd have paid to know a month ago:

• Get the technical basics right, but treat them as the floor, not the ceiling. They stop you losing. They don't make you win.
• A new site's real bottleneck is authority, and that's earned off your own site, in other people's.
• A free, genuinely useful tool earns links better than any amount of content.
• Google rankings are slow, often months. Getting cited by AI answers can be faster, once you're indexed in Bing and answering real questions clearly.

We're running this in public

We're not going to show you a screenshot of a dashboard with a hockey-stick graph. lullaflex is early and still climbing, and we're running the exact playbook we'd run for a client, out in the open. You can watch it work, or watch it not work. That's the point of doing it on our own site first.

Frequently asked questions

Why run your own SEO agent on your own site?

Because dogfooding is the honest test. If the tool can't move a brand-new site we own and control end to end, it isn't ready for a paying client. It also gives us a live case we can point to instead of promising results we haven't earned.

Does good technical SEO make you rank?

No. It makes you eligible. Clean titles, schema, a sitemap and fast pages get you into the race, but rankings are decided by authority, the links and mentions you earn from other sites. Technical hygiene is necessary and nowhere near sufficient.

What's the fastest win for a brand-new site?

Get indexed in Bing (ChatGPT's search runs on it), claim your Google Business Profile, and build something genuinely useful that earns links on its own. Chasing competitive keywords from a zero-authority domain is the slowest path there is.