We need to tell you something slightly embarrassing about this website. Most of its readers are not people.
We found out when our analytics went, in the technical parlance, a bit mental. Thousands of pageviews in a week, nearly all of them landing on one guide, nearly all of them with no referrer, each visitor reading exactly one page and leaving. That is not how humans read a website. Humans wander. They click a second thing. They get distracted by the shop. This traffic arrived, read, and left like it was being paid by the page.
It was bots. Specifically, it was AI companies. OpenAI, Anthropic, Google, Perplexity and friends, crawling our guides on how consulting projects fail, some of them to train future models, some of them to answer a live question a real person had just typed into a chat box. We know this because we started logging them, in public, on the Bot Ledger. It updates as they visit. It is, at the time of writing, the most honest analytics dashboard we have.
And buried in that slightly absurd discovery is the most practical business development lesson we can offer a consultant in 2026.
The referral you never see
Shortly after the crawlers arrived, something else showed up in the logs: human visitors referred from chatgpt.com and perplexity.ai. Follow the chain of events, because it's the whole story. Somebody, somewhere, asked an AI assistant a question about AI consulting. The assistant went and read our page. It cited us in its answer. The human clicked through.
That's a referral. Nobody networked for it. Nobody posted about their exciting journey on LinkedIn for it. A machine read something specific and useful, judged it worth citing, and sent a prospective reader to the source.
Now put your client hat on. A mid market CEO who's decided to do something about AI in 2026 does not start by asking a consultant. They start by asking the assistant that's already open on their laptop. What is an AI readiness assessment. What should an AI consultant cost. What questions should I ask before hiring one. The assistant assembles an answer from whatever sources it considers citable, and the consultants named or linked in that answer have effectively been shortlisted before any human conversation has taken place.
Your next client is doing discovery on you before you know they exist. The question is whether the machines doing that discovery on their behalf have anything of yours worth reading.
What the machines actually cite
Here's what our own ledger and referral data suggest gets read and cited, and it's almost insultingly simple: specific, factual, unhedged content. Our most crawled pages are the ones with numbers in them. Day rates by specialism. Twelve named red flags. A ninety day plan with actual weeks in it.
What doesn't get cited is the stuff most consultants publish. Thought leadership about how AI is transforming everything. Posts that could have been written by anyone, about anything, for no one. An assistant answering a concrete question has no use for vibes. It wants a figure, a framework, a checklist, a claim specific enough to be worth repeating with a source attached.
This should feel familiar, because it's the same rule that has always separated consultants who get referred from consultants who get scrolled past. Specificity reads as expertise. The only new part is that the referrer is now partly mechanical, tireless, and reads everything.
The practical bit
So what do you actually do, as one consultant with a modest website and no appetite for becoming a content farm?
Publish the thing only you know. One page with real numbers or a real framework from your actual practice beats fifty posts of commentary. If you know what discovery questions kill bad AI projects, write them down, all of them, with the reasoning. That page can work for years.
Make it findable by machines as well as people. This is unglamorous plumbing: clean page titles that match the question being asked, a llms.txt file telling assistants what your site contains, pages that load as readable text rather than a JavaScript souffle. We publish our crawler data openly and it gets read precisely because it's easy to read.
Say who you're for. Assistants answer questions like "AI governance consultant for healthcare" far more usefully than "AI consultant", and so does every human referrer you have. Niche positioning was already good advice. The machines have just made it compulsory.
And then check whether it's working. Look at your referrer logs for chatgpt.com, perplexity.ai and friends. Ask the assistants your clients use the question you'd want to be the answer to, and see who they cite instead of you. That last exercise is uncomfortable and worth doing quarterly.
The part where we admit the joke
Yes, we see the loop. We published guides about AI consulting. AI companies crawled them. We started logging the crawlers, then wrote this article about the logging, which the same crawlers will shortly read, and if one of them cites it to a curious consultant, the ledger will log the resulting visit too. Somewhere in that cycle there's a lesson about the 2026 economy that we're too close to see clearly.
But the practical point survives the irony. Being genuinely, specifically useful in public is now distribution. The machines have simply made the returns on it compound faster. If your positioning needs the overhaul before any of this can work, the LinkedIn Positioning Kit is the fastest fix we sell, and the free Career Pivot Playbook covers building the evidence base itself. The machines will be along to read yours shortly.