Thank you. We will be in touch shortly.
Oops! Something went wrong - please email andrew@singlthread.co.uk
The viral first-principles prompt won't fix your positioning. Here's why and what will.
I have been sitting with something for about a week, and it has taken me most of that week to name what was actually bothering me about it.
You may have seen the prompt doing the rounds on LinkedIn that promises to make AI think from first principles. Very Elon. Strip every assumption, ask what is provably true, rebuild. I read it when it first landed, tried it, got a sort of low-level itch I could not name, and moved on. Then it kept coming back to me. Which is usually a sign I have not actually finished the thought.
A week later, I think I have. And it is worth writing down properly, because I suspect this particular conversation is about to get very loud and not especially useful.
Here is the short version, so you do not have to wait for it.
The prompt works. It just cannot do the bit that actually matters when you apply first principles to brand. And the reason matters for anyone trying to build a business that sounds like itself, and not like its category.
Okay. Longer version.
I have sat in something like fifty strategy sessions with founders over the last three years. I am not going to pretend that is a statistically significant sample, but it is enough to notice patterns. And the pattern that shows up every time - honestly, every single time, is that the words a founder is using to describe their business came from somewhere else.
A competitor. A best-in-class deck they saw at a conference. A framework they were taught on a course. A category leader they quietly admire. The last agency they hired and then fired. Very rarely, almost never in the opening session, from the thing that is actually, specifically true about them and their customer.
This is not because founders are unoriginal. It is the opposite. Most of them are bright, fast-moving, and have absorbed enormous amounts of what good looks like in their world. Which is exactly the problem. The more you know about how your category works, the more you reason by analogy. Be the Stripe of onboarding. The Airbnb of legal. The Peloton of continuing education. Fine as a first pass - it is the fastest way to tell someone new what you do, and I use the shortcut myself. But the moment you build on the analogy, the analogy becomes the foundation. And the foundation is borrowed.
I used to think this was a positioning problem. It is not, really. It is a thinking problem sitting underneath the positioning problem. Everyone is reasoning from analogy. Almost no one is reasoning from first principles. Which is why every B2B SaaS brand sounds like every other one, every wellness brand sounds like every other one, every fintech sounds like every other one.
Analogy compounds into sameness. And the thing is, all of this is before AI.
Large language models and I say this as someone who uses them every day now, are very sophisticated pattern-matchers. That is not a criticism. It is the architecture. They are trained on what everyone has already written. Rewarded for stringing it together smoothly. Penalised, quietly but persistently, for going too far off-distribution. You cannot prompt a model into a set of assumptions it does not already have.
You can prompt it to strip assumptions. Which is what the viral prompt does, and it does it pretty well. What the prompt cannot do is tell you which of your assumptions are inherited, and which are load-bearing. That only you can tell. Or, at best, you plus a person sitting opposite you whose job it is to help you sort them.
This is the bit that keeps coming back to me.
First principles thinking, done properly, has two inputs. One is the method of stripping the assumption, ask what is provably true, rebuild. A prompt does that. Any half-decent strategist does that. Honestly, Aristotle did that, and then Descartes, and then Elon, and now a prompt on LinkedIn.
The other input is a person with enough scar tissue to tell the difference between an assumption that is inherited and an assumption that is load-bearing. That bit is not in the prompt. It is not in the AI. It is in the founder, or in whoever is in the room with them doing the work. And it is by far the harder bit. Which is why most positioning skips it.
Before SingleThread, I spent a decade opening restaurants. First in London - Sketch, the National Dining Rooms, Kenza and then across West and Southern Africa. Sixteen openings. A lot of them worked. One of them went very badly wrong.
The pattern I kept seeing, especially in the African years, was that every operator who arrived at a new market turned up with a set of assumptions about their concept. What the margins needed to be. What the wine list should look like. What front-of-house should feel like. What the customer actually wanted. Nearly all of those assumptions were inherited from a city or a market that was not the one they were now standing in.
The operators who survived were the ones who could tell, quickly, which of their inherited assumptions were load-bearing and which were not. The margin model was usually load-bearing. The front-of-house culture was usually not. A brilliant chef in Mayfair does not necessarily know anything useful about a 300-cover brasserie in Accra — and, it turns out, vice versa. What I did not know at the time, but know now, is that sorting inherited assumptions from real ones is the work. Not reinventing everything. Not rejecting everything that came before. Just sorting. Which bits am I carrying because they are the right answer for this situation, and which bits am I carrying because they are what I learned in a slightly different room?
I lost a business to not doing that work well enough once. Genuinely. COVID took most of the blame, and some of it is fair, it took plenty of good businesses down with it. But a real chunk of the reason mine did not survive is that I was carrying inherited assumptions about a market I had never properly stress-tested. You only really learn this one the hard way. Which is, I suspect, why so few people write about it afterwards.
Anyway. Back to brand, which is the reason you are reading this.
It is slower than it sounds. Quieter. Less impressive on a slide.
You sit with a founder or you sit with yourself, if you are the founder, and you go back further than anyone expects. Earlier than the business. Earlier than the product. Earlier than the market research. Back into the operating history of the person building the thing, and the actual, lived, specific problem their customer has. You strip. You ask what is provably true here, in this particular room, not what is provably true about the category. You find the two or three things that are genuinely theirs and nobody else's. You rebuild from there.
If you get it right, the positioning that comes out is not a version of something else. It is a thing that could only have been built by this person, for this customer, in this moment. It will not survive being described as "the X of Y" because it is not that.
If you get it wrong, you get another Airbnb-of-something pitch, and you go onto the pile.
Which is where most of them go, even now.
Use it. Genuinely. The method in it is the same method a decent strategist already uses, and now you have it on your laptop. That is a small win and I do not want to be snobby about it.
What you should not do is expect it to do the sorting. The prompt will happily strip any assumption you feed it, including the ones that are actually right for you. Without someone in the loop who knows their own business - the scars, the customer, the two or three things that are theirs and not borrowed the prompt produces a very convincing-looking pile of rubble. It will feel rigorous. It will read smart. It will not be your positioning.
This is, for what it is worth, the reason I spent a lot of the last year building something called The Foundation. Not as a course. Not as a template. As a sixty-minute video-guided sprint with an adversarial AI that actually pushes back, rejects generic language, and will not let a founder proceed on an assumption they have not defended. It exists because the sorting is the hardest bit, and almost nobody has a tool for it. There are five-figure in-person sessions for that work, and soon there will be an at-home version, and the thing I keep trying to say to anyone who asks is that the AI in the room is the assistant. You are the strategist. Always.
If you have read this far, thank you, and I am a bit surprised, because this one got longer than I meant it to. Here is the question I would ask you to actually sit with.
Where, in your own positioning, are you working from analogy?
Not reflexively. Not defensively. Sit with it for a minute. Which sentence in your pitch came from your competitor's deck? Which came from a founder's podcast you like? Which came from the last agency that pitched you? Which bit of your own story is inherited, and which bit is actually earned?
If the answer is "none" - I gently doubt you. I have not met a founder in fifty sessions for whom the honest answer was really none.
If the answer is "most of it" - well, now you know where the work is.
Either way, the prompt will not find it for you. You can.
See you next month, probably with something shorter.
Cheers,
Marcus
---
P.S. I keep meaning to point people from this newsletter over to the website and I keep getting it slightly wrong. Every time I send one of these the subscriber list goes up and nobody signs up on the site, which I have decided is funny rather than annoying. So instead of pretending I know what I am doing on that front, here is a direct thing: if you want the ten-minute, diagnostic version of what I am describing above, where your positioning is inherited and where it is actually yours, there is a free thing called the Brand Visibility Diagnosis at https://brandvisibility.scoreapp.com/. It is not a sales call. One conversation, no pitch. If it is useful, it is useful. If not, no harm done.