Positioning is the one brand decision everything else depends on. Your name, your visual identity, your website and your deck all get easier once you can say, in one sentence, who you are for and why they should pick you over the obvious alternative. Most companies skip that sentence and go straight to the logo.
AI is useful here, but not in the way most people use it. Ask ChatGPT to “write our brand positioning” and it hands you a statement that could belong to any company in your category. The better use is to make it do the comparison work, surface the options and argue with you. These are the prompts we use for that, refined from the brand development prompts we published earlier.
What the model needs before you start
Positioning prompts fail without context. Before any of the prompts below, paste in:
- What you sell, in plain words, two paragraphs maximum
- Who buys it (role, company type, sector) and who uses it, if different
- The three alternatives a buyer considers instead of you, including “do nothing” if that is real
- Any proof you have: customer names, numbers, quotes
If you have a brand voice file, load that too (here is how to build one). It keeps the output from drifting into the generic register the moment you stop watching.
Prompt 1: Map the alternatives honestly
Positioning is relative. You cannot say what you are without saying what you are instead of. Start there.
Based on what I have shared, list every alternative a [buyer role] at a [company type] would consider instead of us, including doing nothing, building in-house and hiring a person. For each one, write the sentence the buyer would use to justify choosing it. Then tell me which alternative we lose to most often and why.
The answer to the last question is usually not the competitor you worry about. For most of our clients in life sciences and SaaS it is inertia.
Prompt 2: Generate positions, not statements
A positioning statement is the output. A position is the choice behind it. Make the model show you the choices first.
Give me six different positions we could own, each built on a different axis: speed, cost, specialization in [sector], depth of expertise, ease of working with us, and a sixth you think I have missed. For each: the one-sentence position, the type of buyer it attracts, the type of buyer it repels, and what we would have to stop doing to make it true.
“What we would have to stop doing” is the line that separates a real position from a slogan. If nothing has to change, it is not a position.
Prompt 3: Run the brand comparison
This is the prompt people search for most, and the one that does the most damage when run lazily. The model has not seen your competitors’ websites unless you show it. So show it.
Here is the homepage headline, subheadline and first paragraph from three competitors: [Competitor A: paste]. [Competitor B: paste]. [Competitor C: paste]. Here is ours: [paste]. Build a comparison table with rows for: who they say they serve, the main benefit claimed, the proof offered, the tone, and the words all four of us use. Then tell me which claims are crowded, which are empty and which position is open.
Run it again with each competitor’s about page. Then their pricing or services page. The pattern across three pages is more reliable than any single one, and the “words all four of us use” row is where you find the language to cut.
Prompt 4: Write the statement, then attack it
Only now do you write the sentence. And you write it with the critique built in.
Using the open position from the comparison, write our positioning statement in this form: For [buyer] who [situation], [company] is the [category] that [key benefit], because [proof]. Write three versions. Then, for each, respond as a buyer who has just read it: what do they not believe, what sounds like every other vendor, and what question do they ask next?
The three buyer reactions tell you which version to keep. Usually it is the one that provoked the most specific question, because specific questions mean the buyer took it seriously.
Prompt 5: Test it against the real world
A position that only works in a strategy document is not a position.
Rewrite our chosen positioning statement as: a 6 word homepage headline; the first line of a cold email; the answer to “so what do you do?” at a conference; the first slide of an investor deck. If any of these are hard to write, tell me what in the positioning is making it hard.
If the headline is impossible, the position is too abstract. If the conference answer is impossible, it is too jargon-heavy. The prompt diagnoses the problem by trying to use the output.
A note on “high-value” prompts
People ask us which prompts are worth the time. The honest answer is that the value is in the inputs, not the prompt. Prompt 3 with real competitor copy pasted in will teach you more than any ten prompts run on a blank chat. Spend your effort gathering the material. The prompts above are just the structure for thinking about it.
Where AI stops and strategy starts
The model will give you options. It will not choose, because choosing means knowing things it cannot know: which customers you actually want more of, what your team can deliver, what the founders will say yes to. It also cannot tell you when the honest position is one your company is not ready to claim yet.
That is the conversation we have with science and technology companies before any design work begins. If you are between a product that works and a brand that explains it, start a project. For the prompts that come after positioning (naming, personas, visual identity), the original guide covers them.
Want the short version? The 12-prompt brand library has three of these five plus the voice, naming and website prompts.