AI Prompts for Positioning and Messaging
Nine Copyable Prompts, in the Order Positioning Work Actually Happens
Short answer
The positioning prompts that work feed the model real material and ask it to sort, compare or stress-test, not to invent. Paste competitor claims, customer reviews and your own evidence, then work in order: map claims, find alternatives, link attributes to value, draft statements, and test whether a competitor could say the same.
1. Why do AI positioning drafts come out generic?
Because the model has nothing specific to work from, so it writes the average of every positioning statement in its training data. Generic input produces category language: "powerful", "easy to use", "built for teams". Specific input is the only cure.
Positioning is a choice about who you are for, what you are compared with, and why you are the better option for that buyer. None of those are facts a model can know about your company. When the prompt does not supply them, the model supplies the most common answer, and the most common answer is by definition the one that does not differentiate you.
The fix is to split the work. Positioning methods such as April Dunford's break positioning into component pieces, starting from the competitive alternatives a buyer would otherwise choose [s1]. Each piece can be researched with real material and processed with a focused prompt. The model is very good at sorting and comparing what you give it, and poor at deciding what is true about your market.
Collect this material before using any prompt below:
- Competitor homepage, pricing and product page copy, pasted as text, for three to five competitors.
- Thirty to a hundred customer reviews, yours and theirs, with star ratings.
- Notes or transcripts from a handful of sales calls and customer interviews.
- A plain list of what your product does that the others do not, with how you know.
2. Which prompts map what competitors already claim?
Two: a claim inventory across competitor pages, and a table of which buyer belief each competitor is arguing against. Together they show the claims that are taken, crowded or open.
Below is marketing copy from several competitors, each marked with the company name. List every distinct claim they make about themselves. Output a table with columns: Claim (in plain words), Which companies make it, Exact wording used by each (quoted), Proof offered (if any). Then list the claims made by only one company. Do not add claims that are not in the text. COPY: [PASTE COMPETITOR COPY, LABELED BY COMPANY]
Use it first. The claims made by every competitor are table stakes: you need them, but they will not differentiate you. The claims made by one company are either their real positioning or an opening. Check the table by opening two or three quotes at random against the source.
For each company below, read their homepage copy and identify the belief a buyer must hold for this company to win, and the belief they seem to be arguing against. Output one row per company: Company, Belief they need, Belief they argue against, The line of copy that shows it (quoted). If the copy does not make it clear, say "unclear" rather than guessing. COPY: [PASTE HOMEPAGE COPY, LABELED BY COMPANY]
This one turns a list of features into a map of arguments. If Veltrix is arguing "you do not need a specialist tool", Sondera is arguing "the cost of leaving your current system is higher than you think" and Norvane is arguing "cheaper is good enough", you can see which argument your best customers actually reject.
3. Which prompts find alternatives and real value?
Mine customer language for what buyers used before and why they switched, then connect each of your unique attributes to the value it creates and the proof you have. The model groups; you decide what is true.
Below are customer reviews and interview notes. Extract every passage where the customer mentions what they used or did before, or what they would do without this product, including spreadsheets, hiring someone or doing nothing. Group the passages by alternative. For each group give the number of passages, one verbatim quote, and the reason for switching that customers give most often. List separately any passage you could not group. MATERIAL: [PASTE REVIEWS AND NOTES]
Competitive alternatives are often not competitors at all. A group of customers who moved off "a shared spreadsheet and a lot of email" tells you what your real comparison is for that segment, and it rarely shows up in a competitor analysis.
Here is a list of things our product does that the alternatives listed do not, with the evidence for each. For each attribute, write: the practical benefit for the customer, the value that benefit creates (time, money, risk or status), which customer type cares most, and the strongest proof from the evidence provided. If an attribute has no convincing value or no proof in what I gave you, say so. ATTRIBUTES AND EVIDENCE: [PASTE LIST] ALTERNATIVES: [PASTE THE GROUPS FROM PROMPT 3]
The most useful output of this prompt is the list of attributes with no value or no proof. Those are the features you are proud of that buyers do not care about yet, and they should not lead your positioning.
4. Which prompts draft the positioning statement itself?
One prompt that uses the classic For, who, unlike structure, fed with the outputs of the earlier prompts, and asked for three deliberately different variants. Pick one, then rewrite it yourself.
Using only the material below, write three different positioning statements for [YOUR COMPANY]. Use this structure: For [target customer] who [need or problem], [product] is a [category] that [key benefit]. Unlike [main alternative], we [key difference]. Make the three variants genuinely different: one aimed at the segment that values [ATTRIBUTE A] most, one at the segment that values [ATTRIBUTE B] most, and one using a different category label. After each, list which piece of material supports each part. Avoid these words: powerful, seamless, all-in-one, easy, innovative. MATERIAL: [PASTE OUTPUTS OF PROMPTS 1 TO 4]
The structure is the one popularized by Geoffrey Moore in Crossing the Chasm [s2]. Asking for three variants aimed at different segments is deliberate: positioning is a choice, and seeing the options side by side makes the trade-off visible. The ban list stops the model reaching for category filler.
5. How do I use AI to stress-test positioning?
Ask the model to argue as a competitor and as a skeptical buyer. The swap test checks whether a rival could publish your statement unchanged; the skeptic test finds the claims a buyer would not believe without proof.
Here is our positioning statement and the homepage copy of three competitors. For each competitor, say whether they could publish our statement with their name in it and have it be true, based only on their copy. Answer yes, partly or no, and quote the lines that show it. Then identify the words in our statement that do the differentiating work, and the words any of them could also claim. OUR STATEMENT: [PASTE] COMPETITOR COPY: [PASTE, LABELED BY COMPANY]
Act as a skeptical buyer from [TARGET SEGMENT] who currently uses [MAIN ALTERNATIVE]. Read our positioning statement and list the five questions or objections you would raise before believing it, in the words a buyer would use. For each, say what evidence would convince you. Do not be polite. STATEMENT: [PASTE]
Run the swap test on every draft. A statement a competitor could sign is not positioning. The skeptic prompt produces objections you should compare against real ones from sales calls: where the model and your buyers agree, you have found a claim that needs proof on the page.
6. Which prompts turn positioning into messaging?
A message house prompt that builds pillars and proof from the chosen statement, and a customer language prompt that finds the phrases buyers use for your value, so messaging sounds like them rather than like you.
Using this positioning statement and the supporting material, draft a message house. Output: one umbrella message of under 15 words; three pillars, each with a one-line message and two proof points taken only from the material; and for each pillar, the objection it answers. Mark any proof point that is weak or missing as NEEDS PROOF. STATEMENT: [PASTE] MATERIAL: [PASTE ATTRIBUTES, VALUE AND PROOF FROM PROMPT 4]
Here are our three message pillars and a set of customer reviews and interview notes. For each pillar, find the phrases customers use when they describe that value in their own words. Give five verbatim phrases per pillar with the source line, and point out where customer wording differs from ours. Do not paraphrase the phrases. PILLARS: [PASTE] MATERIAL: [PASTE REVIEWS AND NOTES]
The second prompt is where messaging gets its voice. Customers rarely say "streamlined collaboration"; they say "clients stopped emailing me for updates". That phrase belongs in your headline, and only real customer material will surface it.
7. Which positioning prompts do not work?
Prompts that ask the model to decide rather than process: choose our target market, tell us what makes us unique, or describe what competitors are doing now. The model has no access to the answer, so it produces a plausible one.
| Prompt that fails | Why it fails | Ask instead |
|---|---|---|
| "Write our positioning statement" | No material, so it writes the category average | Prompts 1 to 5, in order, with your material |
| "What makes us unique?" | It cannot know your product or market | Prompt 4, with your list and evidence |
| "Who should our target customer be?" | That is a business decision, not a text task | Prompt 3 to see who switches to you and why |
| "What is Veltrix's positioning now?" | Relies on memory that may be out of date | Prompt 2, with their current homepage pasted |
| "Make this more compelling" | Adds adjectives, removes specifics | Prompt 7, then fix what the skeptic finds |
Key Takeaways
Generic input, generic positioning
Without your material, the model writes the category average. Paste competitor copy, reviews and your evidence.
Work in the order positioning happens
Map claims, find alternatives, connect attributes to value, draft statements, stress-test, then build messaging.
Use the model to sort and compare, not to decide
Grouping, comparing and testing are reliable. Choosing your market or your differentiator is your job.
Ask for variants aimed at different segments
Three different statements side by side make the positioning choice visible instead of hiding it.
Run the swap test on every draft
If a competitor could publish your statement with their name in it, it is not positioning yet.
Take messaging words from customers
Mine reviews and interviews for verbatim phrases. They beat anything the model writes from scratch.
Frequently Asked Questions
Sources
- Books: Obviously Awesome and Sales Pitch April Dunford, September 2026.The author's own page for Obviously Awesome, which breaks positioning into five component pieces.
- Crossing the Chasm Wikipedia, September 2026.Reference for Geoffrey Moore's 1991 book, the usual source of the "For, who, unlike" positioning statement structure.
Explore More
Related analyses, benchmarks, and industry insights
Related Guides
Glossary Terms
Turn reading into a reaction
Oppira watches your competitors, keeps your playbook current, and drafts the response. See your market clearly by tomorrow.