Positioning statement: evidence first, AI second
What is a positioning statement?
A positioning statement is a concise internal summary of the market context a company has chosen: the best-fit customer, the problem or value they care about, the relevant category, the alternatives considered, the differentiated capability, and the reason to believe. It records positioning work; it does not replace the research needed to discover that position.
TL;DR
- -Do not start by filling a template. First establish competitive alternatives, differentiated capabilities, customer value, best-fit customers, market category, and proof.
- -Competitive alternatives are what buyers would use if your product did not exist, including the status quo, manual work, and doing nothing.
- -A differentiator needs three links: a capability competitors lack, value a specific customer cares about, and evidence you can inspect.
- -AI is useful for extracting patterns from supplied interviews, deal notes, product facts, and competitor sources. It must not simulate customers or invent proof.
- -Choose a category whose normal expectations your product can satisfy and whose context makes the differentiated value easier to understand.
- -Write the statement last, derive channel copy from it, and validate comprehension and buying behavior with real prospects.
A polished positioning statement can still be wrong.
The usual failure happens before the writing. A team asks an LLM to fill blanks for audience, category, differentiator, and proof. The model produces a confident sentence from product copy and public competitor pages. None of it shows what buyers actually compare, which difference changes a decision, or whether the proof exists.
The statement belongs at the end. Positioning is the decision; the sentence is its checksum.
What positioning has to resolve
A useful positioning record answers six connected questions:
- What would the customer use if this product did not exist?
- Which capabilities are meaningfully different from those alternatives?
- What value do those differences create?
- Which customers care most about that value?
- Which market category makes the value easy to understand?
- What evidence makes the claim believable?
April Dunford’s public framework uses five components: competitive alternatives, differentiated capabilities, customer value, target customer segmentation, and market category. She explicitly warns that a fill-in-the-blank statement assumes the team already knows the right answers (quickstart guide).
Proof deserves its own column even when a framework treats it as part of value or messaging. A claim without an owner and evidence tends to survive long after it becomes false.
Step 1: collect evidence before the workshop
Use information tied to actual behavior:
- recordings and notes from won, lost, and stalled deals;
- customer interviews about the previous workflow;
- onboarding and support conversations;
- churn and cancellation reasons;
- product telemetry with declared definitions;
- proposals, procurement questions, and security reviews;
- competitor documentation, demos, and current pricing pages;
- product architecture and delivery constraints.
Interview questions should recover the decision, not invite compliments:
What triggered the search?
How did you handle this before?
Which options made the shortlist?
What nearly stopped the purchase?
Which capability changed the decision?
What did you need to verify before believing the claim?
Who felt the cost of the old workflow, and who approved the change?
Separate observed facts, customer reports, and team hypotheses. “Five recent buyers named spreadsheet reconciliation” is evidence. “Mid-market teams hate spreadsheets” is an extrapolation. “The market wants an AI copilot” may be nothing more than internal enthusiasm.
If the company is pre-launch, the evidence column will be thin. That is acceptable. Label the position as a hypothesis and design discovery or offer tests rather than asking AI to impersonate buyers.
Step 2: identify competitive alternatives
Direct competitors are only part of the set. Ask what the customer would do without you:
- keep a spreadsheet or internal tool;
- combine several products;
- outsource the work;
- accept the problem;
- delay the decision;
- buy a named competitor.
Use deal evidence to rank alternatives. Internet research can describe a vendor, but it cannot prove that the vendor appears in your buyers’ shortlists. Dunford calls the imaginary set “phantom competitors”: theoretically adjacent products that customers do not actually consider (positioning and competition).
Create an alternatives table:
| Alternative | Evidence it appears | Why buyers choose it | Where it breaks for best-fit customers |
|---|---|---|---|
| Current manual workflow | 7 interview notes | familiar, no new budget | slow reconciliation at higher volume |
| Existing suite module | 4 lost/stalled deals | already approved by IT | missing required workflow |
| Named specialist | 3 competitive deals | strong category trust | requires an integration buyers cannot support |
The numbers above illustrate the format; replace them with your own counts and sources. Do not turn a web search result into a sales-loss reason.
This table is also an input to competitive intelligence, but positioning uses the alternatives that shape current decisions, not every company in a market map.
Step 3: isolate differentiated capabilities
List product capabilities that can be verified today. Compare each against the relevant alternatives under the same conditions.
A differentiated capability can be:
- a workflow the alternative cannot complete;
- a technical constraint your product satisfies;
- a business model or delivery model;
- proprietary data, distribution, or expertise;
- an integration or operating advantage;
- a combination that is hard to reproduce.
Avoid the word “only” unless the search scope and date are recorded. Competitors change.
Use a matrix:
| Capability | Manual workflow | Suite module | Specialist | Our product | Source |
|---|---|---|---|---|---|
| Works without data export | No | Unknown | No | Yes | docs + test |
| Supports approval policy X | manual check | Yes | Yes | Yes | current docs |
| Produces an audit trail | spreadsheet history | partial | Yes | Yes | demo + schema |
“Unknown” is a valid result. It is better than an invented weakness.
Do not use “simple,” “powerful,” or “AI-first” as capabilities. Describe what the product does and under which constraint. A buyer can inspect “runs inside the customer’s cloud account”; they cannot inspect “enterprise-grade.”
Step 4: translate capability into customer value
A feature matters only through a chain:
differentiated capability
→ changed workflow
→ operational consequence
→ customer outcome
→ evidence
Example:
Capability: joins CRM and product events without a daily CSV export
Workflow change: analyst no longer reconciles two exports before reporting
Operational consequence: report can run on the current data model
Customer outcome: fewer delayed weekly reviews
Evidence: workflow observation + report timestamps
Do not jump from “uses AI” to “grows revenue.” There are several unproved causal steps between them.
Build a value ledger:
| Capability | Value hypothesis | Who cares | Evidence | Confidence |
|---|---|---|---|---|
| Deployment in customer account | avoids copying data to another processor | security owner in regulated team | three security reviews | Medium |
| Automated reconciliation | removes a recurring manual step | analytics lead at high-volume account | observed workflow + pilot logs | Medium |
| Custom dashboard | more flexibility | unknown | no buyer evidence | Low |
The final row should not become a differentiator yet. HBR’s work on B2B value propositions describes the risk as value presumption: treating a favorable difference as valuable without understanding customer priorities (Anderson, Narus & Van Rossum).
When value is economic, keep the assumptions in the value-based pricing evidence ledger rather than dropping an untraceable savings number into copy.
Step 5: find the best-fit customer
Do not ask which broad audience could use the product. Ask which observable characteristics make the differentiated value unusually important.
Useful characteristics include:
- a workflow or trigger already present;
- volume, frequency, or cost that makes the old method painful;
- a technical or regulatory constraint;
- an existing stack or migration event;
- a buyer with authority and budget;
- the ability to adopt the product successfully.
Build the segment from evidence:
Weak:
B2B SaaS companies with 50–500 employees
Stronger:
B2B software teams that reconcile CRM and product-usage data
for a weekly revenue review, lack a dedicated data-engineering team,
and cannot add another external processor
Company size may help locate these teams, but it is not the reason they care.
Compare this with the ICP definition. Positioning identifies customers who care strongly about the unique value; the ICP adds commercial and operational fit such as reachable channels, deal economics, and support burden.
Step 6: choose the market category
A category is a frame of reference. It tells a buyer what the product is, what it should normally do, and which alternatives belong in the comparison.
Evaluate each candidate category against four questions:
| Question | Why it matters |
|---|---|
| Do best-fit buyers already understand it? | Reduces explanation |
| Can the product meet its points of parity? | Establishes legitimacy |
| Does the frame make differentiated value obvious? | Helps the buyer compare |
| Can the company afford to teach a new category? | Exposes adoption cost |
Keller, Sternthal, and Tybout argue that positioning needs both a frame of reference and points of parity before points of difference become persuasive (Harvard Business Review). A product positioned as a CRM inherits basic CRM expectations. Failing them will dominate the buyer’s judgment even if one feature is unique.
Options usually include:
- an existing category where you compete directly;
- a narrower segment or use case within a known category;
- an adjacent category that makes the value clearer;
- a new category that the company must teach.
New is not automatically better. Record which assumptions each category creates, which are true, and which must be corrected.
Step 7: use AI as an evidence clerk
Give AI the source pack and require provenance:
You are organizing positioning evidence. Do not create market facts.
Inputs:
- interview and deal notes with dates and IDs
- product capability matrix
- competitor sources with retrieval dates
- value ledger
- current segment and category hypotheses
Tasks:
1. Extract competitive alternatives named in the sources.
2. Map verified differentiated capabilities to stated customer value.
3. Identify characteristics shared by customers who value them.
4. Compare category options and list the expectations each creates.
5. Produce an evidence table with a source ID for every claim.
6. Separate FACT, CUSTOMER REPORT, TEAM HYPOTHESIS, and UNKNOWN.
7. Flag contradictions, stale sources, and claims without proof.
Do not:
- invent quotations, customers, metrics, market sizes, or proof points;
- treat absence from public docs as proof a competitor lacks a feature;
- infer willingness to pay;
- select a winner when evidence conflicts.
Review the evidence table before asking for prose. If the model cannot cite an input, the claim does not enter the statement.
AI is particularly useful for finding repeated language across interviews and contradictions across departments. It is not a replacement for the interview, the deal, or the product test.
Step 8: write the positioning statement last
Once the components agree, use any concise internal format. For example:
For [best-fit customers with observable characteristics]
who need [high-value outcome in a defined situation],
[product] is a [market category]
that [differentiated value].
Unlike [actual competitive alternatives],
it [verified capability],
supported by [reason to believe].
A statement can be longer than a tagline. Its job is to preserve the decision, not win an award for elegance.
Attach an evidence table:
| Clause | Evidence | Owner | Review trigger |
|---|---|---|---|
| Best-fit customer | deal and usage cohort | PMM | segment mix shifts |
| Alternative | loss notes | Sales | new alternative appears repeatedly |
| Differentiated value | interviews + pilot | Product | competitor closes gap |
| Proof | benchmark or customer result | Evidence owner | source expires |
Do not let AI generate three proof points “for inspiration.” If proof is missing, the output should say proof required.
Step 9: derive messaging, then validate it
Positioning is internal context. Messaging is how that context appears in a channel.
Derive:
- homepage problem, category, and proof;
- sales discovery and narrative;
- competitive battlecards;
- demo path;
- onboarding promise;
- short elevator pitch.
Do not paste the full statement into every channel. A homepage visitor, technical evaluator, finance buyer, and existing customer need different depth.
Validate with real people from the best-fit segment. Avoid “Do you like this?” Ask:
What do you think this product is?
Who is it for?
What would you compare it with?
Which claim needs proof?
What would stop you from taking the next step?
Observe whether prospects classify the product correctly, recall the differentiated value, advance in the buying process, and raise the expected objections. A landing-page test can compare messages, but one click-through rate does not validate the entire position. Sales notes, win/loss patterns, activation, retention, and pricing behavior all add evidence.
When to revisit the position
Review the decision when:
- a different customer characteristic predicts success;
- the status quo or shortlist changes repeatedly;
- a differentiated capability becomes table stakes;
- the product no longer meets category expectations;
- proof no longer supports the claim;
- sales must constantly repair the same misunderstanding.
Keep the old version, sources, date, and reason for change. Positioning should evolve with evidence, not with a weekly prompt.
The final sentence is the easy part. The hard part is choosing a market context that matches what the product can prove and what a specific buyer values. AI can keep that evidence organized. It cannot create the evidence for you.