Rebuilds a positioning statement from its five inputs, starting with the competitive alternative buyers would really pick, which is usually a spreadsheet or nothing at all.
We have not measured this one. It is published because its content is a procedure with checkable outputs, not because it beat a run without it.
What it knows, concretely. First, positioning as five inputs rather than a sentence, following April Dunford's structure in Obviously Awesome: the competitive alternative, the attributes you have that the alternative does not, the value those attributes produce, the buyers who care disproportionately about that value, and the market frame you are asking to be judged inside. The audit rebuilds all five before it touches a word of copy, and the order is load-bearing, because the alternative determines which attributes are even interesting.
Second, a falsification test with a binary output: could a direct competitor put their own name on this sentence without it becoming false. If yes, what you have written is a description of the category and not a position. Most statements fail this on the first pass, including good ones.
Third, an explicit branch for the answer nobody wants, which is that the real alternative is a spreadsheet, a part-time contractor, or doing nothing at all. A statement aimed at a funded competitor when the buyer's real alternative is inaction is arguing against the wrong opponent, and no amount of copy repair fixes that.
Fourth, a Four Fits sweep, from Brian Balfour's framework, on the grounds that a break in market and product, product and channel, channel and model, or model and market presents identically from outside: growth that stalls for no legible reason. The audit tests all four rather than the one you already suspect.
Who it is not for. If nothing has sold yet, this will correctly refuse to produce a statement and hand you an interview plan, which may not be what you wanted. If you are a solo operator selling to people you already know, the framework is heavier than the decision. And if the honest problem is that the product does not do anything the alternative cannot, no positioning exercise will conceal it, and this one is designed to say so out loud.
Measuring one skill honestly costs about twenty model sessions: five runs with it, five without, on real material, each output graded alone by a session that is not told the other arm exists, against a rubric written by somebody who never saw the skill. We have not spent that on this one yet, so it ships labelled rather than ships silently.
How it would be measured. Tier B on the sentence, with a Tier A spine underneath it. Both arms receive the same invented product brief, six invented buyer-interview transcripts and one competitor page, and the outputs are first scored on the checkable parts, whether a non-vendor alternative is named, whether the segment is defined by an observable property, and whether the statement survives the competitor-signature test, before the sentences themselves go to blind pairwise preference.
The hard part is that positioning has no ground truth on the day you write it. The verdict arrives months later in win rates and sales cycle length, which no test harness reaches.
There is a partial objective spine, and it is the reason this is worth testing properly rather than quickly. Three of the audit's outputs are checkable by a grader with no taste: whether the named competitive alternative is a non-vendor option, whether the segment is defined by a property an outsider could verify rather than an adjective, and whether the statement survives the competitor-signature test against a supplied competitor page. Those can be scored blind from an invented brief and invented interview transcripts.
The honest doubt is the rest of it. A capable assistant asked to sharpen a positioning statement already writes a decent sentence. The question is whether it reaches for the spreadsheet and the do-nothing alternative unprompted, and that needs interview material designed so the obvious answer is wrong.
The rule that decides pass or fail was written down before any run was executed and it does not move afterwards. It is in the method note on the hub, along with the full results table including every skill that was tested and cut.
Stated plainly, because a skill that claims everything is useful for nothing.
name and description in the file's frontmatter, so you can also invoke it by name.A competent marketer with fifteen recent buyer conversations and an afternoon does this better than any skill, and if you have the conversations you should just do it. The framework is not the scarce input. The interviews are.
If you can afford a specialist, a positioning consultant who conducts the interviews personally is the strongest option available, because they hear the words buyers use rather than the words your team reports that buyers use. That difference is where most of the value sits.
At the other end, the model with no skill at all will produce a tidier version of your current sentence. That is genuinely useful when the inputs are already settled and the job is wording. It is the wrong tool when the inputs are wrong, which is the case this is built for.
Finally, your own closed-lost notes are free, already written, and more honest than any framework. Read a quarter of them before you run anything.
--- name: positioning-audit description: Audits positioning by rebuilding it from its five inputs: the competitive alternative a buyer would really pick, the attributes you have that the alternative does not, the value those attributes produce, who cares disproportionately about that value, and the market frame you are asking to be judged inside. Includes the competitor-signature falsification test, an observable-property segment test, the category choice with its budget-line consequence, a four-fit sweep, and a branch that outputs a research plan when the buyer evidence is too thin to support a statement. This skill should be used when writing or reviewing a positioning statement, a homepage headline, a category claim, a segment definition or a competitive comparison page. --- # Positioning audit ## The claim this skill is built on Positioning is treated as a writing problem and it is an input problem. The usual request is to make the sentence sharper. The sentence gets sharper. Nothing else changes, because the sentence was a faithful rendering of a set of inputs that were never examined: who the buyer would otherwise have hired, what you can honestly claim that they cannot, what that difference is worth, to whom, and which shelf you are asking to be judged on. So this audit does not start with the sentence. It reconstructs the five inputs first, in order, and only then writes. The order matters because each input constrains the next. Until you know the alternative, you cannot know which of your attributes are interesting, because an attribute is only interesting relative to something. Until you know which attributes are interesting, you cannot say what value they produce. Until you know the value, you cannot say who cares a lot about it. And the market frame is the last decision, not the first, because it is a claim about which comparison set you want, and you can only choose that once you know what you win on. The five inputs are April Dunford's, from Obviously Awesome. The structure below is a summary in our own words, with the audit procedure and the tests built on top. ## The five inputs | Input | The question it answers | The failure when it is skipped | | --- | --- | --- | | Competitive alternatives | What would this buyer do if you did not exist | You argue against a competitor the buyer never considered | | Differentiated attributes | What do you have that those alternatives do not | You list features the whole category has | | Value | What do those attributes let the buyer do or avoid | You sell capability to someone buying an outcome | | Best-fit customers | Who cares disproportionately about that value | You address everybody and persuade nobody | | Market frame | What comparison set do you want to be judged in | You are priced and scored against the wrong shelf | Run them in that order. If you find yourself editing input four because input five is inconvenient, the market frame is wrong. ## Input one: the competitive alternative, done properly This is the input people get wrong, and they get it wrong in a predictable direction: they name the funded competitor whose fundraising announcement they read. The correct question is not who else sells software like yours. It is what this buyer would actually have done on Monday if you did not exist. In practice the answers cluster into five classes: 1. **Nothing.** The problem is tolerated. In many categories this is the largest single class and it is almost never in the pitch deck. 2. **A spreadsheet, plus a person maintaining it.** The default answer for any workflow product. 3. **A person or an agency.** A contractor, an intern, a seasonal hire, an offshore team. 4. **An adjacent tool used sideways.** A ticketing system used as a CRM, a chat channel used as an approval queue. 5. **A direct competitor.** Real, but usually less common than the first four combined. The consequence is structural. If the alternative is nothing, your competition is inertia, and inertia is not beaten by feature comparisons; it is beaten by naming a cost the buyer is already paying and had not priced. If the alternative is a spreadsheet, your competition is something free, familiar and infinitely flexible, and you win on the failure modes of flexibility rather than on capability. If the alternative is a person, your comparison is a salary line, and the buyer's real question is a political one about headcount rather than a functional one about features. **How you know the answer is real.** It comes from words buyers used, in the past tense, about what they did. "What were you using the day before you bought this" is a better question than any hypothetical. "Which budget line did the money come from" is the single most informative question in the set, because the answer names the category the buyer already believes you are in. If your answers come from your sales team's impressions rather than from buyers, treat them as hypotheses and go to the research branch below. ## Input two and three: attributes and the value they produce An attribute earns its place only if the named alternative cannot claim it. Write the list as a two-column mapping and delete every row that fails: - Attribute, stated as a fact a sceptic could check. - The value it produces, stated as something the buyer does, avoids, or stops paying for. A row with no value is a feature. A row with a value but no attribute is a wish. Both get deleted, and the deletions are usually where the useful argument happens, because most feature lists survive at about a third of their original length. Watch for values that are real but generic. "Saves time" survives no test. "Removes the seasonal contractor line from the budget" names a line item somebody has to defend in a planning meeting, and it can be true or false. ## Input four: the segment test Name the segment by an observable property. Somebody outside your company, with no access to your data, must be able to look at an organisation and say whether it is in the segment or not. | Not a segment | A segment | | --- | --- | | Innovative, forward-thinking teams | Teams running the previous version of the system they are migrating off | | Companies that value quality | Companies with more than 200 SKUs and fewer than four merchandisers | | Growth-minded founders | Founders who have hired a first salesperson but not a sales manager | | Enterprises undergoing digital transformation | Organisations whose security review requires data to stay in their own network | The reason this matters operationally, rather than aesthetically, is that an unobservable segment cannot be targeted, cannot be qualified against, and cannot be counted. If you cannot build a list of a hundred of them by Friday, it is an adjective, not a segment. ## Input five: the market frame, and what a new category costs You have two options and they are priced differently. **Enter an existing category.** Cheap to explain, because the buyer already knows what the thing is, already has a budget line for it, and already has an evaluation process. Expensive to differentiate, because you are on a shelf next to everyone else and the buyer's default comparison is feature-by-feature and then price. **Frame a new category.** Expensive to explain, and the reason most attempts fail is more specific than difficulty: the buyer has no budget line for a category that does not exist. Somebody has to create one, which means an internal argument, a business case, and usually a delay of at least one planning cycle. That cost lands on your sales cycle, not on your marketing plan. **The budget-line test.** Ask, of the segment you named: which existing line does this money come from today, and who owns that line. If the answer is a real line, you are entering an existing category whether or not your website says so. If the answer is that there is no line, you are asking for one to be created, and you should price the delay accordingly before deciding it is what you want. A middle path exists and is usually the right one: enter an existing category the buyer already funds, and differentiate hard inside it with a sub-segment claim. That keeps the budget line and buys you the comparison you want. ## The falsification tests Run all four on the finished sentence. Each has a binary output. **The competitor-signature test.** Take your direct competitor's name and put it at the front of your positioning statement. Does the sentence become false? If it stays true, you have written a description of the category. This is the fastest and most brutal test in the set, and most first drafts fail it. **The negation test.** Negate the claim. Does anyone sane hold the opposite position? "Secure and reliable" negates to "insecure and unreliable", which nobody claims, so it carries no information. "Deploys inside your own network, which means a two-week security review instead of a two-month one" negates to a real position that real vendors hold. **The exclusion test.** Can you say, in one sentence, who this is not for? A position with no excluded group is not a position. If the honest answer is that it is for everybody, input four was skipped. **The evidence test.** For each attribute, name the artefact that proves it: a benchmark, a public document, a customer who will take the call, a demo the buyer can run themselves. An attribute with no proof artefact is a claim, and it belongs in a different section of the site. ## The four-fit sweep Positioning is often blamed for a break somewhere else. From outside, all four of the following look the same: effort goes in, growth does not come out, and nobody can point at the reason. The four fits are Brian Balfour's framework, and the operational point is that you must test all four rather than the one you already suspect. | Fit | The break, in plain terms | The disambiguating question | | --- | --- | --- | | Market and product | The people you chose do not want what you built | Do the people who use it most look like the segment you named | | Product and channel | The product cannot be sold the way you are trying to sell it | Does the channel's format fit the explanation the product needs | | Channel and model | The channel costs more per customer than the model can fund | What is the fully loaded cost to acquire one, against first-year revenue | | Model and market | The market cannot support the revenue per customer the model needs | How many organisations exist in the segment, times realistic price | A worked instance of the last one: a self-serve monthly subscription priced for individual users, sold into a segment whose security review takes eleven weeks and requires a named account manager. The positioning can be perfect and the machine still does not turn, because the model cannot pay for the motion the market demands. ## The decision rule After reconstructing the five inputs, take exactly one of these branches. - **You can name the alternative from at least eight recent buyer conversations, and at least two attributes survive the competitor-signature test.** Write the statement, run the four falsification tests, ship it, and rewrite the homepage, the sales deck and the pricing page frame together rather than separately. - **You can name the alternative, but no attribute survives the signature test.** Stop. This is a product finding wearing a marketing costume. The output of the audit is a note to the product owner naming the two attributes that would survive if they existed, not a nicer sentence. - **Different buyer groups name different alternatives.** You have two positions. Pick the one with the larger observable segment for the homepage, give the other its own page, and never blend them into one sentence, because a blended statement fails the exclusion test by construction. - **You cannot tell.** This means the alternatives you have are your team's impressions rather than buyers' words, or you have fewer than roughly eight recent purchase conversations. The honest output is a research plan and not a statement. Specifically: ten to fifteen interviews with people who made a decision in the last ninety days, quota split roughly evenly between won, lost, and no-decision, because the no-decision group is the only one that can tell you whether the real alternative is inertia. Five questions, all past tense: what were you using the day before, what happened that made you look, who else was in the room, what would you have done if we had not existed, and which budget line paid for it. Write nothing until those are done. ## Worked example A tool that reviews insurance claim documents, sold to claims operations teams. **The statement it arrived with:** "The AI-powered platform that helps insurance teams work smarter and faster." **Input one.** Eleven recent buyer conversations. Seven described a shared spreadsheet plus two contract reviewers hired every autumn. Three described absorbing the backlog and doing nothing. One had evaluated a named competitor. The company's entire comparison page was aimed at that one. Corrected alternative: a seasonal contractor line plus a spreadsheet, with inaction as the second-largest class. **Inputs two and three.** Of nine claimed attributes, six were claimable by the spreadsheet-plus-contractor alternative once it was described honestly, and were deleted. Three survived: it reads the carrier's existing claim form layouts with no template configuration, it cites the specific policy page behind every decision, and it runs inside the customer's own network. Their values: the autumn contractor line disappears, disputes are settled without a full re-read, and procurement clears it in one security review rather than three. **Input four.** Not "innovative insurers". Observable: insurers running between forty and four hundred claims handlers, with a seasonal contractor budget line, whose security review requires regional data residency. **Input five.** The budget-line test found two real lines, contractors and claims software, and no line called decision infrastructure. Verdict: enter the existing claims software category, differentiate on deployment and citation. **Rewritten:** "Claim document review for insurers whose security review will not let data leave their network. It reads your existing claim forms without template configuration and cites the policy page behind every decision, so the autumn contractor budget goes away and disputes are settled without a re-read." **Falsification.** Competitor signature: false when their name is substituted, because they are cloud-only and require template configuration. Negation: the opposite position is held by real vendors. Exclusion: not for insurers under about forty handlers, who should keep the spreadsheet. Evidence: deployment documentation, a citation shown in the demo, one reference customer. **The four-fit sweep, which is where the real finding was.** Market and product fine. Product and channel fine. Channel and model broken: the segment requires field sales with a security review, while pricing was a self-serve per-seat plan that cannot fund a five-month cycle. **Verdict: the statement is approved and it is not the fix.** The positioning was a symptom. The audit's primary finding is a channel and model break, and shipping the new sentence without changing the commercial model would have produced exactly the same stall with better copy. ## Failure modes **Auditing the sentence instead of the inputs.** Ten drafts of a headline, all faithful to the same wrong alternative. Recognisable because every draft feels better and nothing improves downstream. **Naming the funded competitor.** The competitor you think about most is rarely the one your buyer weighed. Symptom: a comparison page that gets traffic from people who were never going to buy either product. **Psychographic segments.** "Ambitious", "quality-focused", "modern". Symptom: nobody can build a target list, so demand generation defaults to broad channels and blames the copy. **Attributes with no value attached.** A specification sheet reformatted as positioning. Symptom: the statement is accurate, checkable, and produces no reaction in a buyer conversation. **Category invention as an escape.** Naming a new category because the honest comparison is unflattering. Symptom: buyers ask which existing thing this replaces, and the answer takes four minutes. **One statement covering two segments.** Produced by refusing to exclude anyone. Symptom: the sentence has two clauses joined by "and", and each clause was written for a different buyer. **Blaming positioning for a fit break elsewhere.** Symptom: positioning has been rewritten twice in eighteen months, each time convincingly, and the growth curve is unchanged. Run the four-fit sweep before the third rewrite. **Interviewing only won deals.** The people who bought agree with you by definition. The no-decision group holds the information about inertia, and they are the hardest to get on a call, which is precisely why they get skipped. ## What this skill does not do - It does not interview anyone. Every input is only as good as the buyer evidence you supply, and with thin evidence it will produce a research plan rather than a confident sentence. - It does not size a market, model a funnel, or tell you whether the segment is large enough to be worth owning. Those need data it cannot reach. - It does not write the website. It produces the five inputs, a tested statement, and an excluded group. Turning that into pages, decks and a pricing frame is a separate job. - It cannot see your closed-lost notes, call recordings or CRM fields unless you paste them in, and those are better evidence than anything it can reason about unaided. - It does not decide whether you can afford category creation. It will tell you what the budget-line cost is and hand the decision back.
These skills all ask your assistant to check things against your actual codebase, your actual schema, your actual design system. Locul keeps that context current on its own, from the files you already have, on your machine. Mac and Windows, free to start.