AI Product Marketing
Positioning statement

When to reposition your product

Reposition your product when one of four things happens: a new product has nothing written down, the periodic refresh comes due (quarterly works well in our method), research contradicts the current positioning, or two assets disagree about the buyer. Treat it as a revision of the dated positioning you have, and raise any contradiction before you rewrite.

Last updated by Omair Dawood.

On this page

The four triggers

These triggers come from our own positioning method, the one the Positioning statement tool runs. They are listed in the method's order. The positioning work they start uses April Dunford's method from her book Obviously Awesome (2019) for the five core choices. Her books page says the book breaks positioning into "five component pieces" (aprildunford.com/books, read 2026-10-14).

  1. Nothing is written down. A new product, company or engagement starts, and no document names the buyer or the alternative the story has to beat.
  2. The refresh is due. The method sets a periodic refresh, and quarterly works well. In our view, a refresh can end with the same statement, because its job is to check the positioning against what buyers do now.
  3. Research contradicts the positioning. A win-loss note, a run of calls or a review site says buyers see you differently. The method's rule here is strict: raise the contradiction with the owner first, then draft the replacement. Never rewrite positioning silently to match a new finding.
  4. Two assets disagree about the buyer. The homepage speaks to one reader and the sales deck to another, and nobody can point at the document that settles it. The disagreement is the trigger, whichever asset turns out to be right.

When the trigger is something else

The method also names three cases that look like repositioning and are not. If the positioning is settled and you need pillars, proof and boilerplate, that is messaging work (see positioning vs messaging). If you need the buyer described in detail, with triggers and disqualifiers, that is persona work. If the homepage reads badly, that is a copy fix on top of finished positioning; a messaging teardown finds it.

How the method runs a repositioning

Once a trigger fires, three rules from the method change how you work.

  • Start from the current positioning and its date. It is a required input when one exists. You propose a revision to a known thing.
  • Escalate the strategic changes. A new audience, a category the brand has not claimed before, or an offer added or dropped is a decision for a person. The method drafts the options and the tradeoff, then hands them back as open questions.
  • Write the reasoning trail. For each of the five choices, record the runner-up and why it lost. The method keeps it so each choice is still defensible six months later.

Worked sample: Intercom's homepage, 2018 and 2026

The Wayback Machine holds a capture of Intercom's homepage from 1 June 2018, with this hero:

"Catch, convert and keep more customers"

"Modern messaging products for sales, marketing and support – all on the first platform made with customers in mind."

Captured 2018-06-01. Read on 2026-10-14. Source: Intercom homepage, Wayback Machine capture

That capture's tab title, labelled separately, is "Customer Messaging Platform | Intercom". The live homepage hero now reads:

"A complete system for AI and human customer service"

"Intercom is the only helpdesk with a natively integrated AI Agent, Fin, meaning every customer conversation improves the next one, enabling perfect experiences that were never possible before."

Read on . Source: Intercom homepage

On a narrow screen the subhead ends at "Fin." We quote the desktop version. We read only the words on these two pages. We cannot see why they changed and do not guess.

What the copy does well. The 2018 headline names three jobs in plain verbs, "Catch, convert and keep", and the subhead names the three teams who do them. The 2026 headline does what the method's category step (step 6) asks: it commits to one category, a "helpdesk" for "customer service". The subhead then makes a claim an outsider could test, a helpdesk with an AI agent built in, and names the agent.

What changed, read against the method. The frame moved from messaging for sales, marketing and support to a helpdesk for customer service. Read against our method, a copy change of that size would be a strategic change, which the method hands to a person as an open question with its tradeoff. Neither version names what buyers use today, so there is still no Unlike. Two phrases fail the method's attribute test (step 3): "the first platform made with customers in mind" and "perfect experiences that were never possible before". Nobody outside the company can confirm either. "the only helpdesk" is checkable in principle, and the method's proof pass would ask for a source anyone can follow before it counts as proven.

The fix, as a filled-in statement. Each slot uses the live page's words or stays a question. The For line uses the headline's own words; it is not a claim about who buys. The public page does not say what Intercom's buyers use instead, so we do not guess.

For      teams running AI and human customer service
who      [the pressure they are under: needs evidence]
Intercom is a helpdesk with a built-in AI agent, Fin
that     lets every customer conversation improve the next one
Unlike   [what these teams use today: needs evidence]
Intercom [the difference, traced to an attribute an outsider can check: needs evidence]

In our view, a team rewriting positioning would keep the 2018 hero beside this draft, with its date, and write one line per choice on what changed.

Common mistakes

  • Rewriting quietly after one bad quarter. The method raises a contradiction with the owner before anything changes. Bring the finding first and the draft second.
  • Starting from a blank page. In our view, a repositioning that ignores the current statement loses the reasoning behind it. Load the old one, with its date, as the first input.
  • Calling a copy problem a positioning problem. If the positioning is sound and the page reads badly, fix the page. The method sends that work to messaging or copy, not to a new statement.
  • Folding a strategy change into the draft. A new buyer or a new category shipped inside a positioning document is the failure the method's second refusal exists to stop. List it as a question.
  • Skipping the refresh because nothing feels wrong. The method sets a periodic refresh, and says quarterly works well. In our view, that catches drift before two assets start to disagree.

Questions people ask

How often should you revisit your positioning?

Our method sets a periodic refresh and says quarterly works well, with the other three triggers able to start one at any time. Some practitioners disagree: Aventi Group writes that "There isn’t a universal schedule for positioning work" and ties revisits to meaningful change (Aventi Group, read 2026-10-14). In our view, a fixed date catches the slow drift that no single event announces.

Is repositioning the same as rebranding?

In our view, no. Repositioning, as we use the word, changes the five choices behind the story: the alternative you beat, the attributes only you have, the value they give, who cares most and the category. A name, logo or colour change can leave all five unchanged. A rebrand that skips the positioning check can put a new look on an old, unchecked story.

Sources

Revise yours with the Positioning statement tool

The tool asks four things: what your product does today, what buyers use instead, how you know they do, and what you say about yourselves now. That last step takes your current positioning or homepage headline, so the tool revises it instead of writing over it; skip it if you have none. It needs proof for at least two alternatives. You get a For, Who, Unlike statement, the five steps of April Dunford's method with reasons, what you should not claim yet, and questions only you can answer. It takes about 3 minutes of typing.

Revise your positioning statement from real buyer evidence