How to run a positioning sprint

Guide · Frameworks · 4 min read · last verified 2026-07-27

A positioning sprint is a one-week, evidence-first process for deciding how a company describes itself: gather what the market actually asks and says, draft competing positions, test them against outside reality, choose one, and ship it to a live page before the week ends. The unusual clause in that definition is "outside reality." Most repositioning efforts run on inside material — stakeholder interviews, sticky notes, a facilitated workshop — and end with words the room likes. A sprint in the sense described here treats the room's preference as the final input, not the first.

The one-week constraint is not bravado. Positioning projects tend to sprawl precisely because they lack an external referee: with nothing outside the room to settle arguments, every draft is one more meeting away from consensus, indefinitely. Importing the referee — recorded buyer language, observed answers, competitor text — is what makes a hard deadline survivable, because disputes get settled by looking rather than by seniority or stamina.

Before the sprint: assemble the outside evidence

The sprint consumes evidence, so the evidence has to exist first. Two inputs are non-negotiable.

First, the market's actual buyer questions: what people ask in sales calls, community threads, support tickets, and search. Not the questions you wish they asked — the recorded ones, verbatim, because the vocabulary is the data.

Second, a snapshot of how AI assistants currently describe your category and your company. Ask what a real buyer would ask and save the answers with the date attached, because these are observations of a moving system, not permanent facts. This snapshot doubles as your before picture — the baseline any later re-measure will be compared against.

A third pile helps if time allows: your competitors' public language, collected without commentary. The rule that governs all of it: no drafting until the evidence is assembled. Drafting first and retrofitting evidence afterward is the workshop pattern with a sprint label stuck on it.

Days one and two: read before you write

The sprint team — small, with one decision-maker named in advance — spends the first two days reading, and only reading. The discipline is descriptive: establish what the outside world currently believes before proposing what it should believe.

Things worth logging as observations: words buyers use that appear nowhere on your site; the problem language that recurs across unrelated sources; what the category answers include, omit, and treat as the default; where the assistants' description of you is wrong, stale, or generic. Write observations, not conclusions — "buyers say audit trail, our site says compliance layer" is an observation; "we should rename the feature" is a conclusion that can wait for day three.

Day three: draft three positions

Three is deliberate: a single draft tends to produce attachment, and a large spread tends to produce blur. Each draft position must specify, on a single page, the buyer it centers, the problem in that buyer's own recorded words, the alternative it displaces, the one-sentence claim, and the evidence available to back the claim. The evidence line is the sprint's conscience — a position that cannot cite support gets revised or discarded here, cheaply, rather than on the homepage later.

Force the three to be genuinely different. Three shades of the current position is one position and a mirror; the exercise only informs the decision if at least one draft feels uncomfortable.

Day four: test for legibility

The test is legibility, not applause. A position works only if a stranger — or a machine summarizing you — can reproduce it accurately after one exposure. Two checks fit in a day.

Cold-read the drafts with people who match the buyer profile and have them explain back, in their own words, what the company does and for whom. Where the explanation diverges from the draft, the draft loses; translation loss found today is churned pipeline avoided later.

Then run each draft through a message test against AI answers: probe how an assistant describes a company claiming what your draft claims, and whether your existing public material would plausibly earn that description. A position the public corpus cannot yet support is not disqualified — but it is a roadmap item, and knowing that on day four changes what shipping means.

Day five: decide and ship one page

Decide in the morning, with the evidence on the table and the named decision-maker breaking any tie. Then ship the winning position to exactly one page — usually the homepage or the category page — by the end of the day. One page, not a full rebrand: the page is a commitment device, small enough to ship in an afternoon and public enough that the decision is real. The rollout across decks, docs, and campaigns can follow at a saner pace, with the shipped page as its reference.

After the sprint: schedule the re-measure

A sprint ends on day five; the positioning work ends when the outside description changes. Freeze the question set from your pre-sprint snapshot and re-run it on a calendar cadence, watching for your new vocabulary to surface in answers and in inbound conversations. Movement tends to be slow and uneven — corpus-level description tends to shift over months rather than days — so the absence of instant change is expected, not damning.

What the cadence buys you is an honest verdict: if nothing has moved after several cycles, the position is likely illegible or under-evidenced in public, and the sprint format makes a rerun cheap. Teams that use Magrios typically wire this step to a locked benchmark so the before-and-after is measured rather than remembered; however you run it, the re-measure is what separates repositioning from redecorating.