Why the generic STP does not transfer to launch
The standard STP exercise starts from a business that already sells something. Segments are drawn from purchase records, churn and margin by client type, and targeting compares groups whose economics are already visible. A launch has none of that. There is no history to split, no retention curve to rank segments by, and no customer stories to cite as reasons to believe. If a team runs the usual sequence anyway, it fills the gaps with optimism: the segment that sounds largest becomes the target, and the positioning claims outcomes nobody has yet delivered.
The launch version treats each step as a bet with stated odds. It also adds two things the framework page does not discuss: a list of segments the launch will refuse during its opening window, and a scheduled moment when the bets are checked against what early buyers actually did. Without those, a launch STP turns into a permanent strategy built on a guess.
Decisions that change for launch
| STP step | In the generic sequence | For a launch | Decision it forces |
|---|---|---|---|
| Segmentation | Split existing customers and the market by needs or behaviour, using sales data | Build provisional segments from interviews, substitute-product behaviour and adjacent markets | Which evidence counts, and what would falsify each segment |
| Targeting | Rank segments on size, growth, competition and economics; pick a pattern | Choose one beachhead that is cheap to reach and quick to prove | Whether to trade size for speed of evidence |
| Positioning | Frame of reference, point of difference and reasons to believe drawn from proof | Same structure, with proof points flagged as provisional | Which claims can be made before customers exist |
| Exclusions | Implied by the targeting pattern | Named segments the launch declines for a fixed window | What the team will say no to, and for how long |
| Review | At the next planning cycle | A dated rerun once early-customer data arrives | When the hypotheses are tested and who signs off changes |
Build segments from research and adjacent-market data
With no customer base, segmentation runs on borrowed evidence. Interviews with people who have the problem are the strongest source, especially when framed around the job they are trying to get done; the jobs-to-be-done framework is the natural lens, because it splits people by situation and motive rather than by description. Behind the interviews sit the customers of substitute products, search demand for the problem, and data from neighbouring markets where a similar offer already exists. The market sizing guide covers how to bound each segment's size with ranges rather than single figures.
The trade-off is between richness and reliability. Qualitative research reveals needs but samples few people; market data covers many but describes rather than explains. Choose a needs-based split confirmed by at least two independent sources, and write beside each segment the observation that would show it does not exist. That sentence is what the later review will test.
Target a single beachhead segment that gives proof fast
The generic targeting step weighs attractiveness against the ability to serve, and usually favours large segments. A launch should weigh something else first: how quickly a segment can turn into evidence. The best beachhead is reachable through a narrow, affordable channel, has an urgent version of the problem, talks to itself so that references spread, and can be served well by the first version of the product.
The trade-off is size against speed. A large, diffuse segment may promise more revenue but will take longer to show whether the product works and will cost more per customer (CACCost to acquire one new customer.). Pick the larger segment only when the product's economics cannot survive at small scale; otherwise pick the segment that produces a clear answer soonest. The launch planning and tiers guide helps size the effort so that a tier-one budget is not spent on a segment that only needs a tier-two push.
Write positioning with placeholder proof points
A launch can state a frame of reference and a point of difference with confidence, because both are choices. Reasons to believe are harder: there are no case studies, reviews or usage figures yet. The messaging hierarchy and proof points guide covers how claims are substantiated; for a launch, mark each proof point as one of three kinds: verifiable now (a specification, a guarantee, a founder's relevant experience), borrowed (independent research about the problem), or pending (the customer evidence the launch is designed to collect). The positioning statement framework records the result, with the pending items visible rather than disguised.
The trade-off is between persuasive force and honesty. Strong outcome claims with no evidence may lift early conversion but create a gap the product has to close. Choose guarantees and verifiable specifics over promised outcomes until the pending proof arrives.
Book a dated review to re-run targeting
The framework page notes that STP is static and needs a planning cycle to prompt its review. A launch cannot wait for one. Set the review date before launch, timed to when the first cohort will have used the product long enough to show who stays, who refers and who regrets buying. At that review, re-segment using actual behaviour, compare the beachhead's results with its hypothesis, and decide whether to deepen, shift or widen.
The choice here is between reviewing early, on thin data, and reviewing late, after money has been spent on the wrong segment. A fixed date with a minimum cohort size beside it avoids both reacting to the first week and drifting for a year.
Choose which segments to say no to during the launch window
Every launch attracts interest from people outside the target, and the temptation is to serve them all. Name the segments the launch will decline for a stated period, and write down why: unsuited to the first product version, too costly to reach, or likely to dilute the positioning. Saying no is not permanent; it is a sequencing decision. The product launches hub covers how internal teams are briefed so that sales and support know which requests to defer.
The trade-off is lost revenue against a clean signal. Serving an out-of-target segment on request is sensible when it costs nothing extra; actively marketing to it before the beachhead is won is the mistake.
Decision points in a worked STP for launch
An e-bike subscription chooses its beachhead in one city
Pedalwise, an invented e-bike subscription, plans to launch in Leeds at £59 a month including servicing and theft cover. With no customers, the team runs 24 interviews, studies a bike-share scheme's public usage data and reviews search demand. Four provisional segments emerge.
| Segment | Evidence | Est. size in city | Reach channel | Time to proof | Decision |
|---|---|---|---|---|---|
| Hill-district commuters | 11 of 24 interviews, bike-share peak data | 6,000 to 9,000 | Two employer schemes, local commuter groups | About 8 weeks | Beachhead |
| Delivery riders | 5 interviews, courier forum threads | 2,000 to 3,000 | Courier hubs | About 4 weeks | Declined for 6 months: heavy wear breaks the servicing model |
| Students | 6 interviews | 15,000 plus | Campus | Slow, term-bound | Declined until autumn review |
| Leisure riders | 2 interviews | Unknown | Broad social | Slow | Not targeted |
Positioning for commuters: frame of reference, the car or bus for a hilly commute; point of difference, arriving without sweating for less than a bus pass plus parking; reasons to believe, the all-in servicing guarantee (verifiable), council journey-time data (borrowed) and rider stories (pending). The review is fixed for week 12 with a minimum of 150 subscribers. At review, 210 have joined, 70% are commuters as planned, but a surprise 18% are parents doing school runs with a child seat. The team adds parents as a second hypothesis segment and replaces the pending proof with three commuter stories.
The example shows why each decision earned its place. The delivery segment looked fast to prove, but the exclusion protected the product's economics. The pending proof point made the gap visible, so the review had a specific job. And the dated rerun caught a segment nobody had interviewed, which a permanent STP would have missed.
Measuring a launch STP: KPIs and cadence
The measures for a launch STP check the hypotheses, not just the launch. The headline KPIA metric tied directly to an objective. is the share of early customers who come from the beachhead segment, because a launch that sells mainly to someone else has a targeting problem even if revenue looks healthy. Beside it sit acquisition cost by segment, early retention by segment and the proportion of new customers who mention the positioning's point of difference when asked why they bought. For a launch with a visible competitive set, Share of VoiceYour share of market visibility. within the beachhead's channels shows whether the concentrated budget is actually dominating the space it chose.
Cadence follows the review date. Acquisition by segment is worth a weekly look from launch; retention and reasons for buying need the cohort to mature and belong at the dated review. For distribution choices that reach the beachhead without paying to reach everyone, the distribution channels guide and, for professional segments, the LinkedIn campaign objectives and formats guide cover how to keep spend inside the target.
Where a launch STP misleads
Interview-led segments reflect who agreed to be interviewed, so the most vocal group can look larger than it is. A beachhead that succeeds may be unrepresentative: early adopters tolerate rough edges the wider market will not. Provisional proof points tend to become permanent if nobody owns replacing them. And the review can become a ritual that confirms the plan; it needs someone whose job is to argue that the hypotheses failed.
Common mistakes
- Treating research segments as facts. Without a falsifying observation beside each one, nobody can tell later whether the segment was real.
- Picking the biggest segment as the beachhead. Size slows evidence and raises acquisition cost before the product has proved anything.
- Hiding pending proof. Claims written as if customers already exist set expectations the first version may not meet.
- No date on the rerun. A review scheduled for "after launch" rarely happens until results disappoint.
- Marketing to declined segments. Accepting an inbound request is fine; spending launch budget on it blurs the signal.
