Evidence-backed market wedge system
Compare customer segments, separate evidence from assumptions, identify hard disqualifiers and decide what to validate first. The tool can produce a current focus hypothesis, but it will not manufacture a “best ICP” from default scores or input order.
Unknown stays unknown
Missing assessments are never silently converted to a neutral-looking 3/5.
Evidence is inspectable
Tag assumptions, directional evidence and stronger evidence with a real basis and optional note.
No leader is valid
Weak coverage, effective ties and fragile comparisons become explicit next-learning work.
Quick = core wedge criteria. Deep adds technographics, buying committee, evidence notes, customer signals and sensitivity inputs. One candidate can be audited; two or more can be compared.
Market context
This is the shared context. Candidate segments can still have different problems, triggers, alternatives and desired outcomes.
For B2B, the ICP is primarily an account/segment hypothesis: company characteristics, buying situation, economic fit, Product fit, trigger and exclusions. Buyer, champion and end-user roles support the decision but do not turn the ICP into an individual persona.
After a segment becomes the current focus hypothesis, hand its roles and evidence to the Persona Generator and its context/problem/trigger/alternative/outcome to the JTBD tool.
B2C mode prioritizes context, behavior, recurring problem, alternatives, willingness-to-pay proxies and access. Age and region remain optional rather than becoming the default driver. Stated willingness to pay stays a proxy until observed purchase or renewal behavior supports it.
1 · CONTEXT
Define the Product, category, price/capacity, global problem and review trigger.
2 · CANDIDATES
Create 1–4 Quick or up to 6 Deep candidates, each with its own problem, trigger, alternative, channel and exclusions.
3 · EVIDENCE
Assess favorable criteria on explicit 1–5 anchors and record assumption/evidence status and basis.
4 · DECISION
Inspect coverage, trade-offs, sensitivity, audit findings and gap-driven validation actions.
5 always means more favorable. Procurement ease, integration feasibility and adoption feasibility replace confusing inverse risk scales.
A 4.1 based on a few assumptions is visibly different from a 4.1 supported across most important criteria. Missing values are not averaged in.
If changing one uncertain input by one point can erase or flip the lead, the comparison is labeled fragile and that assumption becomes a validation target.
TAM asks who could theoretically buy or use the Product. The ICP / focus segment asks where Product fit, customer value, economics and go-to-market access currently appear most promising. This tool does not estimate market size.
This tool compares segment hypotheses, not individual companies or leads. It answers “Which type of customer should we test or focus?” rather than “Which account should a rep call today?”
Durable output
The audit groups issues by segment definition, problem fit, economics, Product fit, distribution, evidence, decision quality and handoff. The ledger keeps each candidate’s definition, problem/context, alternative, trigger, channel/access, exclusions, evidence, assessments and next validation action together so the decision remains inspectable after the meeting.
Symptom: A segment appears to win before anyone has assessed it.
Cause: Missing criteria were converted into default midpoints.
Fix: Keep missing values null and block leader selection until comparison coverage is meaningful.
Symptom: High procurement risk improves a score.
Cause: An inverse risk scale was multiplied by a positive weight.
Fix: Score procurement ease so 5 always means favorable.
Symptom: Every candidate uses the same problem statement.
Cause: Only a global Product problem was captured.
Fix: Record the problem, trigger, alternative and desired outcome per segment.
Symptom: Reachability says LinkedIn/email with no evidence.
Cause: The tool invented a channel from B2B/B2C context.
Fix: Capture known access or label an explicit channel hypothesis.
Symptom: A polished ICP contains customer results nobody supplied.
Cause: Messaging generation inferred proof, urgency or differentiation.
Fix: Export messaging inputs only; leave unknown proof and differentiation explicit.
Symptom: B2B records disappear after switching to B2C.
Cause: Incompatible candidate state was silently retained.
Fix: Require a deliberate reset before changing the global context.
For B2B, an ICP is an account- or segment-level hypothesis about the organizations most likely to buy, realize value and remain a good fit. It is narrower than a total target market and different from a buyer persona. This tool treats the ICP as revisionable evidence-backed Product judgment, not a permanent label or a list of individual leads.
The ICP answers which company or customer segment should be the current focus hypothesis. A persona describes a specific human role, behavior and motivation inside that segment. B2B buyer, champion and end-user fields are supporting inputs here, but the downstream Persona Generator owns the richer person-level artifact.
Only criteria you explicitly assess are included. Missing values remain unknown rather than defaulting to 3/5. All scales run in the same favorable direction, and importance is transparent. The score is rounded and presented with coverage and evidence status. The tool can return no clear leader when coverage is weak, scores are effectively tied or the apparent lead is fragile.
You can label an assessment as an assumption, directional evidence or validated evidence and record its basis, such as an interview, closed-won observation, retention, usage, support, sales observation or market research. These labels are descriptive rather than statistical confidence claims. One quote can inform a hypothesis without proving it across a market.
Ability to pay describes economic capacity or available budget. Willingness to pay is better supported by observed price acceptance, purchase, renewal or paid workaround behavior. A large company may have budget but no willingness to buy. A consumer may say they would pay without ever purchasing, so B2C willingness-to-pay inputs remain proxies unless behavior supports them.
That is a valid output. The tool will not manufacture a Top 1 from insertion order, default scores or tiny differences. It surfaces missing evidence, ties and sensitivity drivers, then generates gap-driven validation actions with a hypothesis, evidence source and decision rule so the next step improves the decision rather than merely producing more research.
Not as a primary output. It provides messaging and positioning inputs such as evidenced pain, desired outcome, trigger, current alternative, available proof and differentiation hypothesis. Finished landing-page or acquisition copy belongs downstream, where every claim can be reviewed. The ICP tool does not invent results, urgency, proof or claims that the Product is already built for a segment.
No. The share link compresses the current ICP state into the URL; compression is not encryption. The tool warns that anyone with the link can inspect Product context, evidence notes and internal assumptions, and it recommends anonymizing customer or account details. If the payload is too large for a reliable URL, export Markdown or JSON instead.
Once the wedge is explicit, continue with the artifact that owns the next job instead of generating everything inside the ICP tool.
Use CraftUp’s discovery, survey, JTBD and strategy workflows to test the assumptions that could change the focus decision-not to decorate a profile that already looks complete.
Last updated: 2026-09-18