Step six. Positioning — father VP (07), offers (08), buyer journey (09) and all materials: an error here multiplies down the entire flow. Positioning lives in comparison (Dunford), so we build it FROM whitespace of step 05. Cold audit → mostly a hypothesis with a confidence level. Checkpoint GATE 2.5.
Key principle: boldness of the claim is allowed anchored by stage and available proofs. You cannot
claim a category or premium before cases exist (overshoot — agency-stages.md §8).
Input — 02 (stage + SMF, usually a hole in Cr = "don't close because there's no validated positioning"),
03 (market/ICP/persona/JTBD), 05 (competitor matrix + whitespace + battle angles), 01 (verbatim/pains/cases).
Read ${CLAUDE_PLUGIN_ROOT}/reference/agency-stages.md (stages §1, SMF §2, overshoot §8) and
${CLAUDE_PLUGIN_ROOT}/reference/gtm-frameworks.md §3 (2×2 matrix), §4 (Peršolja 8-step, UVP/USP), §9 (anti-patterns).
Reuse
competitor-finder (this pack) / artifact 05 (DO NOT duplicate — read) · persona-insights-analysis (this pack) or
account-dossier (account-management-skills) — the client's language · value-prop-lister (sales-engine-skills) ·
gtm-action-thinker (this pack, angle stress test) · anticopywriting-ai (gtm-skills, human tone). All — if installed.
The "Propose-by-default" principle (cross-cutting)
This is a cold audit — the client often cannot name on their own the enemy, anchor or best customer. Rule:
an empty field = not "—", but a hypothesis from the skill, derived from the site / cases / 01-verbatim / 05-whitespace,
explicitly marked: (гіпотеза · впевненість low/med/high · підстава: …) — the same style as the "stage
hypothesis" in agency-stages.md. Always give 1 main + 1 alternative (not 5). Hypotheses → in the block
"Sources / assumptions" and on GATE 2.5 as "here's my version — do you confirm or replace it?".
Skeleton (8 steps)
1 · Best-customer pattern. From 01/03 identify 3-5 clients who bought quickly + got results + margin. Find the pattern (industry / size / trigger / JTBD / problem) — this is the latent niche. Stage 0-1: 1-2 cases or founder experience before the agency (borrowed) is enough.
2 · Competitive alternatives. From 05 (DO NOT duplicate the matrix). For an agency the main alternative is usually not an agency: in-house team · freelancers · do nothing / status quo. Name what the client actually compares against.
3 · Axis + Anchor + ENEMY (an agency fork that doesn't exist in SaaS methods).
- Axis of specialization (Morgan, 5): Vertical · Horizontal · Market/Stage · Role/Function · Platform/Method.
- Anchor (Estner, 4): Activity · Use-Case · Service-Category · Competitive-Alternative.
- Choose 1 axis + 1 anchor for the stage (gate below). Enns: the goal = the intersection discipline × market expertise; the point of narrowing is depth, not narrowness.
- ENEMY angle (Herubel, mandatory): (1) name the enemy — not a competitor, but the outdated way/status quo you replace ("horizontal full-service generalists", "in-house drowning in clashes", "5 freelancers and a prayer"); (2) why it breaks — 2-3 structural reasons (money/time/risk); (3) your "new way" — a methodology/process that hits exactly this hole. Without an enemy, positioning slides into vanilla "we're like the others, only better" (anti-pattern §9; red flag Z1.4 — "no enemy"). If the client can't — suggest the enemy yourself (propose-by-default).
4 · Score assets vs table-stakes. Assemble the assets (features, process, methodology, team, cases — the story is true). Each against alternatives (step 2): table-stake (everyone has) or differentiator (only you). Keep the mission-critical diff. 2×2 matrix (§3) — the anchor "where we are in the field".
5 · Assets → benefits + proof. For each diff — "So what?" ×5 → functional / social / emotional benefit. Proof by stage level (gate): borrowed (resume/analog/method) → owned (cases with $). Every benefit with proof, otherwise drop it. A borrowed claim is never presented as owned.
6 · USP + positioning style. USP = valued by client AND unique to you (§4). Not "only" → not sharp enough. Choose a Dunford style for the stage (gate): Head-to-Head / Big-Fish-Small-Pond / Create-New-Game.
7 · Land statement + 3 benefits. Canonical template (Moore/Dunford):
For [ECP] who [need/trigger], [agency] — is a [category/anchor] that [key benefit]. Unlike [main alternative/enemy], we [main diff].
- exactly 3 key benefits with proof. This is the artifact consumed by VPs (07) / decks / content / outreach.
8 · "Say no" / standards test (validation, Enns). Formulate 2-3 types of work that positioning forces you to give up. Empty → positioning is not yet real, go back to step 3.
Stage gate (the heart of the step)
Read from 02. Determines the ceiling of boldness:
| Stage 0-1 | Stage 2 | Stage 3-5 | |
|---|---|---|---|
| Positioning | provisional hypothesis (beachhead) | committed on evidence | defended and evolves |
| Dunford-style | only Big-Fish-Small-Pond | Big-Fish → Head-to-Head | Head-to-Head / Create-New-Game |
| Anchor | Competitive-Alt / Use-Case | Use-Case → Service-Category | Service-Category / own Category |
| Proof base | borrowed (resume/analog/method) | repeatable cases in the niche | owned: category authority, $-results |
| Enemy | broad status quo | narrowed failure-mode of the niche | can become the category name |
| «Say no» | aspirational direction | refusal of ≥1 off-pattern deal | non-negotiable standards |
| Risk #1 | niched down → died before proof | rollback to "it's for everyone" | positioning-delivery gap |
| Repositioning trigger | — | $1M–$3M ARR (§4) | new PMF / segment / product / new practice |
Stage 0 separately: "no positioning, we do everything" — this is a diagnosis. Here only 1 beachhead hypothesis from 1 best client. No matrices.
Self-scoring maturity (rubric Z1.4)
At the end — an assessment of the CURRENT positioning by rubric Z1.4 (5 levels, max=4) + "now → ceiling by stage → gap". The levels map almost one-to-one onto the gate:
| Lvl | Z1.4 | State | ≈ Stage |
|---|---|---|---|
| 0 Initial | service list («Web, Mobile, AI — for any business») | 0 | |
| 1 Basic | statement only with the founder; not in sales/website; "category of one" without proof | 0-1 | |
| 2 Structured | positioning canvas (for whom / alternatives / unique value / category / evidence) | 2 | |
| 3 Advanced | diff by segments; battle cards; win/loss aligned; message A/B tests | 2-3 | |
| 4 Optimized | category creation / subcategory ownership; verbatim; annual re-validation | 3-5 |
Scoring discipline: the highest level where ALL criteria are met; 3 of 4 = the previous one; each score — with evidence (a quote from the website/discovery), otherwise score = one level lower; artifacts > words. Ceiling = what the gate allows (do not set Lvl 4 for the target at Stage 1 — overshoot). Output 🚩 red flags that triggered (e.g. "we're like others, only better", "no enemy"), and gap N→N+1 as a checklist for this cycle.
Quality checks
- USP passes the "only"-test + valued-AND-unique.
- The alternative is named correctly (often in-house/status-quo, not an agency).
- No benefit without proof; proof matches the stage (borrowed ≠ owned).
- Style + anchor ≤ allowed by the gate (no category-claim at Stage 1).
- The enemy is named; "say no" is non-empty.
- §9 not violated: "everyone", vanilla UVP, copy of a competitor's positioning.
- No key field is empty: where the client was silent — a hypothesis with confidence+rationale.
Output (Notion)
Sub-page «06 · Positioning»: stage-gate banner (stage + what is allowed) · best-customer pattern · alternatives (link to 05) · axis + anchor + ENEMY (with rationale for the stage) · assets→benefits→proof table · positioning statement + 3 benefits (main artifact) · "say no" standards · maturity Z1.4: now → ceiling → gap · "Sources / assumptions" block. In Ukrainian; statement and category names — in the original language (often EN for US).
GATE 2.5 — checkpoint (light)
Before VP/offers show: axis + anchor + enemy (1 line each) · statement + 3 benefits · maturity now→target. Present hypotheses (where the client was silent) as "here's my version — do you confirm or replace?". "Say no" question: "what work are you ready to refuse?". Only after "yes" does the orchestrator go to 07-value-prop.
Definition of Done
- Stage read from 02; gate applied (style/anchor/proof ≤ ceiling).
- Best-customer pattern derived (≥1 client/experience).
- Alternatives pulled from 05 (not duplicated); a real one is named (often in-house/status-quo).
- Axis + anchor + enemy chosen and anchored by stage.
- Assets filtered to diff vs table-stakes; every benefit has proof at the stage level.
- Statement + 3 benefits recorded; "say no" formulated.
- Maturity Z1.4 (now → ceiling → gap) derived from red flags.
- There are no empty key fields (hypotheses where needed). GATE 2.5 passed. Notion page created.