# Document Assessment Protocol — Business Plans, Pitch Decks & Idea Write-Ups

> Part of the public canon of *Perceptive Innovation* by **Caleb Kow** —
> canonical URL: https://perceptiveinnovation.com/canon/document-assessment.md · index: https://perceptiveinnovation.com/llms.txt
> © 2026 Caleb Kow, all rights reserved. Quotation with attribution to Caleb Kow / perceptiveinnovation.com is welcome.
> Commercial use — client engagements, training, certification, or incorporation into products or models — requires a written licence: https://perceptiveinnovation.com/terms

Source: operationalised from *Perceptive Innovation* by Caleb Kow for document review. This mode answers four questions about any plan the user provides: **What perception position does this plan evidence? Where does the framework touch each part of it? What has the author missed? And what would most improve its perceptive innovation standing?**

## Doctrine

1. **The document is the evidence base.** Every score cites the document — a quoted phrase, a section, a slide. Never score from general knowledge of the industry or from what the plan *probably* means.
2. **Silence is a finding, not a zero.** A plan that never addresses trust-building has a Trust *blind spot*. Score the sphere as a sceptical first reader of the document would perceive it, mark confidence Low, and log the blind spot — the gap in the plan is often more valuable to report than the score itself.
3. **Assess the venture's perception position, not the prose.** This is not an essay grade. A beautifully written plan can evidence a Critical-band perception profile; a rough one-pager can evidence a Strong one.
4. **Never fabricate facts about their market, competitors, or traction.** Where external reality matters and the document is silent, name the assumption.

## Step 0 — Ingest

Read the complete document before responding. Identify: document type (full plan / pitch deck / one-pager / academic case write-up), the innovation in one sentence, and **the audience the plan itself names**. If no audience is named, that is Blind Spot #1 and the assessment proceeds against the most plausible primary user, stated explicitly. For decks, treat each slide as a section; visuals and logos count as *asserted* claims unless data is shown.

## Step 1 — Section-to-Framework Relevancy Map

Map the document's own structure to the framework, so the user sees where PI touches *their* plan. Typical correspondence (adapt to their actual headings):

| Their Section | Spheres It Evidences | Levers It Evidences | What to Extract |
|---|---|---|---|
| Executive summary / positioning | Identity, Value | Thought, Tangibility | The one-sentence promise; who the user becomes |
| Problem & market analysis | Value, Future | Time, Temperature | Urgency framing; category trajectory claims; is the market climate read at all? |
| Product / solution / technology | Value, Effort | Technology, Tangibility | Capability claims vs demonstrable proof; what adoption actually asks of the user |
| Business model & pricing | Value, Risk | Tangibility, Time | The full cost stack; reversibility (trials, guarantees, exit) |
| Go-to-market | Identity, Effort | Time, Training, Talent | The awareness → familiarity plan; whether channels carry the right identity signal |
| Competition | Trust, Identity | Technology, Thought | Perception position vs feature-table comparison; which mental-model mode this is |
| Team | Trust, Future | Talent, Trace, Tenacity | Credibility a stranger would verify; borrowed trace (advisors, partners) |
| Financials & the ask | Future, Risk | Tenacity, Time | Continuity signals an outside audience would actually perceive |
| Risks | Risk | Trouble | Whether *perception-specific* risks appear at all |
| Traction / validation | Trust, Value | Trace, Tangibility | Verifiable proof vs self-reported assertion |

Report this map in the deliverable — it is the "relevancy of the framework to your plan" the user asked for, made concrete.

## Step 2 — Evidence Classification

Tag every material claim before scoring:

- **Evidenced** — independently verifiable within the document (named customers, data, certifications, signed partnerships, published benchmarks). Can support scores of 4–5.
- **Asserted** — claimed but unverified ("best-in-class", "users love it", "10x faster"). Caps the sphere at 3, with a note on what evidence would lift the cap.
- **Absent** — the sphere is simply not addressed. Score as a sceptical reader would perceive, confidence Low, blind spot logged.

## Step 3 — Score

Score all six spheres in canonical order (Identity → Value → Trust → Risk → Effort → Future) using the anchors in `pipas-assessment.md`, each with a cited basis and a confidence flag (High = evidenced; Medium = mixed; Low = asserted or absent). Compute the composite and band. Apply the standing rule: distribution before composite; lowest sphere first.

**Follow-up discipline:** if the two or three most consequential gaps could be closed by asking, offer up to three targeted questions before finalising ("Your deck names no referenceable customer — do any exist?"). If the user declines or is unavailable, finalise with confidence flags as they stand.

## Step 4 — The Blind Spot Library

Check the document against every pattern below. Report only genuine hits, each in the format: **what's missing → the book concept it comes from → why it moves adoption odds → the fix in one line.**

1. **No named audience.** Perception treated as monolithic ("the market", "users"). → *Sphere × Audience matrix.* Perception is audience-specific; unsegmented plans hide contradictions (user love on a regulator collision course). Fix: name the primary audience and any leverage audiences; score them separately.
2. **The technology-sufficiency fallacy.** The plan argues superiority will win adoption. → *The core thesis: perception holds a veto over technical superiority.* Fix: add a perception strategy alongside the capability roadmap.
3. **Mental-model mode unexamined.** No acknowledgment of whether this aligns with, extends, or replaces an existing mental model. → *Mental Model Economy.* Replacement-mode innovations pay a cognitive tax most plans underestimate by an order of magnitude. Fix: name the mode; budget the tax (education, demonstration, social proof) explicitly.
4. **Skipped perception-stack layers.** Conversion and revenue targets with no awareness or familiarity plan. → *Perception Stack (awareness → familiarity → expectation → evaluation → decision).* Users cannot evaluate what they are not familiar with. Fix: add the missing upper-stack activities and sequence them before conversion asks.
5. **Identity absent.** The plan covers features and value but never who the user *becomes* — the single most common miss. → *Identity sphere.* Fix: define the adopter identity and design visible, aspirational adoption signals.
6. **Gains-only value framing.** Benefits listed; the full cost stack (switching, learning, social, opportunity) ignored. → *Value sphere definition.* Users weigh perceived costs the plan refuses to see. Fix: state the full cost stack and attack its largest perceived item.
7. **No verifiable proof plan.** Trust rests entirely on the vendor's own word. → *Trust sphere + Trace lever.* Fix: schedule proof — named references, certifications, benchmarks, published cases — timed to land before the adoption decisions they must influence.
8. **Reversibility unaddressed.** No trial, guarantee, rollback, or exit path. → *Risk sphere: reversibility, control, best-to-worst asymmetry.* Loss-averse users read absent reversibility as a trap. Fix: add a visible reversibility mechanism (often the single highest-leverage move — it can lift Risk *and* Trust together).
9. **Effort journey unmapped.** No time-to-first-value, no onboarding step count, no drop-off thinking. → *Effort sphere.* Fix: map awareness-to-value; cut onboarding to five steps or fewer; instrument drop-off.
10. **Future = own projections only.** Hockey-stick financials, but no signals an *outside audience* would perceive as longevity — roadmap, partnerships, anchor customers. → *Future sphere.* Fix: publish continuity signals, not just internal forecasts.
11. **Trouble unaudited.** Regulatory, legal, ethical, and reputational exposure unmentioned. → *The Trouble Boundary.* Unaddressed trouble compounds and surfaces after launch, when remediation is most expensive. Fix: audit trouble by severity and visibility; decide what to acknowledge transparently and what to resolve substantively.
12. **Temperature unread.** Premium positioning into a cold, sceptical category — or a me-too entry into an overheated one — with no acknowledgment of the emotional climate. → *Temperature lever.* Fix: match message and pricing posture to the actual climate; design for the next temperature turn.
13. **No anchor use case.** Value spread thinly across many claims with no killer application to anchor perception and pull adoption. → *Format-war Killer App pattern.* Fix: pick the one use case compelling enough to adopt for on its own; lead everything with it.
14. **Feature-table competition analysis.** Competitors compared on capabilities only, never on perception position. → *Comparative perception.* Users choose between perceived positions, not spec sheets. Fix: add a perception comparison — who is trusted, who owns the identity, who feels safer — across the six spheres.

## Step 5 — Prescription

For every sphere at 2 or below (and the weakest two spheres of any Significant-band profile), route through `mapping-matrix.md`: primary levers → one or two interventions selected for multi-sphere leverage → metric attached → expected movement stated ("Risk 2→3, Trust 2→3"). For operators, sequence the actions 30/60/90 days with owners. Blind-spot fixes that don't map to a weak sphere still appear in the report — as plan improvements rather than interventions.

## Step 6 — Assemble the Deliverable

Use `assets/document-review-template.md`. Include the Framework Linkage block when the user is a student or asks how this connects to theory. Close in probabilistic language: the review shifts odds and removes veto conditions; it does not guarantee outcomes.

## Anti-Patterns (Never Do)

- Never grade writing quality, formatting, or slide design — perception position only.
- Never produce a generic SWOT or restate their plan back to them.
- Never hedge every sphere at 3 to stay safe — commit where evidence exists; that is the value of the instrument.
- Never exceed three follow-up questions or stall the review waiting for answers.
- Never rewrite their plan wholesale — prescribe perception moves; the plan remains theirs.
- Never soften the Critical-band recommendation: below 15, the honest counsel is restructuring, not a perception patch.