On bad output, see guides/skill-recovery.md

Business-Plan Market-Sizing Critique

Review a business-plan-shaped draft for one question: is the market story defensible —
large enough, addressable enough, capturable enough, and timed right? Plans that survive
moat and unit-economics scrutiny still fail on the market story — TAM derived from a category
report that doesn't match the product, an addressable segment defined so loosely it includes
customers who'd never buy, a capturable share exceeding what any comparable company reached at
comparable stage, a "why now" true for ten years, or a comp set assembled to support the model
rather than test it. Apply five lenses scoped to that question.

Scope (and what's deferred)

Covers market sizing only. Two adjacent dimensions are out of scope, handled by sibling
skills:

• Moat-and-distribution critique (moat type, distribution channel, switching cost, network
  effect, competitive response) — business-plan-critique-moat.
• Unit-economics critique (CAC, LTV, contribution margin, payback period, gross-margin
  trajectory) — business-plan-critique-unit-economics.

Also distinct from cowen-critique, which applies general argument-rigor moves
(boring-explanation-first, invert-the-thesis, revealed preferences, market signals) to any
prose argument. A Cowen critique may incidentally pressure the market claim — asking why no
$1B incumbent exists if the market is large — but does not systematically check TAM derivation,
segment realism, capturable share, "why now," or comparable benchmarks. Cowen pressures the
argument; this skill pressures the market-sizing methodology. Complementary: an
orchestrator can run both without findings collapsing into each other.

If the draft's biggest weakness is moat/distribution, unit economics, or general argument
structure, name it briefly under "Out of scope" in your Goal-Alignment Note and let the
orchestrator route it. Do not attempt those critiques here — keeping this skill narrow lets it
land sharp market-sizing findings without blurring into a generic business review.

Pre-flight: Skip Obvious Stubs

If the draft is under ~500 words AND contains TODO markers or structurally incomplete signals
(empty sections, missing thesis/intro, placeholder text), output the single line
draft incomplete; market-sizing critique skipped and stop. Both conditions must hold —
short pitches without stub markers, and longer drafts with stray TODOs, still get the full
critique.

Using the Fact-Check Report

If provided a fact-check report alongside the draft, treat it as your factual foundation. Do
not independently verify quoted TAM/SAM figures, market-growth rates, named research firm
reports, comparable-company revenue or share figures, or customer-count claims the report
already assessed.

Instead:

• Reference the fact-check findings where they bear on market sizing. A quoted TAM figure
  rated Unverified or Inaccurate directly affects the TAM Definition lens.
• Build on the fact-check where it surfaces ambiguity. A "Mostly Accurate" finding on a
  comp's year-5 revenue can still affect the Comparable Benchmarks lens if the comp operates
  in a structurally different segment.
• Focus on the lenses below — this skill uniquely provides the structural assessment of
  whether the market story holds together end-to-end (definition → segment → capture → timing →
  comparables).

If no fact-check report is provided, emit this warning at the top of your output before the
critique begins:

⚠️ No fact-check report provided. This critique does not include independent factual
verification. Checkable claims in the draft (quoted TAM/SAM figures, market-growth rates,
research firm reports, comparable-company revenue or share figures) have not been
assessed. For full verification, run the fact-check skill first or use the
draft-review orchestrator.

Then proceed, focusing on internal consistency and structural argument. Do NOT attempt your own
fact-checking — an ad-hoc spot-check without proper sourcing creates a false sense of
verification. When discussing numerical claims you cannot verify, flag your actual confidence:
"this $42B TAM figure is doing a lot of work and I haven't verified the underlying Gartner
methodology" beats either false confidence or vague hedging.

The Five Lenses

1. TAM definition and derivation

TAM (Total Addressable Market) is the most rhetorically loaded number in any pitch — it sets
the ceiling on the entire investment thesis, and small definitional changes produce
order-of-magnitude swings. Most common failure mode: a TAM derived top-down from a research-firm
category report that includes adjacent products, channels, or geographies the company does not
serve.

Moves:

• Top-down vs bottom-up. Top-down cites a category total ("global CRM market is $90B per
  Gartner") and claims a slice. Bottom-up multiplies addressable customers by realistic ACV.
  The two should roughly reconcile; if they diverge by more than ~3×, one is wrong. Plans
  presenting only one, especially top-down, skip reconciliation.
• Category-report fit. When TAM cites a research-firm category, check whether the category
  definition matches the product. A "marketing software" total bundles enterprise marketing
  clouds, mid-market automation, SMB email tools, and adtech — a SMB email product cannot claim
  the full category. Ask: what fraction of the cited category is addressable by this product?
• Adjacency creep. TAMs including "and we'll expand into X, Y, Z" mix today's TAM with
  future-product TAM. Separate them — investors fund the wedge product's TAM, not the imagined
  platform's.
• Geography and segment inclusion. A "$50B TAM" including geographies the company cannot
  legally serve, segments requiring certifications it lacks, or buyers it cannot price to
  (government, regulated verticals) is overstated by the inaccessible fraction.
• Methodology disclosure. If the plan does not disclose how TAM was derived, treat the
  number as opaque and say so. A confident TAM with a hidden derivation is worse than a smaller
  TAM with a transparent build.

If the bottom-up build (realistic addressable customers × realistic ACV) produces a TAM
materially smaller than the top-down headline, that's the headline finding. State both numbers
and the gap.

2. SAM realism (addressable segment)

SAM (Serviceable Addressable Market) is the slice of TAM the company can serve given its
product, channel, and operational reach. Most common failure mode: SAM defined by demographic
or firmographic filters that don't predict purchase behavior — "all US SMBs with 10-500
employees" includes a vast majority who'd never buy at this price.

Pressure-test the segment definition:

• Behavioral vs descriptive segmentation. Descriptive filters (industry, employee count,
  geography) are easy but weak — they identify who could be a customer, not who would buy.
  Behavioral filters (current spend on category X, observable pain Y, use of substitute Z)
  better predict conversion. Descriptive-only plans almost always overstate SAM.
• Buyer vs user. In B2B, user and buyer are often different people in different parts of
  the org. Size SAM by buyer count, not user count. Counting users (all engineers in target
  companies) overstates SAM by the user-per-buyer ratio.
• ACV-segment match. SAM × stated ACV must produce a coherent revenue ceiling. If SAM is
  100K accounts at $50K ACV, that implies a $5B addressable revenue line — plausible for the
  segment, or has ACV been quoted from a few enterprise wins while SAM was sized from SMB?
• Regulatory and certification gates. Segments requiring SOC 2, HIPAA, FedRAMP, ISO 27001,
  or industry-specific certifications are not addressable until the certification exists. Plans
  claiming healthcare, financial services, or government SAM without the gating cert — that
  fraction is aspirational TAM, not SAM.
• Wedge vs full segment. Early-stage companies serve a wedge — a narrow beachhead. Year-1
  SAM is much smaller than full SAM. Plans conflating steady-state SAM with year-1 addressable
  skip where the company starts.

If SAM is sized by descriptive filters alone, ask: of the filtered set, what fraction has the
behavioral signal that predicts purchase? That ratio typically reduces SAM by 5-20×.

3. SOM achievability (capturable share)

SOM (Serviceable Obtainable Market) is what the company projects capturing — typically a
percentage of SAM by some milestone year. Most common failure mode: a projected capturable
share exceeding what any comparable company achieved at comparable stage, without naming what
makes this company an outlier.

Check the capture math:

• Implied share at milestone. Back into the share from the revenue plan: year-5 revenue of
  $200M against $5B SAM implies 4% share. Plausible for the category in 5 years? In most B2B
  SaaS, 1% in 5 years is excellent; >5% requires category disruption or an unusually fragmented
  incumbent landscape.
• S-curve vs linear assumptions. Revenue plans growing share linearly (1% → 2% → 3% → 4% →
  5%) overstate near-term capture and understate the late push. Real category capture follows
  S-curves with a long tail. A straight share curve is almost certainly wrong-shaped, even if
  the integral is right.
• Beachhead → expansion logic. Healthy plans name which sub-segment they capture first
  (beachhead), what share they take there (often 20-40% of a small wedge is achievable), and
  how they expand. Plans projecting "X% of SAM" without naming the wedge project capture
  without a path.
• Competitive density at SOM. If the projected share requires displacing named incumbents,
  account for displacement cost — switching effort, competitive response (incumbent price cuts,
  bundling), and the realistic share gradient (you rarely take >30% from an incumbent actively
  defending). High share in a defended market without addressing this is projecting wishes.
• Sales-capacity reality check. Implied SOM ÷ ACV = customer count needed. Customer count ÷
  rep-productivity-per-year = sales reps required. Sales reps × ramp time = hiring plan. Often
  the implied sales org exceeds what the funding ask supports. If so, SOM is smaller, slower, or
  requires distribution efficiency the plan hasn't demonstrated.

If projected SOM exceeds the share captured by the most successful comparable company at the
same stage, the plan implicitly claims top-decile (or better) execution. State the implied
share, name the comp, and call out the implicit claim.

4. Market timing ("why now")

Market timing is whether the conditions making this product viable are newly present. Most
common failure mode: a "why now" true for ten years — trends so persistent they no longer
differentiate timing.

Pressure-test the timing thesis:

• Specific trigger vs persistent trend. "Cloud adoption is growing" has been true since
  2008. "AI is improving" since 2012. These are not "why now" — they are "why ever." A real
        "why now" names a specific change in the last 1-3 years: a new platform primitive
        (function-calling LLMs in 2023), a regulatory shift (a new mandate taking effect), a
        price-curve crossing (a cost dropping below willingness-to-pay), or a behavioral inflection
        (remote work becoming permanent post-2020).
• Adoption-curve placement. Where on the technology-adoption curve is the market —
  innovators (<2.5%), early adopters (2.5-16%), early majority (16-50%), late? Each phase has
  different sales dynamics, expected ACVs, and time-to-revenue. Pitching enterprise SaaS pricing
  at innovator-stage adoption is mismatched.
• "Too early" failure mode. Look for evidence the market is currently spending money on
  substitutes or workarounds (revealed preference) versus evidence it will spend once educated
  (assumption). The first is a real market; the second is category-creation, which is harder and
  slower.
• "Too late" failure mode. Symmetric: if the thesis is "this category is going to be huge"
  but the category already has a $5B incumbent, the plan is either late or positioning against an
  established player. Both viable, but the plan should acknowledge which and adjust strategy.
• Reversibility of the trigger. If timing depends on a single catalyst (a regulation, a
  platform change, a one-time behavioral shift), what happens if it reverses (regulation rolled
  back, platform changes API, behavior reverts)? Strong timing theses rest on multiple
  reinforcing triggers.

Framing: "Name the specific change in the last 24 months that makes this product viable now and
was not viable before. If you can't name one, the timing thesis is a trend, not a trigger."

5. Comparable benchmarks

Comparable companies (comps) are the most empirical input to a market-sizing argument — they
show what is achievable in adjacent or analogous markets. Most common failure mode:
cherry-picking comps that support the model and excluding ones that don't.

Pressure-test the comp set:

• Comp selection logic. Why these comps and not others? Plans naming 3 successful comps
  should also name 3 unsuccessful ones to show the realistic range. Citing only successes makes
  the implicit base rate ~100%, which is wrong.
• Stage-matching. Year-5 revenue at a comp founded in 2010 is not directly comparable to
  year-5 revenue here unless market conditions match. Comparing to a comp that benefited from a
  unique tailwind (the iPhone launch, the COVID remote-work shock) compares to an unrepeatable
  situation.
• Segment-matching. A comp that succeeded selling to enterprise IT is not comparable to a
  plan selling to mid-market marketing. Same category, different buyer, different sales motion,
  different ACV ceiling. Check the comp set's go-to-market matches the plan's.
• Outcome conditioning. Citing only comps that succeeded survives selection bias. The base
  rate for any specific market entry is dominated by the failures you don't see. The realistic
  comp set includes the median outcome and failure modes, not just outliers.
• Implied multiples. If the comp set implies "companies in this category reach $100M ARR by
  year 5," check whether any did so without unusual conditions (a category-defining partnership,
  a viral consumer moment, an acquired distribution channel). The headline may rest on conditions
  the plan can't replicate.
• Negative comps. What companies tried this and failed, and why? A plan that cannot name
  negative comps is either in a genuinely new category (rare) or hasn't done the research. Either
  way, name the failures and state which failure modes this plan's strategy avoids.

Framing: "If the comp set's median outcome is $X by year Y, and the plan projects $5X by year Y,
what makes this company an outlier? Name the structural advantage, or revise the projection
toward the median."

How to Structure the Critique

Output as a Markdown document. Use this exact section layout so downstream consumers
(orchestrators, format tests, human readers comparing critiques) can rely on the structure:

# Market-Sizing Critique: <draft title or topic>

## TAM Definition Assessment
## SAM Realism Assessment
## SOM Achievability Assessment
## Market Timing Assessment
## Comparable Benchmarks Assessment
## Factual Foundation
## Overall Assessment

If a lens does not apply (e.g., no comp set), keep the section heading and state what the plan
is missing rather than dropping the section. The skeleton is fixed; the content adapts.

Each of the five lens-assessment sections MUST include a **Verdict:** line valued one of
Defensible, Plausible, Inflated, Speculative, or Not Claimed (use Not Claimed for a
lens the plan does not address — no comp set provided, or no "why now" named). Each section MUST
also include a **Confidence:** line valued High, Medium, or Low, reflecting how
confidently you can land the verdict given the evidence in the draft (and fact-check report, if
provided). Place both fields at the top of the section, before prose.

TAM Definition Assessment

Reconcile top-down and bottom-up TAM. Identify category-report misfit, adjacency creep,
geographic/segment over-inclusion, and undisclosed methodology. State the gap between the plan's
headline TAM and the realistic addressable figure.

SAM Realism Assessment

Distinguish behavioral from descriptive segmentation. Check buyer-vs-user counting, ACV-segment
match, regulatory/certification gates, and wedge-vs-full-segment conflation. Estimate how much
SAM contracts under realistic behavioral filters.

SOM Achievability Assessment

Compute implied share from revenue plan ÷ SAM. Compare to comparable-company capture at similar
stage. Check beachhead-and-expansion logic, competitive-displacement realism, and sales-capacity
feasibility. Identify the most constraining bottleneck (capacity, share, expansion path).

Market Timing Assessment

Distinguish a specific 1-3-year trigger from a persistent decade-long trend. Place the market on
the adoption curve and check whether the product's pricing/sales motion matches that phase.
Address "too early" and "too late" failure modes. Assess whether timing rests on a single
catalyst that could reverse.

Comparable Benchmarks Assessment

Audit the comp-set selection logic — what was included, excluded, and why. Check stage-matching,
segment-matching, and conditioning bias. State whether the implied trajectory exceeds the
comp-set median, and what structural advantage justifies the gap. Surface negative comps the
plan does not name.

Factual Foundation

If a fact-check report was provided, summarize the findings that bear on market sizing —
especially Unverified or Inaccurate findings about quoted TAM/SAM figures, market-growth rates,
research firm reports, or comparable-company figures. If no fact-check report was provided,
identify the 2-3 numerical claims that would most benefit from fact-checking.

Overall Assessment

Of the five lenses, which surfaces the strongest market-sizing signal and which the weakest? End
with the single most important revision the author should make to the market story. End
constructively — the goal is a more defensible plan, not a takedown.

Output Location

When run standalone (not via the draft-review orchestrator), save your critique as
docs/reviews/business-plan-critique-market-sizing.md in the project root. Create
docs/reviews/ if it doesn't exist.

When run via the orchestrator, the orchestrator specifies the output path — follow its
instructions.

Goal-Alignment Note

When dispatched by an orchestrator, append a Goal-Alignment Note at the end of your critique
using the canonical form from
`patterns/orchestrated-review.md`:

## Goal-Alignment Note
- Answered: [yes / partial / no — one phrase]
- Out of scope: [what was set aside and why, or "none"]
- Escalate: [what the orchestrator should action separately, or "nothing"]
- Questions I would have asked: [1-3 short questions, only if scope was unclear; otherwise omit this bullet]

Use the Out of scope line to flag moat/distribution or unit-economics weaknesses you noticed
but did not critique (those belong to sibling skills), or general argument-structure weaknesses
better handled by cowen-critique. Use Escalate to surface a finding that crosses into
another critic's territory — a SOM argument depending on a distribution claim the moat critic
should re-examine, or a comp-set figure depending on a fact-check claim the orchestrator should
re-examine.

Tone

Direct but constructive. The spirit is "let's see whether the market story actually holds," not
"let me show you why the market isn't there." Founders have already heard generic TAM skepticism;
they need structural diagnosis they can act on. Be specific about which input is doing the
load-bearing work — which definitional choice, which behavioral filter, which comp — not just
what looks optimistic.

Comfort with uncertainty matters. Many market-sizing questions lack a clean early-stage answer —
"the bottom-up build is consistent with the top-down TAM within a 2× factor, acceptable at this
stage; the load-bearing assumption is the 30% behavioral conversion rate, which has not been
validated" is more useful than either "TAM is wrong" or "market looks fine." Calibrate confidence
explicitly, and when the plan's number could be right but the supporting evidence is thin, say
that exactly.