Market sizing assumptions attacked before the board sees them

For: Strategy lead, founder or investment analyst defending a market model

Pattern: Adversarial reviewNeeds live modelsDesigned for 6 to 80 agents

The pain today

A market model is a chain of assumptions, each borrowed from a report. The author believes them. The first hard question in the room finds the one that does not hold.

The ask

I attached our market model notes with every assumption listed and the reports we took them from. Have one team defend each assumption from the sources and another try to break it. Show me which assumptions survive, which the sources do not support, and which rest on a single source.

Plain words, as you would say it to a colleague. Edit it to fit your case before you send it.

What you attach or connect

  • Market model notes with assumptions listed
  • Source reports and dataset descriptions
  • Segment and geography definitions used in the model

The unit of work

One worker task per one assumption with its cited sources.

Why a swarm fits

Assumptions can be tested one at a time against their own sources. Opposing teams from different families keep the author's optimism out of the check.

Not for

A rough estimate for an internal chat: nobody needs an adversarial review of a guess.

The decision tree

6 typed decisions, each with an action for every answer

At fixed moments in a run, the engine puts one narrow question to a decision model. The decision model never writes text: it answers yes or no with a probability, picks from listed options, or gives a score, about a small slice of the material. The engine then does exactly what this tree says, which is what makes the run auditable. The thresholds are the template's design values, not measured results.

  1. Planner, while planning

    Scope checkA choice among options

    Before work starts on a unit

    Where does this assumption's figure come from?

    Sees only: One assumption as written in the model notes and the list of supplied sources

    Why: An assumption with no source cannot be defended or attacked, only owned by its author.

    • A supplied sourcethenAccept
    • A source that was not suppliedthenMark unresolved
    • The author's own estimatethenAsk a person
  2. After workers, the judge checks

    Evidence checkYes or no, with a probability

    After a worker answers

    Does the quoted source passage give the figure for the same segment, geography and year as the model uses it?

    Sees only: The assumption as used in the model and the quoted source passage with its definitions

    Why: Borrowed figures usually fail on scope, not on arithmetic.

    • Yes: 0.85 or higherthenAccept
    • Unsure: 0.50 up to 0.85thenEscalate to a strong model
    • No: below 0.50thenReject and retry
  3. Evidence checkYes or no, with a probability

    After a worker answers

    Does the attack quote a supplied source or a definition that differs from the model's, rather than argue from general plausibility?

    Sees only: The attack's argument and its quoted passage

    Why: Keeps the challenge to evidence, so the author cannot wave it away as opinion.

    • Yes: 0.85 or higherthenAccept
    • Unsure: 0.50 up to 0.85thenMark unresolved
    • No: below 0.50thenReject and retry
  4. Reconciler, while merging

    Conflict checkA choice among options

    While reconciling

    As the model uses it, does the assumption survive the exchange?

    Sees only: The verified defence and the verified attack for one assumption

    Why: Single-sourced and wrongly scoped links are reported as such, not as passes.

    • Survives on the quoted sourcethenAccept
    • The source supports a different scopethenMark unresolved
    • The source says otherwisethenAsk a person
    • Holds, but on a single sourcethenMark unresolved
  5. Run control, between rounds

    Another round?Yes or no, with a probability

    Between rounds

    Did the last exchange produce a new quoted passage from either side?

    Sees only: The quotes cited in the last two exchanges for one assumption

    Why: Ends the argument when both sides are only repeating themselves.

    • Yes: 0.60 or higherthenContinue
    • Unsure: 0.30 up to 0.60thenStop
    • No: below 0.30thenStop
  6. Accountable person, before anything is settled

    Person decidesYes or no, with a probability

    Before anything is reported as settled

    Does a failed or single-sourced assumption feed the headline market figure through the chain of dependencies?

    Sees only: The assumption's ruling and its place in the dependency chain

    Why: The author owns the figure that is presented and decides what to change.

    Accountable: The model's author decides which assumptions to change and owns the figure that is presented.

    • Yes: 0.30 or higherthenAsk a person
    • Unsure: 0.10 up to 0.30thenAsk a person
    • No: below 0.10thenAccept

The fleet: who does what

Model tiers by role, not brands: you choose the models. Strong reasoning models plan and reconcile, small fast models do the wide work, and the judge is a decision model from a different family, so it does not share the workers' blind spots.

  1. Planner

    A strong reasoning model lists the assumptions, their sources and how each feeds the next.

    Decisions here:1. Scope check

  2. Workers

    Defence and attack workers from two different families argue each assumption from the source text only.

    Designed for 6 to 80 agents, one worker task per one assumption with its cited sources. Each worker receives only its own unit.

  3. Judge, from a different model family

    A decision model from a third family rules whether the quoted source supports the assumption as it is used.

    Decisions here:2. Evidence check3. Evidence check

  4. Reconciler

    A strong reasoning model reports survivors, failures and single-sourced links, and shows which conclusions depend on them.

    Decisions here:4. Conflict check5. Another round?

  5. Accountable person

    The model's author decides which assumptions to change and owns the figure that is presented.

    Decisions here:6. Person decides

Checked before anything is accepted

  • Defence must quote the source passage and its definition of the figure
  • Attacks must cite a source or a definitional mismatch, not a hunch
  • Assumptions resting on one source are marked single-sourced
  • The swarm does not produce a replacement market size

What comes back

  • Assumptions that survived, with quotes
  • Assumptions the sources do not support
  • Definition mismatches between model and source
  • Conclusions that depend on a failed assumption

What to measure

  • Challenged assumptions the author changes
  • Hard questions in the room that the review had not raised
  • Hours to prepare the defence
  • Cost per assumption tested

Names of measures only. No result is claimed for this template.

Templates open in the workspace chat with the ask filled in. Nothing runs until you send it.

Get early accessSign in to use

Literature review where every claim cites a passage

For: Research lead, analyst or doctoral researcher scoping a field before committing to a direction

A review means a very long reading list.

Pattern: Map, verify, reduceNeeds scale5 decisionsDesigned for 30 to 800 agents

Competitor teardown from public sources with citations

For: Strategy analyst, product leader or founder preparing a market entry or board discussion

A teardown is stitched from a few web pages and hearsay.

Pattern: Hierarchical decompositionNeeds a connector6 decisionsDesigned for 20 to 500 agents