Challenge impairment and valuation assumptions unit by unit

For: Chief financial officer, valuation lead or audit partner reviewing impairment tests

Pattern: Adversarial reviewNeeds live modelsDesigned for 8 to 200 agents

The pain today

Growth, margin and discount rate assumptions for each cash-generating unit are defended with management's own narrative. Contradicting evidence in board papers, budget-versus-actual reports and order books is rarely lined up against them.

The ask

I attached the impairment models, the assumption memos, board papers, budgets against actuals and order book reports for each business unit. For every key assumption, make the case for it from these documents, then argue against it from the same documents, and tell me which assumptions are poorly supported.

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

  • Impairment or valuation models per unit
  • Management assumption memos
  • Board papers and strategy documents
  • Budget-versus-actual history
  • Order book, pipeline and market reports

The unit of work

One worker task per one assumption for one cash-generating unit.

Why a swarm fits

Each assumption is tested against a narrow slice of one unit's documents, and a group has many units with several assumptions each. Supporting and attacking with different model families reduces shared optimism.

Not for

One unit with ample headroom, or choosing a discount rate from market data the documents do not contain.

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 checkYes or no, with a probability

    Before work starts on a unit

    Is this an assumption the outcome is sensitive to, according to the sensitivity table in the memo?

    Sees only: The assumption and the memo's sensitivity table for the unit

    Why: Debate is spent on the assumptions that can move the result.

    • Yes: 0.60 or higherthenAccept
    • Unsure: 0.30 up to 0.60thenAccept
    • No: below 0.30thenSkip this unit
  2. After workers, the judge checks

    Evidence checkYes or no, with a probability

    After a worker answers

    Do the quoted order book or budget-versus-actual figures support the growth rate assumed for this unit in the years shown?

    Sees only: The assumption and the supporting side's quoted figures

    Why: Management narrative is not evidence; figures from the unit's own records are.

    • Yes: 0.85 or higherthenAccept
    • Unsure: 0.50 up to 0.85thenMark unresolved
    • No: below 0.50thenReject and retry
  3. Evidence checkA choice among options

    After a worker answers

    What does the challenger's quoted document say about this assumption?

    Sees only: The assumption and the challenger's quoted board paper or report

    Why: A challenge has to be on point to count against an assumption.

    • States a lower or later figure than the memothenMark unresolved
    • Concerns another unit or periodthenSkip this unit
    • Opinion with no figurethenSkip this unit
  4. Reconciler, while merging

    Conflict checkYes or no, with a probability

    While reconciling

    Does the input in the model differ from the figure the memo states for the same assumption?

    Sees only: The model input cell and the memo sentence

    Why: Memo and model drifting apart is a finding auditors always raise.

    • Yes: 0.60 or higherthenMark unresolved
    • Unsure: 0.30 up to 0.60thenMark unresolved
    • No: below 0.30thenAccept
  5. Run control, between rounds

    Another round?Yes or no, with a probability

    Between rounds

    Did the last exchange cite a document not already weighed for this assumption?

    Sees only: The citations of the last exchange

    Why: Stops when both sides have run out of documents.

    • 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

    Is this assumption contradicted or untested, so that management and the auditor must decide what to do with it?

    Sees only: The assumption register entry

    Why: Management owns the assumption; the auditor owns the conclusion.

    Accountable: Management owns the assumptions and the impairment decision; the auditor owns the audit conclusion. The swarm only lines up evidence.

    • Yes: 0.40 or higherthenAsk a person
    • Unsure: 0.15 up to 0.40thenAsk a person
    • No: below 0.15thenAccept

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 key assumptions per unit and the documents that could confirm or contradict each.

    Decisions here:1. Scope check

  2. Workers

    Small fast workers argue for one assumption; workers from another family argue against it, both quoting the documents.

    Designed for 8 to 200 agents, one worker task per one assumption for one cash-generating unit. Each worker receives only its own unit.

  3. Judge, from a different model family

    A decision model from a third family rules whether the assumption is supported, contradicted or untested by the evidence.

    Decisions here:2. Evidence check3. Evidence check

  4. Reconciler

    A strong reasoning model summarises weak assumptions per unit and notes where management's own documents disagree.

    Decisions here:4. Conflict check5. Another round?

  5. Accountable person

    Management owns the assumptions and the impairment decision; the auditor owns the audit conclusion. The swarm only lines up evidence.

    Decisions here:6. Person decides

Checked before anything is accepted

  • Historical forecast accuracy is computed in code from budget-versus-actual data
  • Both sides must quote documents; assertion without a source is discarded
  • The model's inputs are checked against the figures stated in the memo
  • Assumptions with no relevant evidence either way are reported as untested

What comes back

  • Assumption register per unit: supported, contradicted or untested
  • Contradicting evidence with quotes and dates
  • Differences between memo, model and board papers
  • Questions for management per unit

What to measure

  • Challenged assumptions management later revises
  • Auditor challenges the review had anticipated
  • Review hours per cash-generating unit

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

Reconcile intercompany agreements with the transfer pricing policy

For: Group tax manager or group controller preparing transfer pricing documentation

Markups, payment terms and service scopes are set in the transfer pricing policy, restated in intercompany agreements and changed in addenda.

Pattern: Map, verify, reduceRuns today6 decisionsDesigned for 4 to 120 agents

Break a group audit plan into entity, process and assertion tests

For: Head of internal audit or group audit senior manager planning the annual audit

A group audit spans many entities, each with its own processes, ledgers and prior findings.

Pattern: Hierarchical decompositionNeeds scale6 decisionsDesigned for 20 to 1,000 agents

Extract lease accounting inputs where the wording needs judgement

For: Group accountant or lease accounting lead maintaining the lease register

Lease term, extension and termination options, indexation and incentives sit in contracts and amendments in several languages.

Pattern: Cross-examinationNeeds live models6 decisionsDesigned for 6 to 300 agents