Calibration

Every workplace strategy is a structure of assumptions: growth projections, attendance factors, sharing tolerances, overlap requirements, space standards, financial constraints, policy and technology futures. Most strategies bury these assumptions inside their numbers, where they age silently until reality diverges and nobody can say which assumption failed. Calibration makes the assumption structure explicit, owned and reviewable — the single practice that separates a strategy that can be updated from a report that can only be replaced.

• In Define, once the demand model exists and before Scenario Development varies it
• Whenever a model produces a number leadership will act on — the number's assumptions enter the register
• At every phase gate, as a standing review of register currency
• Continuously in Recalibrate, where Assumption Recalibration is this method operating in steady state

• Strategist as register architect
• Named assumption owners: Finance owns growth and cost assumptions, HR owns policy assumptions, CRE owns portfolio assumptions, IT owns technology assumptions
• Leadership team accepting the register and its confidence grading
• Project team maintaining linkage between assumptions and models

• Every quantified input to the demand model: growth plans, attendance and presence logic, sharing ratios, overlap requirements, design and space standards, financial constraints, future ambition, policy and technology assumptions
• Confidence evidence per assumption: measured, projected, decided or asserted
• Sensitivity structure from Taxonomy Alignment

1. Extract every assumption from the models into a single register — including the invisible ones embedded in standards and inherited constraints.
2. Classify each by source: measured (evidence-based), projected (extrapolated), decided (a governance choice), asserted (someone said so). Asserted assumptions are the register's risk concentration.
3. Grade confidence and volatility per assumption: how sure are we, and how fast could it move?
4. Assign a named owner to each — the person accountable for flagging when reality diverges.
5. Run sensitivity analysis: for each high-volatility assumption, what moves in the strategy if it shifts by a plausible margin? Attendance and growth assumptions typically dominate.
6. Define trigger thresholds: the divergence level at which each assumption forces model regeneration rather than tolerance.
7. Publish the register alongside the strategy — assumptions hidden from decision-makers are decisions taken from them.
8. Establish the review cadence that Recalibrate will inherit.

• Asserted assumptions wearing measured costumes ('our attendance will be 60%' with no evidence lineage)
• Assumptions without owners — orphaned assumptions never get flagged, only discovered
• Compound risk: multiple volatile assumptions stacked inside one headline number
• Political assumptions: growth figures that are ambition statements, attendance figures that are policy hopes
• Precision theatre: assumptions stated to decimal places their evidence cannot support

• Assumption register: source, confidence, volatility, owner, trigger threshold per assumption
• Sensitivity analysis on the demand model
• Trigger framework for model regeneration
• Review cadence handed to Recalibrate

Scenario Development varies exactly the assumptions this register identifies as volatile. The Workplace Strategy publishes the register as part of its evidence base. Living Workplace Outputs keep register and models linked, so an assumption breach regenerates the affected models. Assumption Recalibration in the sixth phase is this register under continuous review — the mechanism that makes the strategy living rather than archived.

The platform maintains linkage and flags divergence; humans set the thresholds and own the assumptions. Judging when an assumption is decided versus asserted, how much compound risk a scenario can carry, and when a political assumption must be confronted rather than registered — that is governance craft, and it is frequently uncomfortable.

• Registering only the convenient assumptions and burying the political ones
• No named owners, so the register is nobody's job
• Thresholds never defined, so divergence accumulates until the strategy is quietly obsolete
• Treating the register as project documentation instead of a living governance instrument
• Confusing sensitivity analysis with scenario planning — sensitivity tests one variable, scenarios test coherent futures

A media company's register exposed that its entire capacity model stood on an asserted 55% attendance factor sourced from a leadership away-day, not evidence. Reclassified as 'decided' with the COO as owner and a ±8-point trigger, it breached fourteen months later when measured attendance stabilised at 66%. Because the trigger existed and the models were linked, the capacity model regenerated within days — instead of the alternative timeline in which the building is discovered to be too small during a lease negotiation.