Organisations hold more workplace evidence than they realise — and less than they claim. Headcount lives in HR, floor plans in FM, booking data in IT, cost data in Finance, and none of it reconciles. Requesting data early exposes what exists, what is reliable, what contradicts, and what must be gathered fresh. It prevents the two classic failures: commissioning research that duplicates existing data, and building a strategy on data nobody validated.
- In the first two weeks of Discover, in parallel with governance design
- Before scoping Diagnose methods — the inventory determines which are necessary
- When inheriting a project with prior studies of unknown quality
- Before any platform-enabled method is configured, to establish baseline data
- Workplace project team lead (owns the request)
- Named data owners in HR, Finance, CRE/FM, IT and the business
- Data protection / privacy officer for anything person-related
- Workplace strategist defining what each dataset must answer
- Project definition document (defines what the evidence must inform)
- Organisational chart identifying data owners
- The request template itself: a structured inventory across all evidence categories
- Issue a structured request covering: business strategy and growth plans; organisational charts and headcount (current and projected); HR data (attrition, recruitment, contract types); portfolio and lease data; floor plans and building information; financial and cost data; existing surveys and occupancy studies; booking data; sensor and environmental data; technology and collaboration data; change history; policies and protocols.
- For each dataset, record: owner, format, date, collection method, known limitations and privacy status.
- Assess quality: is headcount FTE or people? Are floor plans as-built? Was the last survey pre- or post-hybrid? Does booking data reflect use or reservation?
- Reconcile contradictions between sources (HR headcount vs badge population vs Finance cost allocations) and document unexplained gaps.
- Confirm privacy conditions with the data protection officer: aggregation levels, consent requirements, retention limits.
- Produce the evidence inventory: what exists and is usable, what exists but is unreliable, what does not exist — and therefore which Diagnose methods are required.
- Datasets that answer a different question than claimed (booking data presented as occupancy data)
- Pre-pandemic evidence presented as current behaviour
- Headcount definitions that differ by 15–30% between HR, IT and Finance
- Surveys with response rates too low or too skewed to generalise
- Data owners protective of 'their' data — an early signal of governance friction
- Evidence inventory with quality and reliability assessment per dataset
- Data gap analysis mapped to Diagnose methods
- Privacy and consent register for all person-related data
- Reconciled baseline figures (headcount, capacity, cost) for later modelling
The gap analysis directly scopes Diagnose: if reliable occupancy data exists, the Occupancy Assessment validates rather than collects; if no reported-demand data exists, the Dynamics Assessment becomes mandatory. Baseline figures feed Taxonomy Spatial Benchmarking, Calibration, Sharing Ratio and every scenario. The privacy register governs all platform-enabled methods.
Deciding whether a dataset is good enough for a given decision is judgement, not procedure. A 40% survey response may suffice for hypothesis generation and be indefensible for a sharing ratio. That threshold depends on the decision at stake.
- Sending an unscoped request for 'all workplace data', producing noise instead of an inventory
- Accepting datasets without recording collection method and date
- Treating booking or badge data as behavioural truth
- Skipping privacy review because the data 'already exists internally'
- Failing to reconcile headcount before modelling — every downstream number inherits the error
A public-sector organisation reported 92% desk occupancy and requested expansion advice. The data request revealed the figure came from a booking system with auto-release disabled: reservations, not presence. Badge data suggested peak attendance near 55%. The project pivoted from expansion to a supply–demand recalibration — a conclusion reached from the inventory alone, before any fieldwork.