Risk Intelligence

Vacant buildings, undeclared: finding vacancy risk across a portfolio

5 minute read

Vacancy changes what you are underwriting. Fire is discovered later and protection is often disabled; a leak runs because nobody is there to notice it; the building attracts the kind of attention an occupied one does not. Policy wordings already recognise all of this through vacancy clauses and occupancy warranties, which work only when somebody knows the building is empty.

The declaration gap

Occupancy is declared at inception and rarely updated. Buildings fall vacant mid-term; schedules inherited through renewals carry stale occupancy codes; and in large portfolios nobody is re-verifying line by line. The claims team often discovers vacancy first — at the worst possible moment.

Detection at submission and across the book

Vacancy indicators can be surfaced automatically: signals from imagery, activity data and public records combine to flag locations that appear unoccupied. Applied at submission, the underwriter prices the risk that actually exists. Applied across a portfolio, it produces a worklist of locations to verify — turning an invisible exposure into a managed one.

What the submission says

LOC 0147 · Unit 4, Brayford Industrial EstateOccupancyOccupiedSchedule, tab “SOV”, line 147 — declared at inception, not revisited

Three signals, none of them in the submission

  • Imagery

    Overhead and street-level captures show an empty yard across several passes and boarded openings at ground level.

  • Activity

    The weekday pattern an occupied unit of this kind produces is present earlier in the record and absent since.

  • Public records

    The lease on the unit is recorded as ended, with no later occupier registered at the address.

Needs attention

A question, not a determination.

Three independent signals disagree with the declaration and none of them settles it. The location goes on a list to be verified. The occupancy on the record is left as the broker declared it until somebody checks, and nothing is re-rated on the strength of a signal.

The same check, down the whole schedule

  • LOC 0112Unit 2, Brayford EstateOccupiedNothing raised
  • LOC 012844 Calder Street, retail unitOccupiedQuestion raised
  • LOC 0131Warehouse C, Fenway EstateOccupiedNothing raised
  • LOC 0147Unit 4, Brayford EstateOccupiedQuestion raised
  • LOC 0160Depot, Marsh LaneOccupiedNothing raised

LOC 0147 is the location opened out above. Every line here is declared occupied; the two with a question against them are where the signals and the declaration part company.

One invented location on an invented schedule. The signals shown are the kinds the article names — imagery, activity and public records — not any particular source, and no finding is implied: three signals disagreeing with a declaration is a question for someone to verify, not a determination that a building is empty.

See it on a document you recognise

Bring three of your own submissions to a demo and watch them run in the session. Reading about extraction and watching it fail or hold on your own slip are different experiences.

Book a demo