What we measure, how we calculate it, and why it matters.

OpenBlue is the authoritative source for space inventory. Every metric that touches “what space exists” reconciles to OpenBlue first. Other sources (sensors, badge, reservations) are layered on top — never substituted. This is the single most important decision in the framework. It eliminates the “which number is right?” conversation.

Every metric above is implemented as a Postgres view in our staging environment. Below: the canonical definitions, executable and version-controlled.

-- Metric 1: Reliable seat supply (by building × floor × space type)
create or replace view v_reliable_seat_supply as
select
  s.building_id,
  s.floor,
  s.space_type,
  count(*)                                              as seats_in_inventory,
  count(*) filter (where s.status = 'IN_SERVICE')       as reliable_seats,
  count(*) filter (where s.status = 'OUT_OF_SERVICE')   as out_of_service,
  count(*) filter (where s.status = 'CAPITAL_PROJECT')  as in_capital_project
from openblue.spaces s
group by s.building_id, s.floor, s.space_type;

-- Metric 3: Capacity risk by building × day-of-week (rolling 8-week window)
create or replace view v_capacity_risk as
with windowed as (
  select *
  from v_daily_building_utilization
  where day >= current_date - interval '8 weeks'
)
select
  building_id,
  day_of_week,
  avg(utilization_pct)   as avg_utilization,
  max(utilization_pct)   as peak_utilization,
  case
    when max(utilization_pct) > 0.85 then 'HIGH'
    when max(utilization_pct) > 0.70 then 'MEDIUM'
    else                                  'LOW'
  end                    as risk_level,
  count(*)               as observations
from windowed
group by building_id, day_of_week;