Reserved Bays to Revenue: Car Park Demand & Price Optimisation
Two connected models that rebalance bay allocation in real time and identify where pricing can safely rise, built on more than a million anonymised entry/exit records.
- ThinkCol built two connected models on over one million anonymised entry/exit records: a demand model and a pricing model.
- The demand model predicts monthly-parker presence every 30 minutes (R² 0.93 in pilot back-testing) and recommends the monthly-vs-hourly bay split in real time.
- The pricing model reads historical occupancy and peak-compression patterns around real price changes to flag which lot-days can safely absorb an increase — 89% cleared for a rise in testing, delivering +3.6% hourly revenue.
A major car-park operator allocated bays between monthly and hourly parkers using fixed manual rules. Around 21% of bays sat empty on a typical weekday while hourly drivers were turned away, and pricing stayed coarse, with a single flat rate per day-type.
ThinkCol built two connected models on more than one million anonymised entry/exit records. A per-vehicle presence model predicts monthly-parker demand every 30 minutes, achieving an R² of 0.93 in pilot back-testing, and recommends the monthly-vs-hourly bay split in real time. A pricing model reads occupancy curves and peak compression around real historical price changes to identify which lot-days can safely absorb an increase.
In testing, 89% of flagged lot-days cleared for a price rise, delivering a 3.6% lift in hourly revenue, with modelled revenue tracking official accounting to under 1% variance. Everything runs against the operator's own database, with staff sign-off required on every recommendation.
Why customise? It plugs into the operator's car-park system at licence-plate level and retrains daily — no off-the-shelf tool does that, and price sensitivity is read from real price changes in the data rather than guessed.
