A contractor's win rate was falling and cost overruns rising — without knowing which bids were the problem.

A construction contractor needed to benchmark bids against real historical cost data to stop losing the right work at the wrong price.

Client

Construction Contractor

Duration

4 months

Industry

Construction

+20%Win Rate on Revised Bids

~50%Bids Revised by Engine

−20%Loss Rate Improvement

−15%Cost Overrun Reduction Target

Client problem

In their words

The contractor's loss rate had risen 20% and project cost overruns by 15%, with no visibility into whether bids were priced too high, too low, or misjudged on scope. Pricing decisions were made without external benchmarks, and no system connected past bid outcomes to future estimates.

What ZapSight built

One operating system around the risk pattern

Ingested all historical bids along with win/loss outcomes to build a proprietary pricing benchmark
Created an external bid database from public procurement data for market comparison
Built a cost-variance model mapping expected versus actual costs across all completed projects
Deployed a Bid Assist Engine that evaluates each bid and flags areas of under- or over-pricing before submission

Technologies Used

PythonLangGraphPostgreSQLPublic Data APIsXGBoost