How we calculate the numbers

The scoring and benchmarking engine is tested against thousands of real-world IFC models — buildings, bridges, roads, rail and more — pulled from actual projects, not synthetic test cases built to make the numbers look good. That's what every percentile and benchmark comparison on a report is measured against.

On top of that automated benchmark sits a manual expert-review programme: dozens of experienced BIM professionals have individually gone through hundreds of real files by hand, estimating what each one would actually take to model — broken down by geometry, information and structure, not just a single total. That panel is actively growing, and every new round of expert estimates gets folded back into the calibration, so the model keeps getting closer to how real practitioners actually see these files.

For buildings, the model is cross-checked against published academic benchmarks for the cost of poor data interoperability in construction and facilities management — including the widely cited NIST estimate of the annual cost of inadequate interoperability per square metre of floor area, used as a sanity check on the model's own output rather than taken on faith.

Underneath that, the savings figure is built up from five concrete, recurring operational mechanisms — the actual tasks a facilities team performs over an asset's life, each priced out separately rather than folded into one guess:

Future savings are discounted back to today's value using a standard 4% discount rate, with costs assumed to rise 2% a year — the same present-value approach used in infrastructure and real-estate investment appraisal, not a method invented for this model.

Every version of the model is tested leave-one-out against real expert consensus, specifically to check whether its stated range — not just its central estimate — actually contains the right answer as often as it claims to. That's a harder, more honest test than just checking whether the average looks close, and we keep re-running it every time the model is recalibrated.

Pro users can also tune the assumptions behind the valuation — like the average hourly rate — to match their own market, rather than being locked into one global default.