How we compare — including where we lose.
The advantages worth talking about are the ones a competitor's data model cannot express — not the ones their team has not got round to. Everything below follows from the same root: they store a table of current-value project records, sold by sector, and we store an append-only log of claims with evidence.
This is a structural comparison, not a claim of present data-volume parity. GB, US, BR and IN have project-bearing adapters today. The BR customer write path remains ODbL-gated; IN remains gated on CEA reproduction permission and durable cross-quarter identity. No country is called paid-ready until its own production canaries and independent scorecard pass.
The five weaknesses they share
Shared because they follow from the same design, not because of any one company.
No evidence chain
You get a field. You do not get why. For lead generation that is an inconvenience; for diligence it means re-verifying from scratch, which is most of the work you were paying to avoid.
No history — so nobody can prove their own accuracy
They overwrite. When a date moves the old one is gone, so you cannot ask what they said in March, you cannot measure a developer’s track record, and they cannot measure their own. That last one is why no incumbent publishes accuracy.
Sector silos leave the intersections unserved
A data centre is at once a construction project, a grid interconnection, a power-demand signal and a water abstraction question. One vendor sells you one facet and another sells you the next. Neither joins them, because their products are the boundary.
No policy layer joined to assets
They report that a project exists, not why it exists. “These rules tighten — which projects lose their economics?” has no answer in a table of current values.
Portals and spreadsheets, not APIs
Investors and quants want to join this to their own data. Export to Excel is not that.
What we do not win on
Listed because a comparison that concedes nothing gets discounted entirely — including the parts that are true.
Contact depth
Theirs is decades old and ours starts empty. Human calls surface things no crawler reaches — an internal approval never published, a slip nobody announced. We do not claim parity and will not demo as though we had it.
Turnaround coverage
IIR’s is the best there is and mostly comes from phone calls. We model turnarounds as first-class objects so we accrue them properly, and we expect this to be a multi-year gap.
Analyst narrative
GlobalData ships written market context. We are not competing on prose; our equivalent is an explainable model and a policy landscape you can query.
Brand trust in procurement
Slow to earn, and no argument shortcuts it. Published accuracy is the fastest credible route we have — which is another reason the retained history matters.
The test we invite
Pick a project you know well. Ask us for a fact about it, and ask us what we believed about it three months ago. Then ask the same of anyone else you pay.
Our per-geography accuracy is published, never averaged into one figure, and scored against registers you can check yourself.