We don't lead with artificial intelligence. Intelligence here means the data-to-insight pipeline — the four project vitals, plus three questions your current system almost certainly cannot answer.
Every finished project was priced at a margin and returned a different one. We show it as a bridge — planned margin on the left, realized margin on the right, and the movements between: scope changes billed, scope approved but never billed, cost movement against plan, and what remains unattributed. It reconciles by construction, not by argument, so the bars always add up.
A raw league table mostly ranks who was handed the hardest work. We measure each project against what that kind of work normally returns, so a team running difficult work isn't punished for running it. The cut is restricted to senior roles by default and enforced on the server, not hidden in the interface — a report that can damage behavior should be harder to reach than one that cannot.
| Team | Raw margin | Adjusted |
|---|---|---|
| Complex delivery | 9.1% | +6.2 |
| Standard delivery | 21.4% | +1.1 |
| New initiatives | 18.7% | −0.4 |
| Small & recurring work | 24.0% | −2.8 |
The lowest raw margin is the highest adjusted performer, because that team runs the hardest work. An unadjusted table would rank them last.
A number is only alarming relative to something. We show each period against your own eight-period baseline with a shaded normal range. Inside the band is ordinary, however uncomfortable it looks. Outside it is flagged, named, and worth a conversation. The band is calculated from your history, not an industry benchmark you had no part in producing.
The industry's standard answer to this kind of analysis is to export the data to a BI tool and build it yourself, which decays the moment its owner leaves. We took the other approach: every measure has one written definition, stored in the product, editable by you, and used by every view that quotes it.
The anchor comes from the recorded planned margin; realized margin is computed from actual value and cost. Every bridge step is a partition of the movement in value or cost, so the steps always sum to the realized figure.
Each project is scored against the typical return for its work type, then a team's adjusted figure is the average of those differences. It answers “did they beat what this kind of work normally returns” rather than “who had the easiest work.”
The normal range is calculated from your own trailing history. A period inside the band is ordinary; a period outside it is flagged and named. No industry benchmark is used.
This is the capability customers ask us for most, and it's on our near-term roadmap. Until it ships, the page is honest about it on screen: a figure that isn't yet reconciled to your ledger says so, and where something isn't computed it's marked as such rather than filled with a zero. We'd rather show you exactly where the gap is while we close it.
We'll run these cuts against a slice of your own completed projects — not our demo data. If the answers are dull, that's worth knowing early, and we'll say so.
Book a demo