Intelligence / Benchmarking
Comparison worth trusting, or none at all
Comparative context for cycle time, severity and closure patterns — with an explicit statement of what the comparison set is.
An unlabeled benchmark is worse than no benchmark. “Industry average” against an unnamed population has led more operations to the wrong conclusion than any missing report.
Ask what the comparison set isWhat exactly are we being compared against?
That is the only question that matters, and the one most benchmarking products answer vaguely. Any comparison we surface states its population and its limits — and where the honest answer is that the set is too small or too unlike your book to be meaningful, we say so rather than drawing the line anyway.
Population stated
How comparison is handled
- 01
Your own history first
The most reliable comparison for a book is its own past. Period-over-period movement in your data needs no external population at all.
- 02
Comparable segmentation
Line, jurisdiction and severity band, because a coastal homeowners book and an urban commercial auto book are not usefully averaged together.
- 03
The set is named
Where an external comparison appears, its population and its caveats are stated on the surface, not in a footnote.
- 04
Silence when the set is thin
If the comparison population is too small or too dissimilar, no line is drawn. An absent benchmark is more useful than a misleading one.
Reading a benchmark
Questions to ask of any comparative figure
Ask these of ours and of every other vendor’s. The pattern of which questions get specific answers is informative on its own.
| Question | Why it matters |
|---|---|
| How many organizations are in the set? | A dozen books is an anecdote, not a benchmark |
| Are they like us? | Line and geography mix changes every operational metric |
| What period? | A pre-catastrophe baseline flatters everybody |
| Is our own data in it? | Self-inclusion quietly flattens the comparison |
| What is excluded? | Dropped outliers are where the interesting claims live |
We are early, and the set is small
This is the most honest thing on the page: a young platform does not have a large comparison population, and pretending otherwise would be the easiest number to invent.
Small population today
Our comparison set is early-stage. Where it is too thin to be meaningful, you will see your own history instead of a fabricated industry line.
No purchased benchmarks
We are not reselling a third-party dataset as our own insight. If you want an external benchmark, buy it knowingly.
Comparison is not a target
Being faster than a benchmark is not the same as being fast enough for your clients, whose SLA is the figure that actually binds.