Methodology guide · 02
Two sources build every discount curve, with a third opening up as customers opt in. We keep them separate in the model so any figure traces back to its origin, and we are explicit about the one we will never name.
Commercial procurement and benchmark data sources covering transacted software pricing at scale. This tier gives the curve its breadth: it sees far more deals, across far more buyers, than any single advisory firm ever could.
These sources are confidential. We do not name them now and we will not name them later, including under NDA or as part of a vendor security review. We would rather tell you that at the start than have you build a diligence step around an answer that is not coming. What we do share, in full, is the methodology: which curve a vendor sits on, the bands it was calibrated against, and the confidence we attach to each.
What we have seen directly, advising on these negotiations. The quotes vendors put on the table, what they moved to under pressure, which concessions were real and which were theatre, and where each vendor's true discount ceiling sits once the quarter is closing.
This is the tier that separates us from a data vendor. We are not only reading the market, we have been sitting on the buyer's side of it. It is also the tier that catches a pricing regime change first, because we feel it in a live negotiation months before it shows up in aggregate data.
Everything is normalised and anonymised into the curve. No individual engagement, client or deal is ever surfaced, and no client is identifiable from any output.
Opt in only. Nothing you upload contributes to anything unless you switch it on, and the switch is off by default for every organisation.
If you do opt in, records are stripped of entity identifiers and surface only through a k-anonymous aggregate with a floor of five contributing organisations. Below that floor the aggregate does not exist, not even as a count, so no output can ever be narrowed back to one contributor.
The pool is empty today. We built the consent mechanism before we started asking, which we think is the right order.
For each vendor, the research record fixes six things. This is the working definition of a benchmark on our side.
| Element | What it fixes |
|---|---|
| Product and unit | Exactly what is being benchmarked and on what unit, for example per user per year, per core, per committed dollar, per FUE. |
| Discount curve | Low, median and high discount off list, stated separately for each deal size band. A $200k buyer and a $6m buyer are not on the same curve and are never compared. |
| Maximum credible | The ceiling beyond which a claimed discount is either a different scope, a one-off credit or a story. Stops the model chasing outliers. |
| The two levers | The two structural terms that actually move price for that vendor, such as term length with a renewal cap, or a competitive displacement in play. Encoded into the model, not just described. |
| Confidence | How well anchored the curve is, and specifically where it stops being measured and starts being extrapolated. |
| Sources | Which tiers contributed, and at what weight. |
A benchmark that ignores who you are is a number without a claim behind it. Every reference set is cut three ways before your deal is placed against it.
This is the part most procurement and security teams want in writing, so we keep it short and absolute.
| What you do | What happens to it |
|---|---|
| You upload a contract | It is processed for your use only, inside your own tenant. It does not enter the benchmark set, it does not train a shared model, and no other customer can see it or anything derived from it. |
| You run a benchmark | Your inputs place your deal against the reference set. They are stored against your account so you can revisit the scenario, and are not pooled. |
| You opt in to contribute | Records are stripped of entity identifiers and enter a pool that only ever emits aggregates with at least five contributing organisations behind them. You can withdraw consent. |
| You do nothing | The default. Nothing of yours contributes to anything. |
The reciprocity rule. Where a benchmark surface shows anonymised contributed outcomes, an organisation that does not contribute does not see them. It is not a paywall, it is symmetry: the pool exists because contributors built it.