SDI Intake - Software Defined Corporation

The SDI Intake v1.0 · 2026-08-05

Ten questions. One hour. Everything the score needs.

What this is

In one sentence, the Software Defined Index is simple: the measured value over the promised value, for each key performance indicator of your workflow, combined so that the weakest promise dominates. Everything else on the Methodology page — the four dimensions, the clips, the tail caps — exists to handle the edge cases of that one ratio: promises that must hold every time, promises nobody can measure, and behavior nobody promised.

That means the score is built from two lists you already have, even if nobody has written them down in one place: what you were promised, and which of those promises your workflow actually depends on. The vendor’s written claims determine what is testable. You determine what matters. We determine that neither list has been quietly trimmed.

The ten questions below are the entire intake. Answer them — roughly, in plain language — and the first engagement call becomes a working session instead of a discovery exercise. You do not need precise numbers to start; you need to know where the numbers live.

The ten questions

1. Walk us through the workflow.

Where does it start, where does it end, and what does “done and correct” mean? Why we ask: the SDI is deployment-scoped — we score the system inside your workflow, not the product in the abstract.

2. What were you sold?

Every document containing a promise: the contract, the SLA, the datasheet, the proposal deck, the sales email. What to bring: the documents themselves, not summaries. This builds the claims inventory — and the promises that turn out never to have been written down are themselves a finding.

3. Which broken promises cost you money, time, or risk?

Rank the top five to ten. Why we ask: this is KPI selection. Your ranking becomes the weighting of the score — we do not invent what matters to your business.

4. For each KPI: what exact number was claimed, and where is it written?

“Fast” is not a claim; “p50 under 50 ms” is. The rule: no document, no score — but an unwritten claim is logged and reported as one.

5. Must each promise hold every time, or on aggregate?

An uptime promise is aggregate; a latency promise is every-request. Why we ask: this routes each KPI to the right treatment — repeated-observation claims are scored on their spread and their shortfall, not just their average.

6. What deviation is tolerable before it hurts?

If the promise is 50 ms, does 55 ms hurt? Does 80? Why we ask: when the vendor’s spec is silent on tolerance, yours is the one that counts — and it is published in the finding, so nothing is scored against an invented threshold.

7. What telemetry exists today to check each KPI, and who owns it?

Dashboards, logs, billing exports, status pages — and the person with access. Why we ask: a promise nobody can measure scores against the vendor, not for them. This answer also scopes the engagement honestly before anyone commits to it.

8. What do you pay, per what unit of output?

Dollars per million requests, per seat, per terabyte — whatever the natural unit is. Why we ask: cost-per-unit is the one KPI almost every vendor claims and almost no buyer verifies.

9. What does your worst day look like?

Peak load, close of quarter, product launch, Black Friday. Why we ask: the measurement window must contain your worst day. A score computed on a calm week is the vendor’s score, not yours.

10. Will you sign the KPI list as complete?

The signed list becomes the specification of the audit itself. Why we ask: it stops the two games that corrupt scoring — cherry-picking claims to flatter a vendor already chosen, and quietly adding scope after the number looks inconvenient. Anything discovered later is a documented scope change, not a silent edit.

What happens next

From your answers we produce three artifacts before any measurement begins: the claims inventory (question 2), the signed KPI list with weights and tolerances (questions 3, 6, 10), and the measurement plan (questions 5, 7, 9). Then we measure, and the ratio does the rest. Every published score carries its scope, methodology version, and measurement window, per the Methodology page.

Ready when you are

Answer what you can, skip what you can’t — question 7 alone is worth the call.

Start the Intake