Fixed scope / Five business days

Know whether one AI workflow is safe to scale.

The AI Margin Decision Sprint connects complete cost to serve, successful outcomes, and attributable revenue or price to one live pricing, packaging, model, usage-limit, or scale decision.

01 / Admission

A sprint for a decision already approaching.

Fit

A B2B software product has paying customers.
One AI-supported workflow is in or near production.
AI, tool, infrastructure, or human-review cost materially affects cost to serve or gross margin.
A successful outcome can be defined and counted.
A pricing, packaging, provider, usage-limit, or scale decision is expected within roughly 60 days.
A technical sponsor can provide minimized aggregate evidence.

Not fit

Generic AI strategy or a broad transformation program.
A token dashboard without a named business decision.
Customer willingness-to-pay research or a full pricing redesign.
Metering, billing, continuous observability, or model-routing implementation.
Requests for guaranteed ROI, savings, or unsupported conclusions.

02 / Scope

Comparable evidence in. A bounded decision out.

One bounded scope

One workflow or feature.
One historical period.
One currency.
One successful-outcome definition.
One current revenue or price basis.
One named decision.

Timing

The working format is five business days after complete aggregate inputs are received. Missing, unmatched, or materially inconsistent evidence pauses the analysis rather than forcing a conclusion.

03 / Inputs

The smallest aggregate evidence needed.

Recognized revenue or the current price basis attributable to the bounded scope.
Model, API, or inference cost for the same period.
Materially attributable tools and infrastructure cost.
Human review or escalation cost when material.
Total outcome count and successful-outcome count.
Matched usage or volume for the same period.
The current pricing or packaging rule relevant to the decision.

04 / Deliverable

A decision pack, not another dashboard.

A reconciled cost-to-serve map.
Cost per successful outcome.
Revenue per successful outcome when attributable.
Contribution dollars and contribution-margin percentage when attributable.
Break-even revenue per successful outcome.
A sensitivity table for the three dominant variables.
The current state plus no more than three bounded alternatives.
Evidence gaps and assumptions.
One recommendation class: scale as-is, reprice or repackage, redesign the cost structure, measure before deciding, or stop or pause.
A 30-day action list tied to the recommendation.

If revenue cannot be attributed: cost-to-serve and break-even thresholds can still be reported, but contribution-margin conclusions remain unsupported.

05 / Data boundary

Aggregate evidence by default.

Provide

Only the aggregate cost, revenue or price, usage, volume, and outcome evidence necessary for the named decision.

Do not provide

No prompts, responses, credentials, raw customer records, raw production traces, or production access by default.

06 / Inquiry

Bring one live AI margin decision.

Share only a work email. Do not include operational or customer data.