Proposed lead: David Frise, with an independent accountant and the contractor’s finance approver.
Draft answer framing: Follow enquiries through to completed work and compare them with a credible estimate of what would have happened anyway. Report uncertainty rather than assigning every AI-handled booking to the system.
Why it matters to the buying decision: The desk should not approve a business case that counts duplicates, recovered callbacks, uncompleted appointments or revenue without delivery costs.
Put this question to the expert:
“Which jobs are genuinely additional, how do we know, and what costs or displaced work must be deducted before calling the pilot successful?”
A strong answer should cover this proposed measurement chain:
Unique eligible enquiry → qualified opportunity → confirmed booking → completed job → collected revenue and attributable delivery costs
Use the same measurement period throughout:
Net benefit = incremental completed-job contribution + evidenced cash costs avoided − implementation cost − operating cost − additional supervision and rework cost − contribution displaced elsewhere
This is the desk’s proposed evaluation framework, not an observed market result. The accountant should agree the contribution definition and prevent double counting.
Require:
- Existing callback recovery and repeat enquiries removed from claimed uplift.
- Comparable cohorts or a feasible controlled allocation, with differences disclosed.
- Capacity constraints and changes in advertising, weather, pricing or staffing recorded.
- Released staff hours reported separately unless converted into evidenced cash savings or additional productive work.
- Enough follow-up for cancellations, completion, refunds and collection to become visible.
Evidence to request: An anonymised enquiry-to-job reconciliation and the assumptions behind the comparison.
Expert quote: ____________________