Agree what counts as a sale
Choose a paid-order or other completed-purchase event from the business system. Store the measurement definition with the evaluation window. A message, a positive reply, and a checkout link are intermediate steps, not revenue.
Decide how duplicate leads, existing customers, cancelled orders, and purchases outside the conversation are treated. Use the same definition in the baseline and pilot. Separate new customer acquisition from reorders so a retention workflow does not inflate acquisition performance.
Compare equivalent groups
A before-and-after result can reflect seasonality, promotions, inventory, or a changed offer. Record those conditions and compare similar eligible leads. Where practical, use a contemporaneous comparison group and agree assignment before the launch.
Keep sample size and the observation period visible. A few purchases cannot establish a reliable forecast. Wait long enough to account for your buying cycle and subsequent refunds before presenting a result as a business improvement.
Include the complete operating cost
List setup, service charges, communications usage, integration work, staff review, and any ongoing support costs. Distinguish a one-time launch expense from a recurring expense. Report contribution after costs rather than treating all order value as profit.
For staff capacity, observe time spent on the same task before and after launch. Include exception handling and quality review. Freed time is not automatically cash savings unless staffing or paid hours actually change.
Use a transparent illustrative calculation
Suppose an eligible group has 100 inquiries and 10 completed purchases in one period, then 12 purchases in a comparable period. The observed rates are 10% and 12%. That is a two percentage-point difference; it does not prove the AI caused the change.
If acquisition spend is $1,000 in each period, the acquisition-spend ratio is $100 per acquired customer for ten customers and about $83.33 for twelve. Those figures exclude service and channel costs and assume all purchases represent newly acquired customers. They are hypothetical arithmetic, not Crypture customer results.
Publish the limits alongside the outcome
A useful pilot report names the workflow, baseline, dates, eligible lead count, attribution rules, completed outcomes, refunds, staff time, and complete costs. Include issues that required human intervention. Obtain customer approval before sharing identifying details.
Crypture’s pilot measurement page explains this review process. No conversion percentage or CAC reduction is promised by the public examples. Use your own business records to decide whether to expand the workflow.
Plan your first workflow.
Bring your approved answers, platform names, and the outcome you want to measure.
Book a demoTalk to our AIExplore the sales workflow · Review pilot measurement · Check platform requirements