Pick the number before the build.

One number, chosen before pricing

Replies within the hour. Hours returned per week. Enquiries answered out of hours. Records entered once. Pick one, write it in the scope document, and let it be the thing the project is judged on.

Secondary metrics can be watched. They cannot all be targets: five targets is no target.

A baseline you actually measured

Measure today’s value for two ordinary weeks before the build. Remembered baselines are always flattering to the change, which makes the eventual result unusable in a board meeting.

A remembered baseline always flatters the result.

If the number genuinely cannot be measured today, say so in writing and use a proxy you can measure from day one.

Put the whole cost on the other side

Build cost, running cost per unit at your real volume, your team’s review hours, and the change budget for the first eight weeks. Comparing a full benefit against a partial cost is how projects look brilliant and feel expensive.

Book the review before launch

Put a date in the diary eight weeks after go-live, with the number, the baseline and the cost in one page. Decide in advance what a miss means: tune, extend, or turn it off.

The willingness to write down “turn it off” is what makes the rest of the measurement credible.