A2
6 to 8 weeks
AI Value Realisation
A2 - Workforce Productivity Assessment is a return figure for your AI licences, measured within your own estate rather than modelled from a benchmark. Your dashboard reports assisted hours, which the documentation reveals to be action counts times a fixed six minutes at a US labour rate, so your finance director discounts it and is right to. With the Workforce Productivity Assessment module, we measure what the tool did to the work: your own people record the baseline, then observe the period, and cash and time come back on separate lines.
Over six to eight weeks, I run that measurement, and you have the baseline in hand by week two, before the observation starts. You keep a return figure set against that baseline, and a cost per unit of work in the unit you already report.
Six months in, your telemetry shows the tool in daily use, and your people still rarely hand an hour back. They spend it somewhere else, because somebody checks a draft, a colleague picks up a second pass, and a manager absorbs a review nobody was doing before.
In a survey, all of that looks identical to work that genuinely vanished. So I separate the two by watching it, and by asking the colleague now receiving the work whether their own load moved.
Most organisations do not cut posts after a rollout like this. Your managers move people onto work that was already queued.
Somebody resigns, and nobody is hired to replace them, or the team absorbs a rising volume at the same headcount. Any of those can be the right decision, and none of them puts money back in your budget this year.
At EY, I ran the study behind a fifteen million dollar commitment to twenty thousand Microsoft 365 Copilot licences, covering four hundred and sixty people across global hubs. The headline was strong.
Ninety-two per cent intended to keep using it, at six hundred and twenty-five dollars per person per month, with a two-month payback. Then I put a different number in front of the board.
Only thirty-six per cent of output was good enough to use unedited. The tool was worth buying, and nobody could leave it unsupervised.
Both readings were true, and the sponsor got both.
| Phase | Activities |
|---|---|
| Scope | The teams, the measures, and the unit of work you already report |
| Baseline | Recording how long the work takes now, before anybody changes how they do it |
| Observation | Two to three weeks inside the work, following where the recovered time goes |
| Timing | The work timed across the teams, with the receiving colleague asked whether their own load moved |
| Quality | What the tool produced, set against what your people could use unedited |
| Reporting | Three costs deducted, cash split from time, a figure for each team |
I set the teams before collecting anything, because teams drawn after the data arrives are teams drawn to suit the data. I record the baseline before anybody logs in, since once the tool is in daily use,e the old record is gone and every later figure is a reconstruction.
The observation period makes the figure real, and cutting it short is the most common reason a study like this ends up worthless. You cannot see that failure in the finished report, which will still carry a confident percentage.
Where your renewal date leaves no room, I say so in the first hour, then scope something smaller and mark it clearly as indicative.
The three costs I deduct are training time, the hours spent correcting output that came back wrong, and the work of connecting the tool to anything.
| The problem | What the study settles |
|---|---|
| A renewal is coming, and nobody can prove what the last term returned | A return figure measured against a baseline your own people recorded |
| The only figure you hold came from a vendor dashboard or a survey | Your board reads a number taken from your own estate |
| Usage looks healthy in the telemetry, and your finance function still cannot find the savings | Cash savings and time savings reported separately, so finance knows which portion reaches next year’s budget |
| Somebody has promised your board a headcount reduction | A stated headcount position, tested before the term ends |
| The saving you hold is gross, with nothing deducted | Ramp, rework and integration deducted and itemised |
| You suspect whole teams hold seats they never open | A return figure for each team, so seats move towards the teams that use them |
This is one of two modules under AI Value Realisation, which carries the offer and the fee. Since the spend isn’t committed yet, AI Scale Assessment comes first.
If your problem is what the tools already in your estate returned, this module is for you.
If your problem is knowing what to scale and what to stop, then AI Scale Assessment is for you.
If the study has told you which teams returned something and the job is now to spread that, then AI Adoption and Change Management is for you.