- In McKinsey's 2026 survey, 37% of respondents attribute any EBIT impact to AI, essentially unchanged from a year earlier.
- Identified value (what could be saved) and delivered value (what was measured) must never be added together.
- Without a baseline measured before the change, there is no result to report.
- Hours given back = people x hours saved per task x times per month; multiply by an hourly rate agreed with finance.
- Hours given back are capacity; management decides whether they become savings, growth or better service.
On this page:Why many organisations cannot yet show valueIdentified versus delivered valueStart with a baselineThe sum: hours x rateHours are capacity, not yet savingsAllow for the dip and set review points
Why many organisations cannot yet show value
Use of AI is spreading faster than proof of its value. In McKinsey's 2026 global survey, 80% of respondents said AI had improved their own productivity, yet only 37% attributed any EBIT impact to AI, essentially unchanged from 2025. The small group of high performers is more likely to fundamentally redesign workflows (nearly three quarters, against one quarter of others) and twice as likely to have defined processes to measure the impact of AI initiatives.
A 2025 report from MIT's Project NANDA, based on interviews with 52 organisations, 153 survey responses and a review of over 300 public initiatives, put it more starkly: 95% of organisations were getting zero return, meaning no measurable profit-and-loss impact about six months after a pilot. The authors attribute the gap to approach rather than model quality, citing tools that do not learn, integrate poorly or do not match workflows. It is a preliminary, modest sample, but the message is consistent: value comes from changing and measuring the work, not from buying tools.
Which AI investment would you like to be able to defend with measured numbers? Book a free AI Value Briefing and we will look at it with you.
Identified versus delivered value
Identified value is an estimate: the hours a task could free if AI helped. Delivered value is measured: the hours actually freed once the new way of working is in place. Both are useful, but adding them together, or presenting an estimate as a result, misleads. Label every number as an estimate or as measured.
Start with a baseline
Before using AI on a task, time it on at least three real occasions, start to finish, including checking and corrections. Note how many people do it, how often, and how many errors or corrections it usually needs. Then time it the same way with AI, including the time spent reading and fixing the AI output (Field Guide 28).
The sum: hours x rate
Hours given back per month = people x hours saved per task x times per month. To put a value on it, multiply by an hourly rate agreed with your finance team, and add any outside spend you no longer need.
An illustrative example from Field Guide 28: a team of 6 each writes about 40 complaint replies a month, and timing shows 10 minutes saved per reply including checking. That is 6 x 40 x 10 minutes = 40 hours a month, with the same number of corrections. Then subtract what the change costs: licences, training time and the checking itself.
Hours are capacity, not yet savings
Freed hours become money only through a management decision: taking on more revenue work, absorbing growth without hiring, or reducing outside spend. Record what each block of freed hours is used for, so finance sees the whole picture.
Allow for the dip and set review points
Results often dip before they rise while people learn and processes change (Field Guide 30). Agree in advance which two measures you will track, one for speed and one for quality, and set review dates where a named person decides to continue, change course or stop.
What to do next
- Pick one frequent task done by several people and time it at least three times before AI is introduced.
- Agree an hourly rate and the labels "estimate" and "measured" with your finance team before reporting any number.
- Set three review dates with one speed measure and one quality measure, and name who decides at each.
Goes deeper in our free AI Field Guide: 28. Measuring the value: hours given back, 30. The AI J-curve: why results dip before they rise, 27. Evaluations: testing AI before you trust a change, 16. Choosing the right level: use what you have before you build. Ask for these guides.
Talk it through with us
You get a finance-ready view of value found versus value delivered, measured against a baseline, so decisions to scale or stop rest on evidence.
Book my AI Value BriefingA conversation, not a pitch. We will not ask for confidential data. How we handle your details.How this page was made: researched and drafted with AI assistance from the sources below, each opened and checked on the date shown, and reviewed against our editorial rules.
