Measuring the value: hours given back

What it is

Measuring value means finding out, with real numbers, whether AI saves time or improves work. The simplest measure is hours given back: the time a task took before, minus the time it takes now, multiplied by how many people do it and how often.

The basic sum is: people × hours saved per task × times per month = hours given back per month. For example (illustrative figures), 4 people who each save half an hour on a report written 8 times a month get back 16 hours a month.

Time is only half the story. If the faster work has more errors, the saving disappears in rework. So you also track quality: how many drafts need correcting, how many mistakes reach a customer.

Think of a new route to work. You do not know it is faster because it feels faster. You time a few journeys on the old route and a few on the new one, and you check you still arrive at the right place.

Why it matters

• Real numbers help managers decide where to invest and where to stop.
• Honest measurement builds trust; inflated claims destroy it the first time someone checks.
• Quality tracking can show where AI makes work faster but worse.
• Small, measured wins add up and are easier to repeat across a team.
• What the freed hours are used for is a management decision, and it needs a clear number to start from.

How to do it

1. Choose one task that is done often, by more than one person.
2. Time it before using AI: at least three real instances, start to finish, including checking and corrections.
3. Write down how many people do it and how often.
4. Use AI for the same task and time it the same way, including the time spent checking the AI output.
5. Count errors or corrections in both cases, using the same standard.
6. Calculate hours given back with the sum above, and label it "measured" only if both sides were timed.
7. Record everything in a simple tracking sheet and review it after a month.
8. Report the result with its limits: how many times you timed it, and what you did not measure.

Example

A customer service team drafts replies to written complaints (illustrative figures). Before measuring, a team member guessed AI would "save hours every day".

They timed it instead. Before: 5 replies averaged 24 minutes each, including checking. After: 5 replies averaged 14 minutes, including reading and correcting the AI draft. That is 10 minutes saved per reply. The team of 6 writes about 40 replies each per month. So 6 × 40 × 10 minutes = 2,400 minutes, or 40 hours a month. Corrections by the team leader stayed the same.

Their tracking sheet:

Task: [name]
People doing it: [number]
Times per month, per person: [number]
Minutes before (average of [n] timings): [number]
Minutes after, including checking (average of [n] timings): [number]
Errors or corrections before / after: [number] / [number]
Status: [estimate / measured]
Freed hours used for: [decided by manager]

Common mistakes

• Leaving out checking time. Reading and fixing AI output is part of the task. Always include it.
• Mixing estimates and measurements. A guess presented as a result misleads. Label each number clearly.
• Timing only the best case. One fast example is not an average. Time several ordinary ones.
• Ignoring quality. Faster but worse is not a saving. Track corrections and complaints too.
• Counting freed hours as cost savings. Hours given back are capacity. Whether they become cost savings, more work or better work is decided by management.

Try it this week

1. Pick one task you do at least weekly.
2. Time it once the normal way, from start to final check.
3. Do a similar task with your approved assistant, using made-up or allowed data only, and time it including checking.
4. Note any corrections you made in each case.
5. Fill in the tracking sheet above, marked "estimate" until you have at least three timings each way.

Going further

• Use the median (the middle value) rather than the average if a few timings are unusually long or short.
• Track a "first-time-right" rate: the share of AI drafts that needed no real changes.
• Re-measure after a few months; time saved often grows as prompts improve, or shrinks if checking becomes sloppy.
• Measure across a team with a shared sheet so you can compare like with like.
• Add a short note on what people did with the time, once management has decided.

Key words

• Hours given back: time saved on a task, multiplied by people and frequency.
• Measured result: a number based on timing the task before and after.
• Estimate: a number based on a guess or a single example, not yet measured.
• Error rate: how often work needs correcting or contains mistakes.
• Tracking sheet: a simple shared table that records each task's numbers over time.

Examples and figures in this guide are illustrative. AI tools change often: check what your organisation's approved tools offer today. Use approved tools only, and a person checks and decides.

Prefer to print or share it? Download this guide as a PDF (241 KB), or the complete AI Field Guide (1.7 MB).