Every business has a productivity number it watches. It might be revenue per employee, billable hours per person or jobs completed per crew. Your team knows the definition, someone pulls it every month, and you have probably explained it to your bank at least once.
Most of these numbers have the same weakness. Workforce Intelligence points out that a firm can lay off 20% of its staff on a Friday and see revenue per employee rise on Monday. The business didn’t get better over the weekend. The denominator got smaller. Price increases and reclassified contractors move the number in the same way.
“Headcount is the metric that leaders rely on the most, but tells them the least. It’s a vanity metric.”
Keith Falk, quoted in Workforce Intelligence, Chapter 2
The book’s answer is to multiply your existing productivity number by a second measure that layoffs and price rises can’t move. This post explains that measure, shows the formula with a worked example, and covers the three ways it breaks in a business of twenty people.
The AI Amplification Factor
The second measure comes from a single survey question: when you use AI at work, which best describes you?
| Answer | What it means | Weight |
|---|---|---|
| Transactional | I ask AI a question and use the answer. One prompt, one output | 0.25 |
| Collaborative | I work through problems with AI over several exchanges | 0.65 |
| Co-intelligence | AI is part of how our team works | 1.00 |
Multiply the share of your team in each group by its weight and add the results. That gives you the AI Amplification Factor, which the book calls “A”, on a scale from 0 to 1.
This number only rises when people change how they work with AI. A layoff doesn’t move it, because the people who remain answer the question the same way they did before. Neither does a price increase or a restructure.
The book links this to Microsoft’s 2025 Work Trend Index, which surveyed more than 30,000 knowledge workers and found them split roughly in half between people who treat AI as a command tool and people who treat it as a thought partner.
One note before you use it. Chapter 3 of the book describes a five-tier version of this survey, while Figure 3.1 and Appendix A use the three-tier version above. I recommend the three-tier version because it is simpler and it is the one the book’s own worked example uses.
The Formula and a Worked Example
The Workforce Intelligence Score is:
WI Score = (Output × A) ÷ Workforce Cost
Output is your existing productivity measure at business level, usually revenue. Workforce Cost is total annual salaries plus total annual AI spend.
Here is an example for a twenty-person professional services firm:
| Input | Value |
|---|---|
| Revenue | $2,400,000 |
| Annual salaries | $1,600,000 |
| Annual AI spend | $12,000 |
| Survey result | 55% transactional, 35% collaborative, 10% co-intelligence |
| A | (0.55 × 0.25) + (0.35 × 0.65) + (0.10 × 1.00) = 0.465 |
| WI Score | ($2,400,000 × 0.465) ÷ $1,612,000 = 0.69 |
Three Ways the Score Breaks at Twenty Employees
The score was designed for a firm with a finance team and thousands of survey responses. In a small business, three things go wrong.
A survey of twenty people is too small. One person changing their answer can move A by up to 0.04. A good or bad week for one employee changes your score, so don’t try to manage it to two decimal places.
One hire swings the result. Adding a $70,000 employee to a firm with $1.6 million in salaries raises Workforce Cost by about 4.3%, and the score falls by a similar amount even if the business became more capable. In a 6,000-person firm that hire is invisible. In yours it looks like a decline.
The bands will mislead you. The book reads the score in four bands: Lagging, Developing, Performing and Leading. On those bands, the firm in the example scores 0.69, which is Performing. Yet only 10% of its people use AI as a true collaborator. The score is high mainly because revenue is one and a half times the salary bill. A business with a different cost structure could be making faster progress with AI and still land in a lower band.
How to Use the Score Anyway
Run the survey once a year, in the same month, and record the result. Then ask one question: did it go up? A rising number means the business is learning to work differently. A flat number means it is mostly getting busier. The band label isn’t a grade.
Alongside the score, keep a capacity ledger. This idea is mine rather than the book’s: a one-page list, kept by you, of what the business can do now that it couldn’t do a year ago without adding hours. For example:
- We now send a personal check-in to every client each month.
- We now return quotes within four hours instead of four days.
- We now get a new hire to full productivity in two weeks instead of six.
The ledger is harder to argue with than a ratio, because each item either happens now or it doesn’t. Use the score to check direction and the ledger to see results.
The 2027 AI Transformation Worksheet includes a ready-made page for this, called What Has Actually Changed, with columns for what you can now do, since when, and who owns it.
The J-Curve
The most useful idea in this part of the book is the Productivity J-Curve, described by economists Erik Brynjolfsson, Daniel Rock and Chad Syverson in a 2021 paper. When a business invests in a major new technology, measured productivity usually dips before it rises, so the line on a chart looks like a J.
The dip happens because the technology needs supporting work before it pays off: redesigning processes, retraining people and changing who does what. That work costs time and money now, and the benefit arrives later.
For a small business owner, this creates a specific risk. You see the numbers every week and you absorb the cost directly, so you are the person most likely to stop the program in month five.
The AI Amplification Factor is what tells you whether to keep going:
- If A is flat or falling while productivity dips, treat it as a warning. People aren’t changing how they work, and patience alone won’t fix it.
- If A is rising while productivity dips, you are most likely in the J-Curve. Behavior is changing, and the results usually follow.
Your Next Steps
Send the survey question to your team this week. Keep it anonymous, with three options, and you will have a result within a day.
Start your capacity ledger with three things the business can do now that it couldn’t do this time last year. If you can’t think of three, that tells you something too.
Put a reminder in your calendar for the same month next year, and run the survey again.
If you’d like help setting up the survey or reading the results, the Focus HR team can walk you through it.
Talk to the Focus HR teamSee all 8 posts in The 2027 AI Action Plan for Small Business
About the author: Clint Parry, MBA, SHRM-SCP is a Senior Business Consultant at Focus HR, where he helps Arizona small businesses with HR strategy, compliance and workforce planning. Focus HR is powered by OneDigital. This post is part of The 2027 AI Action Plan for Small Business series.


