The Wrong Ratio: What "80% AI" Misses About Growth

This week an agency announced it now runs on “80% AI, 20% human.” A software company gave its AI a job title and a login. Both are measuring the wrong thing. The number that predicts growth is not how much of the work is machine. It is how much of the leader’s week is spent with people.

Describing your business as a ratio of machine to human decides in advance what the human is for.

Customers are created in conversations, not in output.

Durable companies use AI to move hours toward people, not to shrink the human share.

The Percentage Problem

Two announcements landed in the same week.

On August 31, Optimizely launched “Virtual Teammates”: AI personas with titles like Chief of Staff and Marketing Analyst, each with its own login, its own permissions, and a standing slot on the org chart. On September 3, a St. Louis agency launched an AI marketing platform for contractors and clinics and described its operating model as 80% AI, 20% human strategy.

Both are honest attempts to answer a question every leader is now being asked: how much of this work should a machine do?

It is the wrong question.

When you describe your business as a ratio of machine to human, you have already decided what the human is for. The human is the remainder. The reviewer. The twenty percent that signs off before the campaign ships.

That framing has a predictable life cycle. A percentage is a cost. Costs get optimized. Whatever the human share is this year, the pressure next year is toward less of it. The people who built the 80/20 model will be asked, reasonably, why it isn’t 90/10.

None of this is about whether the tools work. They do, and we use them every day. It is about what you are choosing to count.

What Actually Grows a Company

Peter Drucker said the purpose of a business is to create a customer. He did not say the purpose of a business is to produce output efficiently.

Customers are created in conversations. A founder who spends an afternoon with three clients learns what the market will pay for before any dashboard can show it. A leader who sits with a struggling team member keeps a person the company cannot afford to lose. A phone call to a referral partner opens a channel no ad budget could buy.

None of that shows up in an efficiency metric. All of it shows up in revenue, eventually, and in trust, immediately.

The most valuable thing a leader can do this quarter is often not to adopt a new tool or shave hours off a process. It is to spend more of their time connecting with the people the business depends on.

That is what the AI is for. Not to replace the human share, but to clear the ground so the human can do the one thing no machine can: build a relationship that someone chooses to keep.

The Leader’s Time Ledger

If the ratio to watch is not machine to human, what is it?

Try this for one week. Categorize every hour of your calendar into three buckets.

Leading. Time spent face to face, or voice to voice, with a customer, a prospect, a team member, or a partner, where the purpose is the relationship itself: understanding, deciding together, building trust.

Deciding. Time spent making a call only you can make, with the context prepared in advance.

Operating. Everything else. Status updates. Reformatting. Chasing. Reconciling. Reviewing work that should have been right the first time.

Most founders we work with find that Leading is a small slice of the week and Operating is more than half. That is the real ratio, and it is the one worth changing.

The job of AI in your company, and the job of any partner you bring in, is to move hours from the third bucket into the first. If a tool or a team makes you more efficient at operating but does not give you a single extra hour with a customer, it has not helped you grow. It has helped you stay busy at a lower cost.

A Teammate Has a Stake

One more thing about job titles.

A teammate is someone who shares an outcome with you. Who notices when you are stretched thin. Who learns something at a client dinner and brings it back. Who can be trusted with the parts of the work that are not in the task description.

A scheduled process with a login is a good thing to have. It is not a colleague. Calling it one blurs the exact line a leader most needs to keep sharp: the line between the work that runs and the work that requires a person who cares about the result.

We build systems so the human experts on our team can spend their hours where judgment and relationships live. That is also what we want for the leaders we serve.

Simplify the operation so you have time to lead.

From Insight to Action

• Run the Leader’s Time Ledger for one week. Write the three percentages at the top of your notebook.

• Pick the one Operating task that consumes the most hours and ask what it would take to remove it entirely, not speed it up.

• Book two customer conversations for next week with no agenda other than listening.

• Ask your team which parts of their week feel like operating and which feel like leading. Compare the answers to your own.

• Before adopting any AI tool, write down which hour of your week it will give back and what you will do with that hour.

• Retire the phrase “AI does X percent of the work.” Replace it with “AI gives our people X hours a week for customers.”

  
  

A company that measures AI by the percentage defaults to substitution. A company that measures it by the leader’s time defaults to growth.