Hey Reader,

Welcome to the 27th edition of the 3-2-1 (check out previous issues here).

One idea from me first, three links that go deeper on it, then two things that have nothing to do with work. Let's get into it.

---

1 Idea from Me

Measure the loop, not the task.

The year is 1983. 45,000 American 13-year olds are given a national assessment with this question:

"An army bus holds 36 soldiers. 1,128 soldiers are being bussed to their training site. How many buses are needed?"

70% did the math correctly.

29% wrote "31 remainder 12".

Only 23% answered 32 buses.

Cue panic.

The most famous flashpoint came on April 3, 1986 when teachers in Washington D.C. staged a picket line with signs that read "The Button's Nothin' 'Til the Brain's Trained" (my favorite) and "Beware: Premature Calculator Usage May Be Harmful to Your Child's Education".

The core fear at the time was skill erosion and technological dependency. Sound familiar?

What else can we learn from history?

In 1980, the National Council of Teachers of Mathematics made problem solving the first recommendation in its agenda for the decade. Research throughout the 80s began to show that students who used calculators during math lessons actually performed better on subsequent tests where calculators were forbidden. By offloading tedious arithmetic (multiplication tables anyone?), students could focus on the logic of the problem. Students consistently reported higher self-confidence and a more positive view of math. By 1994 the SAT allowed calculators and asked questions a calculator could not finish.

The catch was the research fiercely validated one major concern of the picket line teachers: premature use causes harm. In 2008, Vanderbilt University did a study that showed for students who had not yet memorized basic multiplication, using a calculator actively hindered their learning and caused lower test scores on future tests (I guess those tables were necessary). For those that had this prior knowledge, calculators became a tool to speed up their work.

Once again, sounds familiar, right?

The new technology led to better performance, but only after careful scaffolding and defining what good performance looked like (in the case of the calculator, using it to solve problems, not produce answers).

McKinsey put out a report this summer (third link below) that surveyed 750 leaders and found that individual productivity with AI does not predict organizational performance, with most leaders not seeing meaningful enterprise value yet. If you're in charge of AI enablement, you're probably seeing similar: task gains are real (answers produced), but outcomes and impact are lacking (problems solved). We are optimizing for the wrong unit of value.

The unit is the loop

Knowledge work happens in loops. You spot a problem, seek the views you're missing, work them into a recommended action, ship it to the people who decide, take the feedback, and repeat the loop. Summarizing a meeting is a task. Deciding what to do next is a loop.

With any AI roll out, it's first about deciding what tools to use, how to make them available, and how to get people confident using them. Then you need to show ROI, right? So you measure the low hanging fruit: the emails drafted, conversations summarized, research collected. You measure the time saved on tasks. And what you measure becomes what you optimize for.

But that's like measuring how quickly someone can get an answer from their calculator... 31 remainder 12.

What we really want is an answer to the problem, an outcome, an impact. We want people to run the loop and realize they need 32 buses to move those damn soldiers.

We coded 9 behaviors that run the loop

We've watched this loop run at clients for two years and coded nine behaviors that separated people who drove impact from people who handed problems up the chain. They group into three motions:

McKinsey's answer is to redesign the org from the top: roles, workflows, decision rights. This loop is the same fix at the altitude where your people actually work, and you don't need a transformation office to start.

An example of the loop from my work

This summer, three companies asked to license our method. Until now, we've delivered everything for clients, so this was new territory for me and the team.

I started with a friend who had bought a licensed sales methodology and rolled it out at his company. He gave me the buyer's view. Then I kicked off a Claude project to research how 12 companies like FranklinCovey license their IP. Claude cooked for ten minutes and gave me a meaty report with all the mechanics.

I shaped a business model that worked for us and put the page live to share with a potential buyer. Within 10 mins the first question came back: could they rotate people in and out of the seats over the year? That showed me where my view was weakest so I called another friend, who had spent years accountable for the commercial results of a licensing business at a coaching company.

He told me what it took to hold the line on seat rotation, but he also raised something I had not asked. The day a license is signed, proving its value becomes somebody's number one job, and nobody budgets for it. We had no owner for that.

Counted as a task, Claude saved me $1,000's in consulting spend. Counted as a loop, the gain was two gaps in my understanding Claude couldn't see.

Your move

Before you hand a task to AI this week, write down what problem you're trying to solve and what success looks like in two or three sentences.

Then ask yourself these questions:

• Seek - what view am I missing? Who can I speak to about it?
• Sense - what information am I considering that isn't relevant?
• Ship - what matters to those involved in making this decision?

Now you're in the loop.

And when your CEO asks for the AI productivity story, look for the loops your people closed and what changed because of them.

If you want to explore teaching a team to run the loop, hit reply. We offer licensing now.

---

3 Things for Work (in L&D)

Math Teachers Stage a Calculated Protest (Anndee Hochman, The Washington Post, Apr 4, 1986)

• The picket line, reported the next morning. About 20 teachers circled outside while 6,000 others attended the conference inside, chanting "Cal-cu-la-tors later, we shall not be mov-ed."

• 🏋 Effort ≈ 3 min read

Generative AI without guardrails can harm learning: Evidence from high school mathematics (Bastani and colleagues, PNAS, Jun 2025)

• The calculator study, rerun with AI. Nearly a thousand high school math students practiced with GPT-4, and their practice grades jumped 48%. Then access was taken away, and they scored 17% worse than students who never had it. A version built to give teacher-designed hints instead of answers erased the damage.

• 🏋 Effort ≈ 25 min read, 3 min for the abstract

From adoption to impact: Three horizons of AI transformation (De Smet, Goldstein, Price, Catlin, Pineault, Rainone and Almasi, McKinsey Quarterly, Jul 8, 2026)

• McKinsey's survey of 750 leaders and employees. What they found: individual productivity with technology is not predictive of organizational health and sustained performance.

• 🏋 Effort ≈ 15 min read

---

2 Things for Life

Paper2Audio (web, iOS, Android)

• A tool that turns PDFs into audio. Free for anyone who has a non-confidential PDF to read (like that McKinsey study or an industry report) but would prefer to listen to it.

• 🏋 Free to start

The Metta of Mister Rogers (Gayathri Narayanan, Lion's Roar)

• A meditation teacher reflects on what Fred Rogers actually practiced. She quotes him: "Love isn't a state of perfect caring. It is an active noun like struggle." It made me think about how I show up with my kids, especially when they're making what seem like unreasonable needs clear to me. My own struggle to meet those needs, and to be with them for who they are, is what really shows love.

• 🏋 Effort ≈ 7 min read

---

That's it for this week - enjoy your Sunday!

I'll be back in two weeks ✌️

Andrew

P.S. know anyone who would find this edition useful? Please share and encourage them to subscribe. It helps support my writing, so thank you!