The 1% Rollout SOP: Stop Swinging for the Fences With AI
This is a Small Business SOP: a free, do-this-today standard operating procedure pulled from a real operator on the Unscripted Small Business podcast. This one comes from Brett Schklar, founder of AI First Leadership, who has asked more than 1,800 CEOs how AI adoption is actually going inside their businesses and wrote AI Without the BS after hearing the same failures on repeat.
Most small businesses took one big swing at AI and missed. This is the opposite approach, and it costs an hour a week. Run it this week.
The Play in One Line
Give every person on your team one small job each week: make one part of their own work 1% better with AI, then say out loud what worked. Nobody owns AI centrally. Fifty-two weeks of that compounds to 68% more done.
Why this matters
MIT found that about 95% of AI pilots fail, and Brett watches the small business version of that every week. Owners get sold the flagship project, walk up to the plate, and swing for the fences. He borrows the fix from Moneyball, where the Oakland A’s stopped chasing home runs and optimized for one thing: getting every batter on base.
In Brett’s words: “And if everybody in a small business gets 1% better each week, you calculate that over 52 weeks and you get 68% productivity gains. 68% more done.” For an owner whose real constraint is time, 68% is not an improvement. It is a different company.
The 6 steps
1
Do not put anyone in charge of AI
No AI department. No handing it to IT. Brett is blunt about this one: “Don’t ever give AI strategy to the IT department if you’ve got an IT department. That’s a formula for failure.” There is no way to centralize the gains. You set the rule as the owner, then get out of the way.
2
Give each person one headache, not one tool
Ask every team member to name one recurring annoyance in their own week and spend an hour trying to solve it with AI. Their headache, their fix, their hour. That hour is the 1%. Handing everyone the same tool and hoping is what fails.
3
Start in the back office. Customer-facing goes last
Proposals, research, drafting, internal reporting. Brett’s rule is that anything touching a customer is the last place you deploy AI, not the first. It is where he has watched it done worst, and it is also where the biggest wins land once the rest is dialed in.
4
Set the guardrails before anybody starts
Write down what data may go in, and what may never go out without a person reading it first. Brett’s warning: publishing unverified AI output is not a style problem, it is “creating misinformation about yourself” under your own name. One page is enough.
5
Keep a human in the loop on anything that ships
Brett’s proof point is the RFP rebuild he ran for paving and concrete companies: “A process to respond to an RFP, which might have taken 30 to 40 hours in the past, only takes about two hours, of which an hour and a half is just reviewing what AI wrote.” AI researches and drafts. Judgment stays human.
6
Share what worked every week, in the open
Create somewhere people post what is working and what is not, as close to real time as possible, so other people see the impact. This is the step that makes it compound. One person’s 1% becomes everybody’s 1%.
The numbers behind the play
Brett has asked more than 1,800 CEOs how AI adoption is really going. These are the four numbers he keeps coming back to.
1%
better, per person, per week
52
weeks of compounding
68%
more done in a year
95%
of AI pilots fail, per MIT
Take the whole SOP with you
Two printable pages: the play, the six steps, and a fill-in checklist for your next team meeting. No email required.
