The top 10% of small businesses earn 2/3rds of all the profit. A shocking 70% earn almost nothing or lose money.
Those are the stark conclusions of a study of German SMBs that the great Steven Wilkinson cited in his recent newsletter, The Pitchfork Papers, and I believe largely mimic what you see in US SMBs as well.
A German researcher named Magdalena Pollok pulled the public financials of thousands of small and midsize companies (Germany change its laws to require this of private companies) and asked one question: who actually makes money? The top 10% earned a 17.1% operating margin. The average was 2.7%. Strip out that top tenth and everyone else (print shops, logistics firms, agencies) earned about 1%.
She asked the winners what they did differently and got the same three answers every time:
Relentless focus on a narrow customer and their problems.
Intentional, hands-on management of the numbers.
Obsessive, constant improvement.
Can you sum up the core lessons of every business book you’ve read any better? The owners who focus on those core tenants took the vast majority of the profits. Maybe we should pay attention?
Steven Wilkinson draws the obvious lesson: do the boring work, join the top 10%. He's right. In my view, AI is going to make this gap worse or better if you’re in the top 10%.
AI supercharges all three
Look at those three habits again, and ask which ones AI can help you with.
Relentless focus on a niche customer: No business owner I know likes this truth until they actually embrace it and see the wisdom. They always want to leave the door open. What if this other type of customer wants to buy? We’re turning down easy revenue. This is a trap set perfectly for fools. I see the results everyday in every pitch on every website. We help clients get outcomes! We’re results focused!
No one is buying that BS. Customers want to hear how their specific problem is solved by your specific product or solution. My new company Rustproof is a good example. We focus relentlessly on trade associations. Our competitors provide technology services for small and medium sized companies and organizations.
We’re bringing you a set of agents specifically tuned to help associations find, retain, manage, and monetize their membership. Do you want to buy that or AI services for small businesses and other organizations?
AI is a superpower here. It can find and synthesize data about your core customer that would take a career to build pre AI. The more focused data you feed it, the more nuanced and targeted the insights you glean can be.
Constant operational improvement. This means adapting all your company processes to the specific needs of that niche customer. That requires constant iteration, measuring, and testing. AI does that, cheaply, for anyone who asks. Think about capturing every customer interaction from your website to a conference to a transcribed sales or delivery call, etc. AI can help you make sense of the data and apply it to your business in no time. That data gathering and analysis work compounds.
What doesn’t compound is learning what processes work for 50 different kinds of companies. Your insights regress to the mean of the model as you expand your scope. In other words, ChatGPT knows about as much as you do and anyone with $20 can match you.
Managing the numbers. Greg Crabtree, Steven Wilkinson, Alan Miltz, and many others have been preaching this gospel for decades. Accounting and numbers are the language of business: the analysis, the forecasting, the "what does this P&L actually say"? If you don’t know the numbers, you are a blind person feeling your way in the dark.
Guess what? AI does that too. It can take you from asking, “What’s a P&L?” to an in depth understanding of how your labor cost impacts you profit and cashflow within a query or two.
Over time you can build a detailed understanding of how you can impact every lever on a financial statement for your kind of customer and service or product. You can use AI to do the type of financial analysis on your company and industry that would have required 100 experienced Wall Street analysts even 2-3 years ago.
Steven Wilkinson, who inspired this post, is one of the great business minds out there. He is using AI to help shape a new program called the Good Books Academy. It’s a financial-fluency program built for owners who freeze at the P&L, not for accountants. It's the fastest way to earn the one form of competence you still have to own yourself. He’s designing the program now and wants your feedback. It’s 5 minutes and 20 question:
[→ Take his survey / join the first cohort: INSERT LINK]
Compounding is the only moat left
The real message of this post is that the only moat left is the one that’s always been there: knowledge and data about your specific customer and how your business can serve them.
It’s axiomatic that the more you know about those two things, the better your company will perform. Profits follow performance.
AI is just an accelerant to that process.
A competitor can clone your product or service in a weekend and rent the same AI you do. However, he can’t clone three years of knowing exactly how your two hundred customers think, what they've already tried, and what finally made them move. That knowledge isn't in the model. It only comes from focus which means picking the smallest market worth owning and owning it completely.
The generic B2B marketing company or amazon reseller or accounting firm is going to get blown away by a new breed of ai-native generalists on the one side. They will compete away all your profits for basic products and services. Remember software is a commodity now. Analysis is a commodity. Execution is close behind.
Most existing companies can’t produce the software, analysis, and execution these new players will bring to the table. They can’t pick their heads up from getting and serving the next customer long enough to learn the tech. Unfortunately, the ai-native competitors won’t stop. There will be waves of them and each one more capable than the last. Each will take another chunk of your business.
One the other side will be the new 10%. My guess is AI will concentrate 2/3rds of the returns of the top 10% in the top 1-2%. These companies will have deep knowledge of very niche groups of customers. They will know seemingly everything and the generic players (ai-native and otherwise) will lose to them repeatedly as they drive specific and verifiable outcomes.
My message for those generic company owners is get out now while you still can. My message for the 10% is different. You have an unprecedented opportunity.
What goes up in value is everything AI can't manufacture: distribution and relationships and what you own that AI can help you manufacture: vertical knowledge. The narrower your niche, the more of that you own, and the more expensive and time consuming it becomes for anyone to outflank you.
Being a bit better at a lot of things is a losing hand. AI does "a bit better at a lot of things" for $20. The winning hand is the opposite: be the single most knowledgeable operator alive about one specific slice of one specific market, and use AI to compound that knowledge.
What I'd do
Pick the smallest segment you'd be proud to dominate; not the biggest you think you can defend. Write down every signal those companies throw off: who's hiring, merging, switching vendors, what are they complaining about, what is their bottleneck, etc. Start capturing it, and point AI at the pile every week to surface what changed. Let the machine do the work for you. Spend your scarce human time building the relationships and distribution while the AI compounds your knowledge.
Do that for a year and you won't be in the top 10% because you worked harder. You'll be there because you own a position no generic competitor can take from you.