All articles

    The 70/30 and 70/20/10 Rules: How to Use AI in a Small Business Without Wasting a Dollar

    Small business owners keep paying for AI subscriptions that sit unused because they asked "which product should I buy" before asking how their time and their operation actually work. Two ratios, the 70/30 Rule for where your week goes and the 70/20/10 Rule for why the software is only 10 percent of implementation, show what to fix before spending another dollar.

    Mike Donaldson
    The 70/30 and 70/20/10 Rules: How to Use AI in a Small Business Without Wasting a Dollar

    The AI subscription is still active. You tried it for a few weeks, it felt like a second job stacked on top of the one you already had, and now it sits in a browser tab you never open. The money keeps going out and nothing comes back.

     

    If that sounds familiar, the software was never the problem. The problem was the question that got asked before the purchase. Owners trying to figure out how to use AI in a small business almost always start with "which product should I buy?" The better question is how you should be thinking about your time and your operation before you spend a dollar on anything. Two ratios answer that question, and neither one is about technology.

     

    The Vending Machine Mistake

     

    Owners treat AI like a vending machine. Money goes in, productivity is supposed to come out. That is not how it works.

     

    My background is lean manufacturing. Picture a production line with a problem somewhere in the middle. Parts back up, defects slip through, people work around it. Now drop a faster machine into that line. Nothing gets fixed. The faster machine feeds the same problem more material, more quickly, and spreads it through the whole operation.

     

    AI does the same thing to a business. AI doesn't fix a bad process, it just makes the mess move even faster. Before you look at technology, you have to look at your time, your process, and your business. That is where the first ratio comes in.

     

    The 70/30 Rule: Where Your Week Actually Goes

     

    Take an honest look at your workweek. Not the ideal week in your head, the actual one. Track it for two days and the pattern is hard to miss. Roughly 70 percent of what you do is work for the sake of work. Responding to email. Checking statuses. Chasing down answers. Updating presentation slides so someone else can present them.

     

    None of that is optional. It has to get done. It just does not require your judgment, your relationships, or the twenty years of experience you bring to the room. In lean terms it is waste, not because it does not matter, but because it consumes capacity that could go somewhere higher.

     

    The rule is simple. Give the 70 to AI. Protect the 30.

     

    The 30 percent is the work only a person can do well. Building trust with a client. Reading a situation and deciding the right course of action. Sitting across a table from someone instead of trading instant messages. AI will not carry the context of those moments. You do.

     

    Think about a restaurant kitchen. Most of what happens there is prep: chopping, portioning, checking inventory, setting up the station. All of it necessary, none of it needs the chef's palate. The chef earns their keep in the final 30, tasting, adjusting, and deciding whether a plate is ready to go out. That is what you are paying for when you sit down at the table. So the question for you is a direct one. Are you spending your day on the prep, or on the part where you shine?

     

    The 70/20/10 Rule: Why the Software Is Only Ten Percent

     

    Here is where the 70/30 Rule alone falls short. An owner hears it, buys a subscription, hands over the 70, and nothing changes. Same process, same mess, no benefit. The 70/20/10 Rule explains why. This ratio is about implementation, where the real work of making AI effective actually lives.

     

    Ten percent is the AI itself. Think of it as the engine in a car. You can have a great engine and still go nowhere without fuel and roads.

     

    Twenty percent is your data. Files, reports, customer records, everything AI needs to know in order to do the job right. If that information lives in someone's head, it is doing no good. It has to be digitized, organized, and stored in formats the system can actually find.

     

    Seventy percent is your process and your people. This is where AI works or it does not. Run a modern AI product on top of workflows that have not changed since 2005 and all you have done is put a fresh coat of paint on cracked walls. Your house must be in order before AI is fully effective.

     

    Change management gets treated as a soft skill. When it comes to AI, it is 70 percent of the equation. Hand every employee a Copilot license and, if nobody understands what it does for them or how it makes their day better, nobody uses it. Adoption is a people problem long before it is a technology problem.

     

    Both Rules in the Same Office

     

    Two scenarios show how the ratios work together.

     

    Picture a doctor's visit from a few years ago. The doctor has their back to you, typing into a computer, asking questions without much eye contact. That is the 70, and it is a poor use of a doctor. Now picture the same office with a dictation system that records the visit, with permission, and logs the notes automatically. The doctor spends the appointment looking at the patient and using judgment. The software was the 10. The organized records were the 20. Changing how the office runs its day and manages its data was the 70.

     

    Now a real estate agent. Most of the day goes to phone calls, email replies, and the same square footage questions over and over. Very little of it goes to negotiations or face to face relationship building. Hand the repetitive work to AI and the agent's calendar changes. They are at the showing, in the negotiation, and building the relationship that produces a referral two years from now. The agents who have not done this are not doing anything wrong. They simply have not built the foundation that lets them use their time differently.

     

    Where to Start This Week

     

    You do not need a big budget or a technology team. You need a different way of looking at your work. Four steps.

     

    1. Audit your toil. For two days, write down every task you do and mark it a 70 or a 30. Repetitive and doable by someone else is a 70. Requires your knowledge and experience is a 30. The results surprise almost everyone.

     

    2. Pick one platform and start. The specific model does not matter yet. Gemini, ChatGPT, Perplexity, any of them can handle this stage. Models matter later, for specific tasks and agents. Right now you are getting a feel for how AI helps you.

     

    3. Hand off one thing from your 70. Not everything. One task. Figure out how to give that single task to AI and get it working.

     

    4. Invest the recovered time in the 30. This is the step owners miss. When AI gives you an hour back, do not spend it on more 70 percent work. Take a client to lunch. Go network. Work on the next product or service. Put the hour where only you can add something.

     

    The software is ten percent of this, and the other ninety is a foundation you can start building before you ever open another subscription.