What this book is
Which of your goals would you give up first? If you can't answer in ten seconds, someone is already answering for you: fatigue, haste and whoever shouts loudest. It is guessing, with extra steps, and the bill for it comes at the end of the month, in the bank balance.
Steer, Don't Guess is one method, about 250 pages, for owners, managers and freelancers who are the only thing holding it together. It is not a motivational book and it does not say "delegate more". It gives mechanics: goals in an order you chose, your real position in numbers rather than feelings, red lines drawn before the crisis, and a hard wall between the levers you can pull and the weather you can only watch.
What you will learn
- The refusal test. The one question that turns a wish list into an order, so a crisis never chooses for you: if circumstances force you to drop goals one by one, which goes first, and which do you defend to the last?
- One number, one source. How a coffee-shop owner who was "doing fine" found his real margin once the write-offs were counted.
- Red lines before the crisis. Three zones, normal, acceptable and emergency, that tell you when to keep going and when to drop everything. And a fourth, "undefined", for the moment you have no data and must not pretend it is green.
- Levers versus weather. What you can actually pull (prices, the menu, terms, hours) and what you can only account for (the season, a competitor, your own lack of time), so you stop spending energy on the second kind.
- Three ways to control. By plan, by deviation, by forecast, and why the first two always leave you reacting too late.
- Why good plans fail. Three offsets to check before acting: a badly defined goal, a wrong starting state, a dying environment.
- The backward move. Planning from the goal back to this week, with a table that answers "what must be true one step before the result".
- The Red Card. A one-page protocol for the day everything burns.
Who it is for, and who it is not for
For owners of small businesses, managers, freelancers and families who are tired of being the only thing holding it together, and who are willing to read their numbers. The method works for a thirteen-year-old saving for a bike and for a two-chair dental practice, because every chapter is written on three levels: a simple everyday example, a practical business case, and a formal level with definitions.
Not for anyone shopping for a guarantee. Forecasts can be wrong. Data can be incomplete. The method does not replace knowing your trade. It replaces guessing.
Your AI assistant, working by the method
The book includes the model's canon and a ready operating prompt. Attach them to any capable AI assistant and it stops giving generic advice: it works with your numbers, in a fixed order of analysis, and marks missing data as missing instead of filling the gap with a guess.
How much difference does that make? A blind test put it on ten realistic business crises, from a food truck 72 hours from losing its permit to a freelancer six days from rent. On its own, the AI slipped something made up into seven answers out of ten, and nothing marked which ones. With the method: one answer in 120. In the same blind reviews, that is more than 100 invented facts against one, and the one was copied from an example inside the prompt.
The weakest answer with the method scored higher than the best answer without it. The full record is public, with a permanent DOI: check every number on Zenodo. Don't believe it. Run it.
What early readers say
Six early reviews agree on the same points, including the critical ones. "Most management books give advice. This one gives mechanics." The refusal test is the idea reviewers quote most. The distinction between levers and conditions is called the most liberating in the book. Every reviewer notes that the author promises nothing, and one adds that this caution is what makes the approach credible.
What they criticise: the terminology is dense in places (goal vectors, state vectors, zones, three offsets), some examples repeat, and readers fluent in OKR and PDCA will find parts familiar. The novelty, as one review puts it, is in how the ideas are arranged and applied, not in a brand-new principle. All reviews →
Inside the book, chapter by chapter
Four parts. Each chapter below is listed with the situation it answers.
