The rules of Profit or Burn are not new ideas. Each one stands on research that anyone can read.
Every link opens the full text free of charge, with no sign-up. Where the journal charges for an article, the link opens the free version published by the authors or their university, and the line under the title says so.
The rule here: a verdict counts only results that can be tied to the action, and a result inside plus or minus 15 percent stays "keep digging".
eBay stopped its paid search ads in large experiments to see what it would lose. When the ads on its own brand name went off, 99.5 percent of the lost clicks came back through the free search results, and the measured return on its other search ads was negative.
663 advertising experiments at Facebook, each compared with methods that estimate the effect without an experiment. Even with more than 5 000 facts about each user, the better method showed effects about three to five times larger than the real ones.
25 large advertising experiments at retailers and brokerages in the United States, 2.8 million dollars of advertising in total. Sales differ so much from one person to the next that the typical range around the measured return was about 100 percentage points wide, so a small gap between two actions proves nothing.
The rule here: every action gets its own verdict, the final one waits 90 days, and brand bets are not ranked.
538 published estimates of how much advertising lifts sales for consumer brands, corrected for known biases such as weak results being published less often. The corrected effect is more than five times smaller than earlier reviews found, close to zero in the short term, and internet advertising comes out as the most effective medium.
Campaigns from 1998 to 2018 in the databank of the Institute of Practitioners in Advertising: business-to-business brands grew best with about 46 percent of the budget on brand building and 54 percent on activation, and brand effects pass the short-term effects only after six months. The authors call this tentative, because fewer than 50 of the cases are business-to-business.
1.9 million purchases traced to the discount codes of Instagram influencers at a large European company that sells directly to consumers, plus three field studies with paid posts. Influencers with few followers returned far more per euro than those with many, on all three measures of return, because their followers engage more.
The rule here: the weekly brief carries one reallocation to approve, and each quarter ends with a ranking of what paid and what did not.
More than 1 600 companies in the United States, from 1990 to 2005. The third that moved the most capital between their business units earned 30 percent higher shareholder returns a year than the third that moved the least, while a third of the businesses received almost exactly the same capital every year.
The same question tested on several thousand companies over 18 years. In most cases the companies that moved more capital between their business units performed better, so the common mistake is to move too little.
The rule here: forecasts stay hidden from each other until the deadline, and the case page shows the median of the people asked next to the owner's target.
Why forecasts from the people behind a project come out too high: optimism, and the wish to get the project approved. The fix, built on the work that won Daniel Kahneman the Nobel Prize in economics, is to start from the real results of comparable past projects and not from the plan.
144 people estimated facts, alone or after seeing what the others had answered. Seeing the other answers made the estimates more alike and the people more confident, but not more accurate.
How to use them: as ranges to hold a target against, not as targets. They are surveys of companies that chose to answer.
342 companies that sell software as a service or artificial intelligence products report their 2025 results. The median company earns back the cost of winning a customer in 16 months and the best quarter in 6 months or less, and every number can be filtered by deal size, growth rate and way of selling.
173 business-to-business technology companies report their marketing budgets. The median budget is 9 percent of revenue, companies growing faster than 20 percent a year spend 12 percent, and those fast growers get 46 percent of their sales pipeline from buyers who came to them.
200 software-as-a-service products report how many free users become paying customers. With a free trial that asks for no credit card, 4 to 6 percent of signups paying within six months is a normal result and 10 to 15 percent a strong one.
A January 2026 survey of sales and revenue leaders at more than 150 business-to-business software companies. It gives the share of leads that pass each step from first contact to signed contract, the cost of a sales opportunity, and the length of the sales cycle by deal size.
Most of the advertising research above was done on consumer products. Read it as a direction for a software company, not as numbers to copy. Links checked on 3 October 2026. How the rules are applied: definitions and how it is built. The same numbers as inputs you can move: the Lever Board.