Base rates for strategy
The Decision Record
Every decision judged on this site, 229 of them, coded by what kind of decision it was and who made it. Before a strategy meeting, look up how often decisions like yours have worked, what separated those that did, and where the evidence runs out.
By kind of decision
Each kind has its own page with the cases, the base rate for each kind of decision-maker, and what separated the decisions that worked from those that did not.
| Kind of decision | Cases | Paid off | Mixed | Backfired | Too early | Paid off, of resolved |
|---|---|---|---|---|---|---|
| Technology bets Which architecture or technical approach should we bet on? | 43 | 27 | 8 | 7 | 1 | 64% |
| Opening or closing Should we open our platform, standard, code or data to others, or keep control? | 36 | 20 | 11 | 4 | 1 | 57% |
| Product design What should the product be, and how should it work? | 29 | 11 | 9 | 3 | 6 | 48% |
| Business model and pricing How should we charge, and who should pay? | 27 | 14 | 2 | 8 | 3 | 58% |
| New categories and markets Should we create a new category or enter a new market? | 21 | 13 | 3 | 5 | 0 | 62% |
| Distribution and positioning How do we reach customers, and where do we launch first? | 16 | 7 | 5 | 3 | 1 | 47% |
| Focus, exit and investment Where should we concentrate, and what should we stop? | 12 | 5 | 4 | 2 | 1 | 45% |
| Trust, safety and privacy How do we protect users and keep their trust? | 10 | 6 | 2 | 2 | 0 | 60% |
| Acquisitions and sales Should we buy this company, or sell ourselves? | 9 | 4 | 1 | 4 | 0 | 44% small n |
| Make, buy or partner Should we build this ourselves, buy it in, or depend on a partner? | 8 | 5 | 1 | 2 | 0 | 63% small n |
| Executing large change How should we roll out a large system or change? | 8 | 1 | 0 | 7 | 0 | 13% small n |
| Organisation and governance How should we organise, govern and run the company? | 6 | 3 | 2 | 1 | 0 | 50% small n |
| Public rules How should the state set rules for this market? | 4 | 0 | 2 | 0 | 2 | 0% small n |
By who decided
| Decision-maker | Cases | Paid off | Mixed | Backfired | Paid off, of resolved |
|---|---|---|---|---|---|
| Market leader The established leader in the market where the decision applied. | 93 | 39 | 18 | 27 | 46% |
| Challenger An established company that was not the leader in that market, often entering it from elsewhere. | 38 | 21 | 9 | 7 | 57% |
| Young company A start-up or a company in its first years. | 45 | 31 | 8 | 5 | 70% |
| Industry-wide A choice made by many firms at once, or by an industry as a whole. | 11 | 0 | 6 | 4 | 0% |
| Researchers and communities Researchers, individuals, open-source communities, standards bodies and consortia. | 26 | 18 | 6 | 1 | 72% |
| Public body Governments, regulators, armed forces and state-owned organisations. | 16 | 7 | 3 | 4 | 50% |
Market leaders’ decisions backfire far more often in this record than young companies’. Part of that is real: leaders have more to protect and more power to misuse. Part is selection: the young companies in these essays are the ones that survived to be written about, while the leaders’ failures are famous because the leaders were.
How well the process grades work
The site grades each decision twice where the evidence allows: once for how it turned out, and once for how good it was given what could be known at the time. If the two grades were truly independent, some strong decisions would have met bad luck. In this record, 44 of the 45 resolved decisions graded Strong paid off. That is too neat. It almost certainly means the process grades were influenced by knowledge of the outcome, the hindsight bias and outcome bias the site warns against elsewhere. Treat the process grades as the author’s considered judgement, not as an independent measure, until they can be assigned blind.
| Process | Paid off | Mixed | Backfired |
|---|---|---|---|
| Strong | 44 | 1 | 0 |
| Sound | 21 | 23 | 7 |
| Weak | 0 | 6 | 24 |
| Poor | 0 | 0 | 8 |
Method
- What is included
- Every decision in the decisions boxes of the essays and roads not taken, and the central decision of each annotated memo. Decide cases are left out so they are never spoiled.
- Kind
- Each decision is assigned one of 13 kinds by its main purpose. Some decisions belong to more than one, such as an acquisition that was also a technology bet; each is filed under the kind a decision-maker would most likely be weighing.
- Decision-maker
- Coded by position in the relevant market at the time: market leader, challenger, young company, industry-wide, researchers and communities, or public body.
- Outcome and process
- As judged in the source essay, memo or road not taken, which explain each judgement with its evidence. Memos carry an outcome but no process grade.
- Coding
- All codes were assigned by one person, the author. They have not yet been checked by a second coder, which is the main weakness of the record and the first thing a reader should test.
- Changes
- The record grows as essays are added, and corrections to codes or judgements are logged on the corrections page. Cite it with the version date.
Use the data
The whole record is free to download, with each decision’s source link, year, decision-maker, codes and judgements. Reuse with attribution is welcome under the terms on the copyright page.
Shrikant Kumar. “The Telos Decision Record.” Telos, version of 2026-11-03. https://shrikantk.com/record/