Bezos's Three Tests for a New Bet
Big idea, right owner, legitimate plan — an idea needs all three or it dies
- Difficulty
- Moderate
- Time to result
- ~weeks to results
- Steps
- 4
- Confidence
- 93%
Asked directly what his criteria were for investing in something new, Bezos gave McAllister three tests: is it a big idea, is it something we should be doing, and is there a legitimate plan to succeed. All three must hold. The framework's power is in the AND — most rejected ideas fail exactly one test, and teams routinely mistake passing two for passing.
Origin
Jeff Bezos, given to Ian McAllister over lunch around 2008 at Amazon. McAllister maps the third test onto the FAQ section of Amazon's working backwards PR/FAQ, which he says is the part of the process least written about publicly.
Core principles
- 01Size, fit, and feasibility are independent axes — passing two is failing.
- 02'Should we be doing this' is a strategic-fit question, not a capability question. Amazon could learn to extract shale oil; it shouldn't.
- 03The legitimate-plan test is stage-appropriate, not exhaustive — it asks whether you've been thoughtful enough for where the idea is.
- 04The FAQ is where the third test gets answered: components, finances, key technical hurdles.
How to run it
- 1
Test one — is it a big idea?
Assess whether the opportunity, if fully realised, is materially large. Small-but-good ideas fail here and that's fine — they just don't warrant a new bet.
Pro tip Before answering, deliberately try to expand the idea. Ask whether it could be bigger than the version you were handed.
- 2
Test two — is it something WE should be doing?
Bezos's own illustration: a new way to extract oil from shale might be a genuinely big idea, and it is definitively not something Amazon should be doing. Strategic fit is a hard filter, independent of size.
Pro tip Ask what unfair advantage your company brings. If the answer is 'we have money and engineers', that's not fit.
Watch out Big companies fail this test constantly because capability masquerades as fit.
- 3
Test three — is there a legitimate plan to succeed?
Work through the internal components, the finances, and the key technical hurdles. This is the FAQ. The bar is thoughtfulness proportional to the stage, not certainty.
Pro tip Passing this test is itself a trust-building act — it shows you've earned the right to the resources you're asking for.
Watch out A compelling press release plus a hand-waved FAQ is the most common failure mode. The big idea seduces; the plan is what actually funds it.
- 4
Require all three before greenlighting
Use the working backwards review as the gate where all three are interrogated together. Only pass ideas that clear every test.
In the wild
Bezos's own worked example: suppose you invent a new way to extract oil from shale. Is it a big idea? Yes, probably enormous. Is it something Amazon should be doing? No. Test two kills it regardless of how well it does on test one.
→ The example isolates strategic fit as a standalone veto, preventing the very common move of letting idea size override the question of whether you're the right owner.
McAllister was tasked with driving direct-traffic loyalty — customers starting their shopping on Amazon rather than elsewhere — without a paid mechanism like Prime. He profiled several concepts against the working backwards process and reviewed them with Bezos and Wilke every couple of months.
→ Amazon Smile was the concept that survived all three tests and was greenlit at the review; the team was then built and the product launched.
Common mistakes
Greenlighting on size alone
A big idea with no strategic fit and no plan is a distraction dressed as ambition. It consumes the scarcest resource — leadership attention — before failing.
Skipping the FAQ because the idea is early
The third test is stage-scaled, not stage-exempt. An early idea still owes a plan proportional to its stage; 'it's too early to know' is not the same as 'we haven't thought about it'.
Is it for you?
Best for
Product leaders and founders deciding which new initiatives deserve funding, headcount, and executive attention
Not ideal for
Prioritising incremental improvements within an existing product where strategic fit is already settled
From the transcript
“one is it a big idea and then second is it something we should be doing”
“and the third test is their legitimate plan to succeed and you gotta have all three of those things”
“the fact basically says okay I've thought through the the internal components or the finances or the the key technical hurdles”
From the episode
What it takes to become a top 1% PM
Ian McAllister (Uber, Amazon, Airbnb)