Structural Conditions for Innovation
Companies that say they tolerate failure but punish it in comp and career don't get innovation. Change the structure.
- Difficulty
- Expert
- Time to result
- ~months to results
- Steps
- 5
- Confidence
- 90%
Most companies claim they are fine with failed bets, then run performance reviews, bonuses, and budgets that punish anyone on a failed bet. Carr identifies the concrete structural changes Amazon made instead: eliminate performance bonuses so compensation tracks the long-term stock rather than your unit's quarter; evaluate people on what they built and contributed (inputs) rather than only on the outputs of the business they happened to sit in; and have a CEO who personally runs interference for the new-thing teams rather than delegating them.
Origin
Bill Carr's synthesis of what Amazon did structurally, prompted by conversations with executives at other companies who crave innovation but can't get it. He explicitly connects it to Safi Bahcall's book Loonshots, which describes the structural conditions — separate decision-making and approval structures — needed for innovation to occur inside a large company.
Core principles
- 01Most structures inside a big company are designed to crush and impede a small innovative team trying to build something new.
- 02Compensation, not slogans, tells people whether failure is survivable.
- 03Evaluate contribution by inputs — what you built and improved — because plenty of people sit in businesses going up and to the right for reasons unrelated to them.
- 04Innovation cannot be delegated by the CEO; it requires their personal engagement.
- 05A failed product may be a good idea at the wrong time, not a bad idea.
How to run it
- 1
Remove the financial penalty for taking the risky assignment
Amazon had no performance bonuses. Compensation was based on the stock price, so a killer year running the book business earned no kicker and a bad year cost nothing — aligning everyone to what was right for the company over the long term.
Pro tip Test your comp design against Carr's own move: he left the largest P&L (books, then music and video) for digital media, where there was zero business, and his compensation did not change one way or the other.
Watch out Any bonus tied to a business unit's short-term financials silently prices people out of risky bets.
- 2
Grade people on inputs, not just outputs
Run performance management on what the person actually delivered and built — how they improved selection, lowered prices, shipped a capability — rather than only on the results of the business they happen to sit in.
Watch out Outputs still matter; Carr says they cared about them too. The point is not to let a rising tide be mistaken for individual contribution.
- 3
Staff the new bets with your best senior leaders
Amazon put two of its smartest leaders — Steve Kessel and Andy Jassy — on digital media and AWS. Seniority is not a reward here; it is the mechanism by which the new team survives contact with the company's approval structures.
Pro tip Their seniority itself runs interference on the approvals and process friction that would otherwise strangle a small team that needs speed.
- 4
Put the CEO in the room, regularly
Bezos met with those leaders regularly and was deeply engaged in reviewing what they were going to build, including deciding what to build. That personal engagement is what let them cut through the approval-here, approval-there friction.
Watch out If the CEO delegates innovation to someone else, the conditions for it do not exist — however much they crave it.
- 5
Post-mortem failures on the customer problem, not the execution
When a product fails, first ask what problem it solved for customers. If execution was fine, the fault is in the concept. And distinguish concept failure from timing failure — some good ideas simply arrive before the technology or the scale exists.
Pro tip Carr: nine times out of ten, if the product was executed correctly, the answer is in what was wrong with the concept.
In the wild
In the early 2000s Amazon shipped a feature called Slots, where third parties could bid on search terms and place a small ad. It failed — not because the idea was bad, but because Amazon simply didn't have the scale it has today. The same idea works now.
→ Carr's rule: a perfectly good product idea can fail purely because the timing or the technology isn't there. Bezos wrote in 2004 about a kitchen puck you'd talk to and shop from — roughly a decade before the technology made it Echo.
Plenty of people inside Amazon doubted the Kindle, including in contentious board meetings. Carr spent years describing the Prime Video vision — watching on your TV, Amazon making its own movies and TV shows — and people laughed at him.
→ Volume of internal disagreement is not a signal of whether a product is right or wrong, which is precisely what makes the decision hard and makes the structural conditions matter more than the debate.
Common mistakes
Saying you tolerate failure while your systems punish it
If working on a project that fails puts your career in the garbage can and costs you your bonus, no amount of leadership messaging about risk-taking will produce risk-taking.
Using disagreement volume as a kill signal
Teams look for 'lots of people disagreed, so it must be a bad idea'. Carr rejects this — the Kindle and Prime Video were both widely doubted internally. It makes the decision harder, not easier.
Chasing near-term financial goals under Wall Street pressure
Carr is sympathetic: the dynamic between public companies and Wall Street creates a genuine fear of failure and a focus on short-term financials. But it is a structural choice, and companies that don't restructure simply will not get innovation.
Is it for you?
Best for
CEOs and senior leaders at scaled or public companies who genuinely want product innovation but keep watching new bets die on process friction and career risk.
Not ideal for
Small startups where everyone is already on the risky bet and there is no bureaucracy to run interference against.
From the transcript
“there were no performance bonuses”
“our compensation was based on the stock price”
“most of the structures inside a big company are are kind of designed to crush and impede a small Innovative team”
“we put like two of our smartest you know leaders in the company on those things Steve kessle and Andy jasse”
“if you don't set up the organization in the right way like you're just not going to get it”
From the episode
Unpacking Amazon’s unique ways of working
Bill Carr (author of Working Backwards)