The Decide-Now Protocol
Ask four questions to decide at speed without pretending you know everything
- Difficulty
- Easy
- Time to result
- ~days to results
- Steps
- 6
- Confidence
- 90%
Hoffman adopted this decisioning mode at PayPal after wishing he'd had it at SocialNet. Confronted with a decision, you run four questions in sequence. First: can I make this decision right now? Second: what would I decide — the answer is usually already there, X not Y. Third: is there anything I could learn by researching or talking to people that would change X to Y? Fourth: what is the cost and time of learning that? If the cost is too great, you go with X and live with it. Layered on top is the one-way door / two-way door test: if you can recover from the decision, you're less likely to justify the research at all. The mechanism's real product is comfort — it trains you to decide at intense speed while uncomfortable, because you don't know everything and never will.
Origin
Extracted from Young and Profiting
Core principles
- 01Speed of decision is itself a competitive advantage
- 02Most decisions can be made right now with what you already have
- 03Research only earns its cost if it could flip the answer
- 04Reversible decisions deserve less deliberation than irreversible ones
- 05Discomfort with incomplete information is the price, not a warning sign
How to run it
- 1
Ask whether you can decide right now
Confront the decision and test immediately whether it can be closed in this moment rather than queued for a future meeting.
Pro tip If the answer is yes, say great and decide — the queue is where speed dies.
Watch out The reflex to gather more is often about comfort, not information.
- 2
Name your default
State what you would decide if forced right now: I would decide X, not Y. Making the default explicit is what makes the rest of the protocol testable.
Pro tip The default is usually available instantly — Hoffman notes the usual answer is already there.
Watch out A vague default turns the next questions into an unfalsifiable stall.
- 3
Test what could flip X to Y
Ask specifically what you could learn — by researching, talking to people, or running a test — that would actually change the answer from X to Y.
Pro tip If nothing you could realistically learn flips the answer, the research is theatre.
Watch out Information that is merely interesting is not decision-relevant.
- 4
Price the learning
Estimate the cost and time to obtain that flipping information. If it's too great relative to the decision, stop.
Pro tip Pair this with the cheap de-risking toolkit: one conversation or a paper test may be affordable.
Watch out Time cost is the one founders systematically underprice.
- 5
Apply the door test
Ask whether this is a one-way or two-way door. Recoverable decisions justify far less investigation than irreversible ones.
Pro tip Two-way doors should be walked through immediately and reviewed later.
Watch out Some decisions look reversible but aren't — culture and infrastructure especially.
- 6
Commit and live with it
Go with X, accept that you decided without complete information, and get comfortable with that discomfort as the standing condition of the job.
Pro tip The practice compounds: each committed decision raises your tolerance for the next.
Watch out Relitigating a committed decision costs the speed you bought.
In the wild
Hoffman describes PayPal's decision speed as one of its many ethoses, and the thing that let the company navigate a period when, as he puts it, lots of things almost blew them up. He compares it to an X-wing going into the Death Star — genuine uncertainty about whether they'd live or die on any given call. The protocol made intense-speed decisions possible precisely because it stops requiring certainty: you name your default, price the information that would change it, and if the price is too high, you commit anyway.
→ PayPal navigated repeated near-death pivots — encryption on mobile, to cash, to Palm Pilots, to online merchant — and succeeded.
A founder faces a tooling decision and instinctively books a week of vendor calls. Running the protocol: can I decide now — yes. Default: vendor X. What could flip it to Y — only a real usage test at scale. Cost of that test: three weeks and an engineer. Door test: switching later costs a migration, but a recoverable one. Result: pick X today, ship, and revisit if the data says otherwise. The week of calls would have produced comfort, not a different answer.
→ A week of deliberation collapses into a five-minute decision with an explicit revisit trigger.
Common mistakes
Researching information that can't flip the answer
The third question exists to kill this. If no realistic finding changes X to Y, the research is a delay dressed as diligence.
Treating two-way doors like one-way doors
Recoverable decisions get the same agonizing treatment as irreversible ones, which is where most decision latency actually accumulates.
Waiting to feel certain
The protocol deliberately leaves you uncomfortable because you don't know everything when deciding; waiting for comfort means never deciding at speed.
Is it for you?
Best for
Founders and operators facing a constant stream of decisions under uncertainty who are losing more to delay than to error.
Not ideal for
Genuinely irreversible, high-stakes, bet-the-company calls where the cost of research is trivial next to the cost of being wrong.
From the transcript
“when I'm confronted with a decision I say can I make this decision right now and if I can make the decision right now I…”
“is there anything that I could learn by researching, talking to people, etc. that would change from X to Y? Okay, what's the cost and…”
“it gets you comfortable with making decisions at a really intense, you know, kind of, you know, speed where you're uncomfortable with it because you…”
From the episode
Reid Hoffman: LinkedIn Co-Founder on Building and Scaling Massively Valuable Companies Fast
Reid Hoffman