Ruthless Referencing
Diagnose talent by extracting the right information from many references, not by trusting a 45-minute interview.
- Difficulty
- Moderate
- Time to result
- ~weeks to results
- Steps
- 5
- Confidence
- 92%
Interviews are a poor signal for assessing strangers; references, done ruthlessly, are teachable and far more accurate. The technique is to do many references, keep going until you hit a negative one, and — critically — frame the question to match what you're actually deciding. A short post-hire feedback loop closes the learning cycle so the skill compounds.
Origin
Rabois synthesises this from his own hiring failures at PayPal, Tony Xu's practice at DoorDash (20 references per senior hire), and David Sze at Greylock (reference until you hit a negative), plus research he cites on a 30-day post-hire check.
Core principles
- 01Interviews barely move you within the middle of the talent bell curve and can't identify the top 10–50 basis points
- 02References are a teachable muscle you improve by dedication and reps
- 03The framing of the reference question determines the answer — same person, wrong question, wrong result
- 04Volume matters: keep referencing until you hit a negative reference
- 05Close the loop fast: a 30-day post-hire check is about as predictive as waiting a year or two
How to run it
- 1
Do many references, not one or two
For senior hires, run a high volume of reference calls. DoorDash's Tony Xu does 20 references on every single senior hire.
Pro tip High reference volume is a teachable craft — being 'incredibly dedicated' to it makes you measurably more accurate.
- 2
Reference until you hit a negative
Don't stop at glowing references. You've only exhausted the reference set when you finally hit a genuinely negative one.
Pro tip Every strong candidate has at least one detractor — surfacing it tells you the real risk profile.
Watch out Stopping after positive references gives a falsely clean picture.
- 3
Frame the question to the actual decision
Ask what you're really deciding, not a generic proxy. 'Was X a good employee?' can be mixed for someone who is nonetheless 'capable of being a world-class entrepreneur.' Match the question to the role you're betting on.
Pro tip Use a general arc: 'What would lead to this person being most successful?' and 'If something didn't work out, what's the primary root cause?'
Watch out Great investors passed on Fair because they asked whether the founder was a good employee instead of whether he could be a world-class entrepreneur.
- 4
Probe the candidate's strategic mind directly
For senior/leadership candidates, ask 'If you were CEO of [their current company], what would you have done differently?' Then follow up with 'Why weren't you able to persuade the CEO to do it?'
Pro tip You're testing whether they've absorbed the business model, understand tradeoffs, and can create unfair advantage — plus their ability to drive change.
Watch out Don't read it as an invitation to trash their company; you want insight into value creation, not criticism.
- 5
Run a 30-day feedback loop
Thirty days after any hire, ask yourself and the whole hiring team: would you make the same decision again? Log the answer and what you now see that you missed.
Pro tip This tight loop is roughly as accurate as waiting one to two years, so the lesson compounds far faster.
Watch out Most people hire and never revisit how it went, so their gut never actually improves.
In the wild
When VCs reference-checked Max Rhodes (co-founder of Fair, ex-Square), most asked Jack Dorsey 'Was Max a good employee?' — which drew a mixed answer, and several talented investors passed. The right question was whether Max could be a world-class entrepreneur, to which the answer was yes.
→ Investors who asked the wrong question passed on Fair and regret it — same person, wrong question, wrong result.
Tony Xu runs 20 references on every single senior hire. Rabois bets that this dedication makes Xu both very good and very accurate at senior hiring.
→ Held up as the exemplar of referencing as a teachable, high-volume craft.
Common mistakes
Asking the generic question
Defaulting to 'was this person good at their last job?' instead of the question that maps to your actual bet produces misleading answers about capable people and causes you to pass on great hires.
Never closing the feedback loop
Interviewing and hiring without a structured 30-day review means the 'gut' never gets calibrated — people repeat the same misreads for years without learning.
Is it for you?
Best for
Founders and hiring managers making senior/leadership hires who want to build accurate talent assessment as a durable skill
Not ideal for
New-graduate or entry-level hiring, where the reference context is too thin and off to be reliable
From the transcript
“Tony at DoorDash does 20 references on every single senior hire. 20.”
“you couldn't stop referencing a founder until you hit a negative”
“Same person, wrong question, wrong result.”
“if you ask yourself 30 days after any hire, would you make the same decision?”
“If you were CEO, what would you have done differently?”
From the episode
Hard truths about building in the AI era
Keith Rabois (Khosla Ventures)