Jon McNeill

Elon's hiring method: interrogate one problem to test for world-class work

Jon McNeill describes Elon Musk's interview method: rather than reviewing a resume, he picks one problem (ideally one he knows deeply) and goes layer by layer into it to determine whether the candidate can do world-class work and isn't bluffing.

“So his hiring method is he wants to determine whether or not you can do world-class work And his method is to interrogate a particular problem and go super deep on it. And it's better if he tries to skew it to a problem that he knows.”

Steal thisIn interviews, drop the resume and go deep on one problem you understand well to test whether the candidate can actually do world-class work.

Founder-led final interviews to imprint culture as you scale

To protect culture while scaling Tesla from 4 to 40,000 employees, Elon and co-founder JB Straubel made themselves the last interview for every manager-level-and-above hire. McNeill says 60% of his calendar was interviews.

“And he and JB Straubel, the co-founder of Tesla, came up with this method to imprint the culture on every hire and to test for culture. And that was, they would be the last interviews That's a hack that I've taken forward now from, from that Elon experience.”

Steal thisAs you scale, make founders the last interview for every manager-and-above hire to imprint and test for culture.

9,000 un-called test drives: how McNeill fixed Tesla's quarter before he was hired

Before joining Tesla, McNeill mystery-shopped 8 stores, never got a callback after a test drive, then learned 9,000 test-drivers in 30 days had never been called back. He had sales ops block new leads until reps called back prior test drives, and sales took off.

“So he cranks through the CRM and he comes back and says 9,000. And I'm like, 9,000? You're gonna make your quarter if you just call these people. I said, okay, look, this problem has to get solved now.”

Steal thisBefore chasing new leads, work your existing pipeline; un-followed-up prospects are often the cheapest revenue you have.

Your two eyes and two ears are your most powerful analytics

McNeill's mentor taught him to sit and watch frontline work (support calls, a factory line) because direct observation surfaces insights faster than data can reach you. At Tesla he and Elon watched the Model X door line and spotted that workers were threading bolts blind and needed a jig.

“I'm going to introduce you to the most powerful analytics you have as a leader: your two eyes and your two ears. Nice. Use them because you will get insights so much faster than the data can get to you to give you insights.”

Steal thisGo watch the frontline work in person before reaching for dashboards; your eyes and ears surface bottlenecks faster than data.

Follow Me Home: watch real customers use your product

McNeill cites Intuit founder Scott Cook's 'Follow Me Home' practice: senior execs regularly watch real customers use the product over their shoulder to see the friction firsthand. One such session surfaced the idea to put payroll in QuickBooks, now a huge Intuit business.

“I learned this technique from Scott Cook, the founder of Intuit, and he's got this process he calls Follow Me Home. And what it means is you look over the shoulder of real customers who are using your product.”

Steal thisMake execs regularly watch real customers use your product over their shoulder to find friction and missing features.

Cybersecurity built for SMBs, not the cloud

McNeill's venture firm noticed ~1,300 cyber platforms were built for the cloud in five years versus zero for SMBs, even though ransomware attacks concentrate on under-protected small/medium businesses (law firms, accountancies). They built a modern cyber platform for variable SMB tech stacks, now one of the fastest-growing companies in cyber.

“We asked, we just happened to ask the question, How many cybersecurity platforms have been created for the cloud in the last 5 years? And the answer is like 1,300. And then we said, how many have been created for the SMB in the last 5 years? And the answer was zero.”

Steal thisWhen a market is built '100% for the cloud,' look at the underserved on-prem SMB segment where the competition isn't.

Order-of-magnitude goals force a different way of solving

McNeill explains Elon's goal-setting principle: a 5-7% goal yields incremental tweaks, but a 10x or 20x goal forces you to rethink the whole approach. Elon told him to improve Tesla digital sales 20x, then benchmarked Tesla's 64 clicks-to-buy against Domino's 10 taps.

“And so if you set a goal for 5 to 7% improvement, you're probably going to get 3 to 5. If you set a goal for an order of magnitude improvement, now you got people that have to think way differently about that problem because you can't tweak the status quo to get to a 100% improvement or a 20x improvement.”

Steal thisSet a 10x goal instead of a 10% one to force a rethink of the whole approach rather than incremental tweaks.

Tesla's 360,000 configs collapsed to 2 because customers only bought two

McNeill's team found that of 360,000 possible build-to-order Tesla configurations, customers were effectively only buying two (standard and performance). Cutting to a few configs killed decision fatigue and clicks, simplified the factory and supply chain, and freed engineering, despite internal 'build-to-order is religion' resistance.

“But that led to like 360,000 different combinations of a car that you could choose, which is driving a lot of clicks plus a lot of decision fatigue for people. And so we went and ran the data, my team ran the data, and we came back and figured out that people weren't buying 360,000 versions. They were buying two”

Steal thisAudit what customers actually buy; you may be paying for thousands of options that map to just two real choices.

Run the company on 3-sentence emails: problem, root cause, solution

McNeill says Elon proposed limiting emails to three sentences: what's the problem, what's the root-cause analysis, and what's the proposed solution (with cost and economics). A CEO's scarcest resource is information-absorption and decision time, so terse communicators become very valuable.

“He's like, let's try to get down to 3-sentence emails. What is the problem? What's, what's your analysis of the root cause and what's your proposed solution? And, uh, and he's like, I think we could run the company on 3-sentence emails.”

Steal thisForce every status update into three sentences: problem, root cause, proposed solution with economics.

18-day cycle time, 6-8 hours of work: the collision-repair epiphany

Investigating body shops, McNeill found cars take 18 days to return (cycle time) but only 6-8 hours of actual labor (touch time) because commissioned techs hoard cars as cash registers. He applied assembly-line methods and team bonuses on throughput, building a chain (Sterling, now Service King) with several billion in revenue.

“And I went, 18 days to get 6 to 8 hours of work done. That looks like an opportunity because that's the difference. Cycle time is the 18 days. That's from the beginning to the end of the process. And touch time was the amount of time that those technicians were actually touching the car, 6 to 8 hours.”

Steal thisFind industries where cycle time dwarfs touch time and re-architect the workflow to collapse the gap.

From human switchboard operators to 1-800 call centers: the second-order effect of automation

McNeill uses Bell Labs' electronic switch to illustrate that technology kills first-order jobs but creates bigger second-order industries: 800,000 operator jobs vanished, but free long-distance enabled toll-free 1-800 businesses and an entire call-center industry employing millions, which seeded his first software startup.

“humans are really good at seeing the first-order effect which is the job destruction, but they're not good at seeing the second-order effect, which is the job creation that happens by entrepreneurs on the back end.”

Agentic AI supply-chain platform that deploys in days vs 9-12 month ERP

McNeill's team (ex-Tesla supply-chain builders) uses agentic AI to learn a client's work rules and design a supply-chain workflow within hours, competing against standard ERP systems that take 9-12 months to implement. One new customer is a major grocery-delivery platform.

“Like, what they're competing against is standard ERP systems that take 9 to 12 months to implement, and they can implement them in a period of days and deliver that value back. And that's the kind of business, like, I think we're going to see more and more and more of as this capability is unleashed.”

Steal thisUse agentic AI to compress a months-long enterprise implementation into days as your wedge against legacy ERP.

The AI labs are building the tooling layer; the real GDP gets built on top

McNeill argues the AI hyperscalers are building a commodity tooling layer, like browsers in the internet era. The Netscape excitement faded, but Facebook, Airbnb and E-Trade were built on the browsable web, so the bigger prize is the businesses built on top of AI, not the models themselves.

“but what got him what got built on top of a browsable web were things like Facebook and Airbnb and, uh, E-Trade and you name the, the businesses. Like, tons of GDP got created on top of that tooling. And it sort of feels like to me we're getting really excited, like we did about Netscape and Internet Explorer.”