Ex-Tesla President: Elon Behind-The-Scenes, Saving Tesla & Scaling a Trillion-Dollar Company

Why does it matter? Because Elon's edge isn't genius — it's an algorithm anyone can run.

The ex-president of Tesla sat down to reveal what it actually looks like inside one of the most consequential companies ever built — and the answer is less "visionary alien" and more repeatable operating system. John, who scaled Tesla from 4,000 to 40,000 employees, shows exactly how the method works, step by step.

• Elon's 10x goal-setting isn't motivational theater — it's a forcing function that makes incrementalism structurally impossible
• 9,000 uncalled test-drive leads were sitting in Tesla's CRM while the company was panicking about missing its quarterly number
• 360,000 build-to-order configurations were religion at Tesla — until purchase data showed customers were actually only buying two
• The real AI money won't be made by the hyperscalers, for the same reason the real internet money wasn't made by Netscape

Set a 10x goal and incrementalism dies on its own

"If you set a goal for 5 to 7% improvement, you're probably going to get three to five." That's the trap most leaders never escape.

Elon's fix is blunt: set an order-of-magnitude goal and the incremental path becomes physically unavailable. You can't tweak your way to a 20x result. John lived this firsthand when Elon handed him a target to improve digital sales by 20x on a product nobody was buying sight-unseen online for $120,000. The immediate consequence wasn't a roadmap — it was a completely different question. How many clicks does it take to buy a Domino's pizza? Ten. Tesla was at 64. "We are 64, Domino's is 10. Let's go to 10."

That single reframe cascaded into questioning one of Tesla's core religious beliefs: fully customizable, build-to-order cars. Nobody challenges a belief that everyone's proud of. The 10x goal forced it into the open.

The principle transfers cleanly. Pick your weakest metric. Set a goal so large that your current process can't survive it. The impossibility of the number is the point — it's the only forcing function that produces structural change instead of marginal improvement. Percentage goals produce percentage results. Order-of-magnitude goals produce new thinking.

Tesla's proudest feature was strangling its own sales

360,000 possible car configurations sounds like a customer-centric triumph. It was actually a conversion catastrophe — and nobody questioned it because everyone loved the idea.

John's team ran the purchase data. Customers weren't buying 360,000 versions. They were buying two: standard or performance. That's it. The entire build-to-order model — the thing Tesla was most proud of, the thing that made them "not like other car companies" — was generating decision fatigue, piling on clicks, and making it impossible to reach a 10-tap checkout.

When John brought the data to Greg, head of manufacturing: "Oh my god, I've been waiting for this day... you know what that would do to our factory? Totally simplifies the factory. It simplifies the supply chain." Doug in engineering was equally ready: stop designing and testing parts that nobody buys, redirect that time to what actually matters to the customer.

The senior management meeting was harder. Someone pushed back immediately — "you're the new guy, it's religion here, we do build to order." John's response: "It might be religion. But that religion might take you to the grave."

Elon looked at the data, listened to Greg and Doug, and said yes. If you buy a Tesla today, you pick from two or three configurations. The lesson isn't about car companies. It's about the belief you're proudest of — audit it with actual purchase data before it quietly kills your conversion.

Tesla's sales crisis was already solved — nobody had made the calls

A month and a half into a quarter where Tesla had promised Wall Street 12,000 cars and sold three, John's diagnostic move wasn't strategy. It was mystery shopping.

He visited eight Tesla stores across a week of travel, used different email addresses each time, took test drives — the supposed fulcrum of the entire sales funnel. He didn't get a single callback. One call to the head of sales ops surfaced the number: 9,000 people had taken test drives in the last 30 days and hadn't been contacted.

"9,000? You're going to make your quarter if you just call these people."

The fix took two hours. Cut off new leads to any salesperson until they'd worked through their existing test-drive backlog. Block it in the system, done, globally. By the next morning, "sales are flying" out of Asia and Europe.

Then it dawned on John: "I made this decision like I was the CEO. I don't even work for Tesla yet." He called Elon, explained what he'd found and what he'd done. Long silence — at least 60 seconds. Then: "I think you're going to fit in here just fine. You've proven you can be useful. Why don't you join?"

The deeper problem was two-layer: a bad incentive structure (commission tied to test drives, not closed sales) and a hiring profile that brought in people who didn't know how to ask for the sale. Fixing the callback backlog was the fast move. The constraint was never top-of-funnel.

Elon's interview is a video game designed to find your ceiling

Two sentences of pleasantries, then straight into a manufacturing problem Elon already knew cold. That was John's first conversation with him — and it was also his job interview.

The method is explicit: go super deep on a problem you understand well, so you can tell the difference between someone who actually did the work and someone who was adjacent to a team that did. "Almost like a video game — how many layers can this player get through before they're stumped."

Elon's target: say "wow" three times in the first 20 minutes. If it's not happening, it's not there. Trust the conversation, not the resume.

The trap John watches for is credit laundering — people who were on an eight or ten-person team that did something remarkable but weren't the ones driving the curiosity or the discovery. Fully matrixed companies like Nike make this especially hard to detect; the nichy job structure means almost nobody had full ownership of anything.

His counter-move: flip the table. Present a real current problem of his own, then watch the candidate's curiosity, their analytics instincts, the quality of their questions. "Go into a problem of theirs, then flip the table and go into a problem of yours."

Sam added the hiring corollary: hire the salesperson from the company with the terrible product. The one closing quota at the tier-three company with nothing going for it — that's the fisherman. The Apple Store rep is just an order-taker. As John put it: the Microsoft Store rep selling a Surface two doors from a MacBook Air, and not starving — "go hire them."

Your two eyes and two ears beat any analytics dashboard

The Falcon Wing door problem at Tesla was solved in 10 minutes. Not by data, not by a consultant, not by sleeping on the factory floor — though Elon had been doing that. By standing still and watching.

John walked the line to where inventory was stacking up — that's always where the bottleneck lives. He and Elon just stood there. Within minutes it was visible: workers were threading bolts blind, missing the angle, door going on crooked. They needed a jig. Maintenance built it. Done.

"I'm going to introduce you to the most powerful analytics you have as a leader. Your two eyes and your two ears."

The same logic built Service King. Before John even had the job at Tesla, he mystery-shopped eight stores in a week. The same logic produced the Swiffer — Chip Berg at P&G watched moms suffer with regular mops and invented the solution. Scott Cook at Intuit has senior executives do "follow me home" sessions, watching real customers click through the product until a payroll accountant said "it'd be easier if QuickBooks just did payroll" — and that became a massive business line.

The bank executives version is the most damning: John asked a room full of them to raise their hand if they'd used their own consumer app in the last week. No hands. "I could have guessed that because I'm a consumer of two of your banks, and your apps are terrible." The leaders who rely on dashboards get insights weeks after a 30-minute observation would have surfaced the same thing — and they miss the physical root cause entirely.

A $50B industry ran like a hair salon — one assembly line fix built a billion-dollar chain

18 days. That's how long it took a collision repair shop to complete 6 to 8 hours of actual work. John spotted this gap after his wife's car sat in a shop for two Fridays straight and came out unfinished both times.

The root cause was incentive structure, not laziness. Every technician controlled two bays on commission. The car sitting in their bay was a cash register — there was zero reason to finish it fast and hand it back. They'd cherry-pick hours, surround themselves with three or four jobs, and let cycle time balloon. "Essentially a hair salon."

300 platforms had been built for cloud cybersecurity in five years. Zero for SMBs. The collision industry equivalent: 38,000 shops averaging $1.2M in revenue, none of them running assembly lines. John applied Henry Ford's model — put cars on a production line, pay the whole team a bonus based on cycle time and throughput instead of individual bay hours, and the incentive structure flips entirely.

The pitch to customers became simple: get your car back in a day instead of 18. The company became Sterling, then Service King, now owned by Carlyle with several billion dollars in revenue. The business didn't require a technical breakthrough. It required reading the incentive structure correctly and being the outsider who wasn't locked into the incumbent logic. Any fragmented industry where individual actors are locally optimizing in ways that destroy customer value is the same setup.

The hyperscalers are Netscape — the money is in what gets built on top

Everyone's watching the foundation model arms race like it's the main event. It isn't.

The browser wars felt like the main event too. Netscape made careers. But browsers became a commodity — and what got built on top of a browsable web was Facebook, Airbnb, Uber, and trillions in market cap that nobody could have predicted while they were arguing about Internet Explorer versus Netscape.

"I sort of wonder if they're creating a tooling layer" — and if the hyperscalers are the tooling, the question isn't who wins the model race. It's what gets built on top.

John's clearest example: a team that built Tesla's supply chain automation platform in 2017 with ML, now rebuilding it with AI. They went into one of the largest grocery delivery platforms in the country, had agents learn its work rules within hours, and designed a full system in days. The competition is standard ERP implementations that take 9 to 12 months. "Humans who have got AI as exoskeleton like super strength."

Stop watching the foundation model scoreboard. The asymmetric opportunity is the application layer — domain experts using AI to compress what used to take a year into a day.

The pattern that keeps repeating itself

Every insight in this conversation traces back to the same move: go look at the actual thing. The factory floor. The CRM. The checkout funnel. The repair shop bay. The customer trying to navigate your site. The executives who skip this step — who rely on dashboards, receive filtered reports, and set modest goals — are not just slower. They're looking at a map of a territory that already changed.

What the next wave of AI application companies will have in common with Service King and Tesla's two-configuration checkout is exactly this: someone willing to stand in front of the broken thing long enough to see what's actually wrong. The tool changes. The move doesn't.

---

Topics: Elon Musk, Tesla, hiring, goal setting, operations, entrepreneurship, AI, business models, sales, incentive design, management, product design