Ch2 01: 64 Clicks to Buy a Car: How Tesla Deleted Its Way to Growth#

Buying a Tesla should’ve been the easiest car purchase on the planet. No dealership. No haggling. No trade-in theater. Just hit the website, spec your car, and buy. Simple.

Except it wasn’t simple at all.

When I first mapped the online purchase flow at Tesla, I counted sixty-four clicks to complete a single order. Sixty-four. That’s more clicks than filing a tax return. For a company that prided itself on reimagining every corner of the car business, our e-commerce experience was, frankly, embarrassing.

But here’s the kicker — nobody inside the company thought it was a problem. Every one of those sixty-four clicks had a reason. Legal needed certain disclosures. Finance needed certain confirmations. Engineering needed certain configuration checks. Marketing wanted certain upsell moments. Each team had bolted on their requirements independently, and each requirement made perfect sense on its own.

The result was death by a thousand reasonable decisions.

The first round of cuts was easy. We scanned the flow and asked: which steps are obviously redundant? Duplicate confirmation screens. Repeated data entry. Pages asking for information we already had. Low-hanging fruit — stuff nobody would defend once it was called out.

We went from sixty-four to about forty. A thirty-seven percent improvement. At most companies, that would’ve been declared a win. A project team would’ve written a case study. Someone might’ve gotten a promotion.

We weren’t done.

Round two required a sharper blade. For every surviving step, we asked one question: Would the customer pay extra for this?

Think about what that question does. It vaporizes every internal justification — “legal needs this,” “finance requires that,” “we’ve always done it this way” — and replaces them with the only standard that ultimately counts: does this step create value the customer recognizes and would pay for?

The answer, for most steps, was no. Customers don’t pay for disclosure screens. They don’t pay for marketing upsells. They don’t pay for internal compliance rituals. Those steps serve the organization, not the customer. That doesn’t make them worthless — but it makes them candidates for deletion or radical compression.

We went from forty to about twenty-five.

Round three was the toughest, because it meant going toe to toe with the people who held the most authority: legal and regulatory.

“We need this disclosure for compliance.” That sentence had been a conversation killer for years. In some cases, it was genuinely true — real legal requirements you can’t skip. But in many cases, the “compliance requirement” turned out to be an internal interpretation of a regulation, not the regulation itself. The actual law required certain information to be provided. It didn’t specify that the information had to live on a separate page, require its own acknowledgment click, or appear in a particular order.

When we went back to the actual text of the law — not the compliance team’s reading of it, but the statute itself — we found room. A lot of room. Disclosures could be consolidated. Acknowledgments could be merged. Information could be presented in ways that were fully legal but dramatically less intrusive.

We went from twenty-five to fifteen.

Round four was the most counterintuitive. We deliberately deleted too much.

We stripped the flow to eight clicks. We knew some of those cuts would break things. That was the point. By overshooting — by pulling steps we suspected might still be needed — we created a situation where we could discover which steps were truly essential through observation, not speculation.

Three of the deleted steps turned out to be genuinely necessary. Customers got confused at certain points. Critical info was missing. So we added those three back.

Final count: twelve clicks. An eighty-one percent reduction from where we started.

There’s a reason this took four rounds instead of one. Complex systems have layered redundancy — like Russian nesting dolls of waste. The surface-level bloat is easy to spot and easy to cut. But cutting it exposes the next layer, which was previously hidden behind the “reasonable” outer shell.

I call this the Progressive Exposure Effect. Each round of deletion doesn’t just remove waste — it reveals the waste beneath. And counterintuitively, each deeper round often exposes more waste than the one before, because the inner layers have been shielded by the “necessity” narrative of the layers above.

This is why single-round optimization projects consistently underdeliver. They shave the surface and celebrate, never reaching the structural waste that accounts for most of the problem. The real payoff lives in rounds three and four — the territory that feels uncomfortable, challenges authority, and forces you to distinguish genuine requirements from institutional habits wearing a compliance costume.

The overshoot strategy in round four deserves a closer look, because it runs directly against how most organizations make decisions.

In a typical optimization effort, the unspoken goal is to remove exactly the right amount — no more, no less. The fear of cutting too deep dominates the room. “What if we break something?” “What if customers complain?” “What if legal comes for us?” Those fears are real, but they create a systematic bias toward caution. You always end up cutting less than you should.

The overshoot approach flips this. Instead of hunting for the minimum viable process from above (by carefully trimming), you find it from below (by deliberately going too far and adding back). The result is more precise, because you’re working with empirical evidence — you can see what actually breaks — rather than theoretical guesses about what might break.

Deleting too much and adding back ten percent is almost always more effective than deleting too little and wondering what you missed.

Guidance#

Pick your most customer-facing process — the one that touches the most people, most often. Map every step. Count them. Then run four rounds:

Round 1: Cut anything obviously redundant. Duplicate steps, repeated inputs, unnecessary confirmations. The easy round.

Round 2: For every remaining step, ask: “Would the customer pay for this?” If not, it’s a deletion candidate.

Round 3: For every step defended by “compliance” or “policy,” go back to the original source. Read the actual regulation, contract, or law. Does it truly require this specific step, or have you over-interpreted?

Round 4: Deliberately overshoot. Delete more than feels safe. Watch what breaks. Add back only what’s proven necessary.

If your process runs more than twenty steps, I guarantee you can get it below ten. The waste is always bigger than you think. And the deeper you dig, the more you find.