Ross Perot's Idle Machine Play: how an IBM salesman sold what the computer could do instead of the computer, the 7 idle machines sitting in businesses today, and the pitch to land your first client
In 1962 Ross Perot was one of IBM's best salesmen. He had already filled his whole year's quota in January. He kept telling IBM that customers needed more than a machine. They needed someone to make the machine do the work. IBM passed. So he quit, wrote a $1,000 check, and started Electronic Data Systems. EDS did not sell computers. It sold what computers could do. In 1984 General Motors bought it for $2.5 billion. This page breaks down the exact deal where he found the idea, why the leader of a technology wave never fills this gap, the 7 places the same gap is wide open with AI right now, and the word-for-word pitch to land your first client.
Overview
Every time a new machine changes business, the same thing happens. The company that makes the machine races to sell as many as it can. The customers buy it because they are told they must. Then most of them cannot get the work out of it. The machine sits there, paid for and half used.
Ross Perot saw this up close in the early 1960s. He was selling IBM mainframes in Dallas. The machines were huge, expensive and new. Most companies had no one who knew how to run them.
He built a company on that one gap. He did not compete with IBM. He did not build a better computer. He sold the part IBM did not want to sell: the work around the machine.
The IBM years: he asked for the worst accounts
Perot joined IBM in Dallas after the Navy. His manager, a man named Wendler, gave him the toughest accounts in the Dallas office.
That suited Perot. IBM paid higher commissions on new business and on difficult accounts. So the accounts other salesmen avoided were the ones that paid the most. He asked for them.
There was a catch. IBM also set a yearly limit on how much commission a salesman could earn. Once you hit it, the rest of the year paid you very little.
Perot filled his quota earlier every year. In 1962, his last year at IBM, he filled the whole year's quota in January. It took him 18 days. That left eleven months where selling more made him almost nothing. He had time, and he had an idea IBM did not want.
The 7070 nobody needed
One of his problem accounts was Southwestern Life Insurance Company in Dallas.
A middle manager at Southwestern had signed an agreement for an IBM 7070. It was the largest commercial machine IBM made at the time. Southwestern did not need anything close to that much computer. Run the way they planned, it would have sat idle most of the day.
The man at the top made it worse. Southwestern's chairman, James Ralph Wood, did not like salesmen in general or IBM in particular.
So this was the account: a signed deal for a giant machine the customer could not fill, and a chairman who did not want to hear from IBM. Most salesmen would have let it die.
Perot got a meeting with Wood through IBM's president, Gilbert E. Jones.
The fix: sell the hours the machine would waste
Perot did not try to talk Wood into needing a bigger computer. He found someone else who needed one.
A government agency, the Agricultural Commodity Price Stabilization Service, had a problem of its own. Its staff were flying to New Orleans several times a week to use a rented 7070 there.
Perot put the two problems together. Southwestern keeps the big machine. In the hours Southwestern does not use it, the agency rents that time. The machine earns money when it would have sat idle. It stops being a waste and starts paying for itself.
Wood bought. Perot also sold Southwestern on running the machine for a second and a third shift, so it paid itself off even faster.
Look at what he actually sold. He sold hours of computing to someone who needed them, using a machine someone else already owned.
EDS: from idle hours to Medicare
Perot pitched IBM again and again on the same idea: sell customers the computer time, the software and the people to run it, not just the machine. IBM was not interested.
Then came the barbershop. In 1962, at 31, he was waiting for a haircut and flipping through Reader's Digest. He found a line from Thoreau: "The mass of men lead lives of quiet desperation." He said it gave him the strength to go out on his own. He kept that page framed in his office.
On June 27, 1962, he wrote a $1,000 check to start Electronic Data Systems. The money came from his wife, Margot.
His first move was the Southwestern play again, this time for himself. He bought idle time on that same 7070 at Southwestern at wholesale, during the hours it sat unused. Then he sold that time at retail to companies that needed computing and had no machine. His first customer was Collins Radio.
He had no factory, no product and no computer of his own. He had a machine other people owned, hours they were not using, and buyers who needed those hours.
The big break came a few years later. In 1965 the US government created Medicare and Medicaid. Millions of claims suddenly had to be processed. EDS won contracts to handle them, starting in Texas.
EDS went public in 1968. In 1984 General Motors bought it for $2.5 billion and Perot became GM's largest single shareholder.
Why IBM passed, and why the leader always will
It is easy to say IBM was foolish. It was not. IBM was in the middle of the biggest technology wave of its time. Every machine it sold was worth more than any service it could add. Every hour its people spent running a customer's computer was an hour not spent selling the next one.
The leader of a new wave has one job: build the best machine and sell it to as many people as possible, as fast as possible. Services slow that down. So the leader leaves them on the table on purpose.
That creates the gap. It opens when a technology is new, spreading fast and hard to use. It stays open until the technology gets easy enough that customers can run it themselves. EDS built a company in that gap and stayed in it for decades.
The rule we take from Perot: when a new machine is spreading faster than people can learn to use it, do not sell the machine. Sell the result the buyer was promised and is not getting.
The 7 idle machines sitting in businesses today
AI is today's 7070. Businesses are paying for ChatGPT, Claude, Copilot and a pile of AI tools. Most of them got the same pitch Southwestern got: this will change everything. Most of them use a sliver of it. The AI companies are racing each other to build better models. They are not coming to set it up for a dentist's office.
Here are 7 places the machine is paid for and sitting idle. Each one is a Perot-style business: you sell the result, not the tool.
1. The front desk that misses calls
Idle machine: AI phone and chat tools that can answer, book and follow up around the clock.
Who has it: clinics, salons, contractors, law offices.
What you sell: "Every call answered and booked, set up in a week." You set up the AI receptionist on their existing number and calendar, then check it every month.
2. The quotes that never get sent
Idle machine: AI that can turn a phone call or site visit into a written estimate.
Who has it: roofers, landscapers, renovation crews.
What you sell: "Quotes out the same day." You build the template and the process so the owner talks into his phone and the quote is in the customer's inbox before dinner.
3. The leads nobody follows up
Idle machine: AI email and text tools inside the CRM they already pay for.
Who has it: real estate agents, mortgage brokers, car dealers, gyms.
What you sell: "Every lead gets 5 follow-ups without you lifting a finger." You write the sequences and switch them on.
4. The paperwork pile
Idle machine: AI that reads invoices, forms and receipts.
Who has it: bookkeepers, small accounting firms, property managers.
What you sell: "Your data entry done every night." You connect the inbox to the books and check the exceptions.
5. The reviews and posts that never happen
Idle machine: AI writing and image tools.
Who has it: restaurants, local shops, clinics.
What you sell: "A post a day and a reply to every review." You run it for them on a monthly fee.
6. The team that got ChatGPT and never opened it
Idle machine: company-wide AI licenses.
Who has it: offices of 10 to 100 people that bought seats because they felt they should.
What you sell: "Your team using it on real work in 30 days." You find the 3 tasks each role repeats and build the prompts and shortcuts for them.
7. The knowledge stuck in one person's head
Idle machine: AI assistants that can answer questions from a company's own documents.
Who has it: any business where new hires keep asking the owner the same questions.
What you sell: "A helper that answers like your best employee." You gather the documents, set it up and keep it current.
The test: if the AI tool got 10 times cheaper tomorrow, would this business get better or worse? A business that sells the result gets better: its costs fall and the customer still pays for the result.
The pitch: land your first client the way Perot landed Southwestern
Perot's sale had 4 parts. He found an account that already paid for a machine. He found the waste. He found someone who would pay for that waste. And he went straight to the person who signs.
Step 1. Pick one of the 7 and one kind of business (10 minutes)
Do this: choose the business you understand best. If you have worked in a clinic, pick clinics. If your uncle runs a roofing crew, pick roofers. One narrow group is easier to sell to than a clever idea.
Step 2. Find 20 of them that already pay for the machine (30 minutes)
Do this: look for the signs. A chat bubble on their site that nobody answers. A "call us" number that goes to voicemail. A Google profile with reviews and no replies. These are businesses with an idle machine.
Step 3. Go to the person who signs
Do this: Perot went to the chairman, not the middle manager. In a small business, that is the owner. Call, walk in, or send this:
Hi [name], I called your office at 6pm yesterday and got voicemail. I set up AI receptionists for [type of business] so every call gets answered and booked, even after hours. You likely already pay for most of what it needs. Could I show you in 15 minutes how many calls you are missing?
Step 4. Show the waste in their own numbers
Do this: Perot did not sell a feature. He showed Wood a machine that would sit idle and a buyer for its empty hours. Do the same. Count the missed calls, the unanswered reviews, the leads with no follow-up. Put a dollar number next to it using their own prices.
You told me a new patient is worth about [their number]. Last week I called 5 times after 5pm and got voicemail every time. If even 2 of those a week were patients, that is [2 x their number] a week walking to the clinic down the road. I can fix that in 7 days. You only keep paying me if it books them.
Step 5. Make it pay for itself, like the second and third shift
Do this: price it so the first month's result covers your fee. Offer a short trial with a clear number to hit. Perot made the 7070 pay for itself. Your service should too.
Here is what I suggest. I set it up this week. We track booked appointments for 30 days. If it books more than my fee, we keep going month to month. If not, you owe me nothing for the setup.
Step 6. Do it again for the next one
Do this: once one works, you have the thing Perot had after Southwestern: a working example in one industry. Every other business of that type has the same idle machine. Call the next 20.
The numbers
| Figure | Number |
|---|---|
| Year's quota filled | January 1962, in 18 days |
| The machine | IBM 7070, IBM's largest commercial machine at the time |
| EDS founded | June 27, 1962, with $1,000 |
| Age at founding | 31 |
| First customer | Collins Radio |
| Big break | Medicare and Medicaid claims processing, from the mid 1960s, starting in Texas |
| Public | 1968 |
| Sold to General Motors | 1984, about $2.5 billion |
Want 50 more breakdowns like this?
Real businesses with their revenue, tools, costs and how they grew, plus step-by-step guides to start your own. Free.
Get free accessMore breakdowns
Sources
- Selling Power, "Inside the Mind of H. Ross Perot" (2010)
- D Magazine, "Perot's Biggest Battle" (June 1992)
- The Texas Observer, June 5, 1992
- Encyclopedia.com, Perot, Ross; Electronic Data Systems Corp
- Company Histories, Electronic Data Systems Corporation
- Associated Press obituary, July 9, 2019; Biography.com; CNBC
