A John Deere tractor sits in a barn in Iowa. The dash screen shows a fault code: three letters and four digits. The farmer who owns the machine paid $3,160 a year for the software that reads that code. The screen tells him something is wrong. It does not tell him what. For that, he drives to the dealer and pays $58.90 more per hour than an independent mechanic charges. The vendor sold him the tractor. Then the vendor sold him the lock.

"Go with the best vendor, and the service pays for itself." Most operators who have signed a long-term vendor contract have heard some version of that line. It sounds like a quality argument. It is not. It is a framing trick. It keeps you measuring what you paid to get in. It never asks what it costs to get out.

The research on this is worth reading. The FTC and five state attorneys general spent years building an antitrust case against Deere. What they found was simple and ugly. Deere made the only diagnostic software that could perform all electronic repairs on its equipment. Then Deere made that software available only to its authorized dealers.

Farmers could buy a lesser version. That version showed fault codes but redacted what the codes meant. It blocked reprogramming. It disabled key repair functions. The price of that hobbled tool was $3,160 per year, per the FTC complaint.

The price of staying locked in was much higher. A U.S. PIRG study called "Out to Pasture" put the total cost of repair restrictions across all farm equipment makers at $4.2 billion a year. That figure includes $3 billion in tractor downtime and $1.2 billion in excess labor costs paid to dealers who held the only working keys.

The advice did not fail the farmers. It failed the system they were trying to run.

Deere tried to solve this with a promise, not a fix. In early 2023, the company signed a memorandum of understanding with the American Farm Bureau Federation. The deal: Deere would open repair access. In exchange, the Bureau would drop its support for right-to-repair laws.

One month later, PIRG audited the customer version of Deere's diagnostic tool. The fault descriptions were still redacted. The troubleshooting guides were still missing. Reprogramming was still disabled. The promise was designed to stop the law, not open the system.

The MOU had no enforcement clause. Either side could walk away with 30 days’ notice. A federal court would not have stepped in to hold it together. That is not a contract. It is a stall.

It took the FTC and five state attorneys general to break the lock open. In early July, they secured a 10-year consent decree forcing Deere to give farmers and independent repair shops the same tools its own dealers use. That is not a vendor dispute. That is an antitrust action.

And the after-the-fact remedy tells you everything. A separate $99 million class action had settled months earlier. After legal fees, each of the 200,000-plus farmers received roughly $395. Less than one dealer service call. The lock cost billions. The refund covered a lunch.

The problem is not that operators pick bad vendors. The problem is that "go with the best vendor" trains you to measure quality at the point of entry. The cost is incurred at the point of exit. That gap is the vendor's margin.

The same pattern runs in your software. SaaS inflation hit 13.2% as of early 2026, per the Vertice SaaS Inflation Index. That is nearly two points higher than the year before. And 72% of vendor growth now comes from price hikes on current customers, not from new ones. The vendors are not earning more. They are extracting more.

SaaS cost per employee climbed from $7,900 in 2023 to $9,100 by the end of 2025. That is a 15% rise in two years. A Zylo survey of IT leaders found that 61% of their organizations cut projects because of unplanned vendor cost increases. The lock is the same. The equipment is different.

Your switching cost is the real price of the product. Not the monthly fee. Not the onboarding cost. The cost of leaving. Once that is clear, three moves follow from it.

For every vendor in your stack, write down what happens if you leave in 90 days. Where does your data go? What format is it in? Who owns the integrations? How long does migration take? What breaks during the gap? This is not a thought exercise. It is a price calculation you have been skipping.

Deere's power came from one thing: no other tool could do the job. Look at your own stack the same way. If only one vendor can run a core function, and that vendor controls the data layer, you do not have a tool. You have a lease with no end date. Where a single-source lock exists, build a second path. Even a partial backup shifts the price of leaving.

Most operators review a vendor when the invoice arrives. By then, the switching cost has already been set. Instead, set a calendar trigger 90 days before each renewal. Use that window to price the exit. If the exit cost has grown since the last check, the vendor is tightening the lock. That is your signal, not the renewal email.

Running this for one quarter does something the advice never did:

You see which vendors have made leaving harder since you signed. You see which tools hold your data in formats only they can read. You see where a 10% price hike would force you to cut something else. And you see which vendors made it easy to leave, which tells you who earned the renewal.

At the end of the quarter, ask three things.

→ Which vendor's exit cost went up since the last review?

→ Which renewal looked routine but locked in a new dependency?

→ Which tool could you replace in 30 days if the price doubled?

That is the difference between advice that sounds right and a system that proves itself.

The farmer in Iowa can now read his own fault codes. It took a federal consent decree to make that happen. Your vendors have not had that visit. The question is not whether the lock exists. It is whether you priced the exit before you signed, or whether you are pricing it now.