Moats are broken from the inside, and the breaking move is always extraction
BMW replaced its own tagline. Unity changed the pricing unit. Delta tied status to card spend. Forty-two companies on record, and only four moats were lost — every one of them to the company that owned it.
Nobody in this record lost a moat to a competitor
Forty-two moat decisions are on record and 29 were never reversed. Of the four that were, none was taken by a rival with a better product. Each one was undone by the company that owned it, and each time the move that undid it was an attempt to get more out of the thing that made the company defensible.
BMW ran 'The Ultimate Driving Machine' for decades, replaced it with the 'Joy' campaign in 2009 because it decided it had a better idea, and restored the original by 2012. Unity charged per seat, switched in September 2023 to a per-install runtime fee aimed at extracting more from the successful games it most needed to keep, and canceled the fee entirely twelve months later. Delta tied SkyMiles elite status primarily to credit-card spend in September 2023; its chief executive acknowledged the change had probably gone too far, and Delta walked key provisions back within months. Etsy let outside manufacturers into a marketplace whose premise was handmade authenticity; the policy stood, the network grew, and Etsy is now cracking down on the inauthenticity it allowed while gross merchandise sales fall.
The extraction always looks reasonable on paper
Broadcom killed VMware's perpetual licenses after the acquisition, and the record is unusually direct about the reason: having paid far more than the assets were worth and booked the goodwill, Broadcom needed revenue fast from a captive base. Subscription revenue spiked 275% year over year. Around half of VMware's midmarket base had let licenses lapse.
That is what the trade looks like when it is measured properly. The extraction works, in the quarter you measure it. What it costs does not appear for two or three years, and it appears in a different line.
What the durable ones protect
The moats in this record that were never touched share a property that has nothing to do with strength. The company knew which specific thing it was not allowed to monetize.
A moat is not a barrier, it is a reason a customer does not leave. Every one of those reasons has a price at which the customer will reconsider, and the company is usually the only party in a position to find that price. The useful question is not how wide the moat is. It is which part of it you have decided is not for sale.
How to run this decision
Drawn from what separated the records that stuck from the ones that were undone. A step nobody in the record can be seen taking is not here.
- 01
Write down the reason your customer does not leave, in one sentence
Not the feature list and not the switching cost. The reason. BMW's was a claim about how the cars drive that had been made consistently for thirty years. If the sentence takes a paragraph, nobody in the company can protect it, because nobody can recognize an attack on it.
- 02
Name the part that is not for sale, before finance asks
Decide in advance which element of the moat is off the table regardless of the revenue case, and write it where the pricing committee can see it. The four reversals here all cleared an internal approval, because at approval time nobody was holding a list of things that were not allowed to be traded.
- 03
Treat a change in the pricing unit as a product change
Unity did not raise prices, it changed what customers were buying — from a seat to an install. Delta did not devalue miles, it changed what earned status. Customers respond to a changed unit the way they respond to a changed product, which is far more violently than they respond to a higher number.
- 04
Price the extraction against three years, not the quarter
Broadcom's subscription revenue rose 275% while half the midmarket base lapsed. Both numbers are real, and only one of them was in the paper that approved the decision. Ask what the base looks like at the next renewal cycle, and at the one after that.
- 05
Find out who inside the company is measured on the moat
Usually nobody, which is why it gets traded. Revenue, margin and growth all have owners with targets. The reason customers stay is typically owned by a brand or product function with no line in the model, so it loses every internal argument on points.
- 06
Put a reversal test on it before you launch
Define the customer reaction that would make you stop, and the date you will read it. Unity took twelve months and a canceled fee. Coca-Cola took seventy-nine days. The companies that recovered cheaply are the ones that had decided in advance what a refusal looked like.
What it looks like when this is going wrong
- The proposal changes what the customer is buying, not what it costs.Unity, 2023. Delta, 2023.
- The revenue case is built on a base described internally as captive.Broadcom and VMware, 2023.
- A long-running brand claim is being replaced because leadership has a better idea.BMW, 2009.
- The premise of the marketplace is being relaxed to let the network scale.Etsy, 2013.
- No one in the room is measured on the reason customers stay.The four reversed moat decisions on record.