Leave the Keys: The Economics of Not Being Locked In

The fourth of Ariadne's principles is "leave the keys": you receive source code, documentation and full account ownership. Clients occasionally read that as a generous gesture. It is not generosity. It is the correct response to a well-documented market structure.

What lock-in is worth to the vendor

Shapiro and Varian state the definition plainly: "When the costs of switching from one brand of technology to another are substantial, users face lock-in" Shapiro & Varian 1998. Their central analytical result, from chapter 5, is the one worth memorising:

Read that as a diagnostic. If a vendor's profit on you exceeds their honest advantage over the alternative, the difference is switching cost — and switching cost is a transfer from you to them that produces nothing.

They also lay out the lock-in cycle in four stages, which is the shape every marketing-stack entanglement I have unwound follows.

The archetypal case Shapiro and Varian cite is Bell Atlantic's investment of roughly $3 billion in AT&T 5ESS digital switches in the mid-to-late 1980s. The switches were fine. The position afterwards was not.

The formal result

Paul Klemperer's Review of Economic Studies survey has a section titled, without hedging, "Switching costs yield monopoly power" Klemperer 1995:

The most obvious effect of switching costs is to give firms some market power over their existing customers, and thus to create the potential for monopoly profits.

In a mature duopoly with homogeneous products where each firm has an installed base, if the switching cost $s$ exceeds the margin a rival could offer, both firms price at the reservation price $R$ — the pure monopoly outcome, in a market with two competitors. The effect weakens with many firms or asymmetric shares, because a firm with a small share has little to lose from cutting price.

Farrell and Klemperer's Handbook of Industrial Organization chapter is the authoritative survey, and states the welfare consequence: lock-in hinders customers from changing suppliers in response to changes in efficiency Farrell & Klemperer 2007. In plain terms: even when a better option appears, you rationally stay with the worse one. The market stops disciplining quality.

How big is it, actually

Most writing on this topic is theory or assertion. Oz Shy derived switching costs from observed prices and market shares and applied the method to two real markets Shy 2002.

Finnish demand-deposit banking, 1997, four largest banks, in USD, lifetime fee sums discounted at a 4% real rate:

Israeli cellular, 1998: Pelephone NIS 1,298 and Cellcom NIS 945, against handset prices of NIS 700–1,400.

Note bank 4: $23, effectively zero. Switching costs are not a law of nature. They are a design choice made by the supplier, and some suppliers choose not to impose them.

What lock-in looks like in a marketing stack

The specific forms I find in audits, mapped to Shapiro and Varian's categories:

Contractual. Annual commitments with auto-renewal. The most visible and least dangerous, because you can read it.

Durable investment. Three years of automation logic expressed as a proprietary visual workflow that exists only inside one vendor's product. Nothing exports. The logic is yours; the expression of it is not.

Information and databases. Your customer history living in a schema you cannot query, exportable only as a flattened CSV that drops the relationships.

Search costs and learning. Your team knows one tool. Retraining is real money, and vendors price with this in mind.

What "leave the keys" commits me to

Four concrete deliverables, each aimed at a specific mechanism above:

1. Source code you own, in your repository, under a licence that is yours. This makes durable investment portable.
2. Every account in your name, with you as owner and me as a delegate that you can remove. This eliminates the hostage case entirely.
3. Data in an open, relational, exportable form — the relationships as well as the rows.
4. Documentation written for a successor. The explicit test: could another engineer take this over without speaking to me?

That last one is the honest measure of the principle, and it is deliberately against my own short-term interest. Shapiro and Varian's equation says exactly why: giving up $S$ means the only thing I can be paid for is $A$ — the actual advantage of my doing the work rather than someone else.

That is the correct arrangement. If I am not better than the alternative, you should leave, and it should be cheap.

References

1. Shapiro, C., & Varian, H. R. (1998). Information Rules: A Strategic Guide to the Network Economy. Harvard Business School Press (Ch. 5, 'Recognizing Lock-In', pp. 103–134). https://www.hbs.edu/faculty/product/863XTrade/academic book, not peer-reviewed; the analytical content restates the authors' and Farrell's refereed work.
2. Klemperer, P. (1995). Competition when consumers have switching costs: An overview with applications to industrial organization, macroeconomics, and international trade. The Review of Economic Studies, 62(4), 515–539 (quote at p. 519). https://doi.org/10.2307/2298075
3. Farrell, J., & Klemperer, P. (2007). Coordination and lock-in: Competition with switching costs and network effects. In M. Armstrong & R. Porter (eds.), Handbook of Industrial Organization, Vol. 3, Ch. 31, 1967–2072. Elsevier/North-Holland. https://doi.org/10.1016/S1573-448X(06)03031-7
4. Shy, O. (2002). A quick-and-easy method for estimating switching costs. International Journal of Industrial Organization, 20(1), 71–87. https://doi.org/10.1016/S0167-7187(00)00076-X