Lock-in is when leaving costs more than staying. Unmeasured lock-in is the problem.
Vendor lock-in is when leaving a supplier costs more than staying, even when staying is the worse deal. In software it comes from data that is hard to export, processes shaped around one product, and integrations that only work inside one ecosystem. Some lock-in is unavoidable. Lock-in you have not measured is the problem.
The definition, and the three places it hides
Lock-in is a switching cost. Every supplier has some; the question is how much and whether you knew. The ACCC digital platform services final report (March 2025) described cloud customers facing “high impediments to switching,” including “technical barriers to interoperability, and high egress fees when moving data out of a cloud ecosystem.” The same report put Amazon at roughly 39% of the global infrastructure market, Microsoft at 23% and Google at 8.2%, with combined cloud revenue of US$66.26 billion in the December 2024 quarter.
That is the big-end version. In a small business the mechanism is the same and the numbers are smaller, but there are three distinct places to look.
- KNDR·01Data lock-inYour records are in the product and getting them out is incomplete, slow or expensive. Five years of job history, photos and notes in a job management system is the common case for a trade.
- KNDR·02Process lock-inYour people have learned the product and your procedures assume it. Retraining an office of four is a real cost; retraining a workforce of forty is a project.
- KNDR·03Integration lock-inEverything else you run connects to this product. Change it and every connection has to be rebuilt. This is the one that surprises people, and the one you can design around.
Egress fees: the cloud version, and what changed in 2024
The purest form of lock-in is a fee for leaving. Cloud providers charged for data transferred out, so the more you stored the more it cost to go. That changed under regulatory pressure. Google Cloud announced free egress for departing customers on 11 January 2024, the day the EU Data Act took effect; AWS followed in early March 2024 and Microsoft Azure on 12 March 2024 (CIO Dive, 14 March 2024).
Read the conditions. Google and Microsoft require you to close the account and finish within 60 days; AWS does not require closure but scrutinises repeat requests. Free to leave means free to leave completely, once, on their timetable.
You deal with this indirectly: the SaaS products you rent run on those clouds, and their costs and terms flow down to you.
The small-business version: a product gets retired
Lock-in bites hardest when the vendor leaves you, not when you try to leave. Xero announced on 22 March 2023 that it would retire WorkflowMax and switched it off on 26 June 2024 (Xero blog). Firms that had built their whole job-to-invoice process on it had fifteen months to migrate to BlueRock’s rebuilt version or start again. The data was exportable. The process, the training and the integrations were not.
The quieter version is a price you cannot refuse. Xero’s Australian plans rose on 1 July 2026 — Grow from $75 to $78, Comprehensive from $100 to $107, Ultimate 10 from $130 to $143, and the multi-organisation discount was removed (Plus 1 Group summary of Xero’s 1 July 2026 changes, 17 June 2026). Nobody moves their books over $3 a month. That is exactly why the increase is $3.
And there is the version that reaches you through the add-ons. From 2 March 2026 Xero replaced its developer revenue-share with paid tiers — $35 AUD a month for 50 connections, $245 for 1,000, $1,445 for 10,000 (Xero developer pricing, September 2026). The small add-on you rely on now has a bill it did not have before. Xero vs MYOB compares the two accounting incumbents on what they publish.
What you can actually take with you
Before you sign anything, find out what leaves with you. Xero exports the general ledger, invoices and bills, contacts, inventory, bank transactions and fixed assets as CSV; the audit trail and custom reports do not export directly (Boxkite, 3 March 2026). That is a good outcome by industry standards. It is still not everything.
The API is the other exit, and it is rationed. Xero allows 5,000 calls a day per organisation (Xero developer limits FAQ); ServiceM8 allows 20,000 a day (ServiceM8 developer documentation). A full extraction of a large history through the API is measured in days, not minutes. Plan for that before the day you need it.
MYOB states there are “no lock-in contracts” on its plans (MYOB pricing, September 2026), and that is true and beside the point. Nobody is locked in by the contract. They are locked in by seven years of ledger. Our Xero integration page lists what the API exposes and what it does not.
Zapier and n8n: lock-in at the joins
The connections between your products carry their own lock-in, and it is the one you can design away. A Zap lives inside Zapier. Zapier’s free plan allows 100 tasks a month, its Professional plan starts at US$19.99, and it connects around 9,000 apps (Zapier pricing and developer platform, September 2026). None of that logic can be exported to run anywhere else. Leave Zapier and you rebuild every Zap.
n8n is the counter-example. Its workflows are JSON files. Its Community Edition can be self-hosted from GitHub at no licence cost, and its hosted Starter plan is €20 a month billed annually for 2,500 executions (n8n pricing, September 2026). Move from hosted to self-hosted and the workflows come with you. Zapier vs n8n weighs that against the convenience you give up.
Neither is wrong to start on. The mistake is letting forty business-critical Zaps accumulate without ever costing what it would take to move them.
Custom software can lock you in harder
We build custom software, and this is the part most builders skip. A custom system where the developer holds the code, the hosting account and the only understanding of how it works is worse lock-in than any SaaS product. There is no export button. There is one person, and they might move to Perth.
The test is whether you could hand it to a stranger. The repository in your name. The documentation good enough for a competent developer to pick up. The hosting and credentials under your control, not ours. If a builder cannot say yes to all three in writing, the lock-in is to them. Ownership and security is where we say yes to all three, and you own it is the plain-English version.
We have not published measured client results, and we have built against nine systems, not every system. Neither of those facts changes this one: code you hold is the only software you cannot be locked out of.
How to measure it before you sign
Cost the exit on the day you enter. Four questions, an afternoon: what exports and in what format; what the API allows and how fast; how many other systems connect to this one; and how many hours of retraining a switch would cost at your people’s loaded rate. Write the total down. That is your lock-in, in dollars, and it will grow every year you stay.
The Productivity Commission’s final report on harnessing data and digital technology (summarised by Johnson Winter Slattery, 19 December 2025) recommended making the Consumer Data Right more flexible across industries — the regulatory version of forcing exports open. Do not wait for it. Most business software is not covered and will not be soon.
Then decide with the number in front of you. Some lock-in is worth it: Xero at 1.14% monthly churn (Xero FY26 results, 14 May 2026) is a product people stay with because it works, not only because leaving hurts. The aim is not zero lock-in. It is to know the figure before the vendor does.
The questions people actually ask.
01Every vendor has some lock-in. Why does it matter?
Because the amount decides your negotiating position for the next decade. A vendor who knows you cannot leave prices accordingly — $3 a month at a time. Knowing your own exit cost does not remove it, but it tells you when a price rise has crossed the line, and it tells you which parts of your stack to keep portable.
02Is a custom build the way to avoid lock-in?
Only if you hold the code, the hosting and the documentation. A custom build where the developer holds those is the worst lock-in on this page. Ask any builder, including us, to put the ownership terms in writing before the deposit.
03Should I avoid Zapier because of this?
No. Start on it if it gets you moving. Just count the Zaps every six months, price what rebuilding them would cost, and move the critical ones to something portable — self-hosted n8n or owned code — before the count gets frightening. Lock-in you have costed is a decision. Lock-in you have not is a surprise.
Stop bending.
Start shipping.
Bring the one flow that keeps costing you a Tuesday afternoon. We map it live and tell you what we'd build, what we wouldn't, and what it costs.