The Locked Door: Negotiating Your Exit Before You Sign the Entry

Every hosting negotiation begins in the same mood. The commercial team is attentive, the pricing is competitive, and the question of how the relationship might one day end feels impolite. It is also the single most valuable question a UK buyer can ask, because it is the only one that will be answered generously while the contract is still unsigned.

Once the workload is live, the leverage reverses. The provider knows what a move costs in practice, and the customer discovers it only by attempting one.

An Asymmetry Built Into the Contract

Entry to a hosting platform is designed to be frictionless: a signed order form, a migration credit, a technical account manager, onboarding calls. Exit is not designed at all, or is designed by the billing department. The result is an asymmetry that most procurement processes fail to examine, because the evaluation criteria usually stop at price, performance and support responsiveness.

Four cost centres account for most of the surprise.

Egress: The Meter That Only Runs One Way

Data transfer into a provider's network is commonly free or generously bundled. Transfer out is frequently metered, and migration involves transferring everything out at once. For an application serving modest traffic but holding years of large media objects, logs, snapshots and backups, the egress bill for a single migration can exceed several months of normal charges.

Watch for the definition. Is egress measured at the network edge, at the application gateway, or per service? Are replication and backup traffic counted? Is there a credit for migration, and is it capped? Is the rate fixed for the term or indexed to a price list that can move?

The Migration Project Nobody Budgeted For

The contractual exit cost is easy to quantify. The internal one is not, and it is usually the larger figure. It includes re-architecting against a different network model, retesting integrations, rebuilding monitoring and alerting, retraining staff and running two environments in parallel during the cutover window.

A useful discipline is to size this before signing, not after. Ask the prospective provider to describe a migration off their own platform — genuinely, in steps — and note where they hesitate. The hesitation tends to sit where the proprietary components are.

Tooling That Does Not Travel

Platform-specific functions are the quietest form of lock-in, because each one is individually modest and collectively decisive. Managed queue services, serverless functions, proprietary load balancers, built-in secret stores, bespoke backup formats, log pipelines and identity integrations all create dependency. None is inherently wrong; the question is whether the dependency is documented and priced.

For each managed service in the proposed architecture, ask three things: what format the data can be exported in, whether that format is open and complete, and how long the provider will continue to support the export after notice is served. A commitment to "data portability" is meaningless without a format.

Notice Periods, Auto-Renewal and the Calendar

Commercial terms frequently include a 60- or 90-day notice period, with renewal dates fixed to the original signature. Miss the window by a week and the customer is committed for another full term. In practice, the exit from a hosting contract is often delayed not by technology but by a diary entry nobody owned.

Three details deserve attention: the notice period relative to the term length (90 days on a monthly rolling contract is a contradiction), whether auto-renewal is opt-in or opt-out, and whether prices may be varied mid-term with only a notice period as protection. A change-of-control clause is also worth negotiating — if the provider is acquired, the customer should have an exit rather than inheriting a new owner's roadmap.

Nine Questions for the Pre-Contract File

Put these in writing and keep the answers:

1. What is the notice period, what is the minimum term, and how does auto-renewal operate?
2. What are the egress rates for migration traffic, and can a one-off migration allowance be written in?
3. In what formats can data be exported, and how quickly is an export delivered on request?
4. Which services use proprietary interfaces, and what is the supported path off each of them?
5. Is there a documented exit assistance or transition service, and is it chargeable?
6. Are internet addresses, DNS zones and TLS certificates portable, and who controls them?
7. How long is the post-termination grace period for data retrieval, and what happens to the data afterwards?
8. How may prices change during the term, and on what notice?
9. What happens to terms on a change of control at the provider?

A provider confident in its service will answer all nine. One that deflects questions three and four is telling you where the wall is.

Bringing the Exit Into the Evaluation

Procurement teams routinely score suppliers on onboarding support and give the leaving experience no weight at all. A simple amendment corrects this: score the exit plan. A written transition procedure, a fixed egress allowance and a defined export format are worth more to most UK businesses than a modest discount on the first year.

Lock-in is not inherently sinister; it is a commercial position, and it is negotiable at exactly one moment in the relationship. After that, it becomes leverage — someone else's.