Skip to content
Operations desk with laptop displaying CI/CD pipeline, a labelled 'Recovery Runbook' binder and a neutral third‑party escrow token, representing code and data escrow for supplier reversibility.

Insights

Data and code escrow as delivery‑grade mitigations: what boards must require before approving supplier‑managed services

Antares Consultancy

Executive decision at stake: should the board or Senior Responsible Owner (SRO) require active data, configuration and code escrow (including verification and live recovery tests) as a condition of funding or contract award for supplier‑managed critical services?

Why escrow matters now for supplier‑managed services

Public bodies increasingly accept supplier‑managed platforms (SaaS or hosted) for essential services. This creates a real delivery risk: if a supplier fails, insolvency or prolonged outage can stop a service and block access to the data and operational artefacts needed to continue or migrate the service. Escrow converts a fuzzy commercial control into an executable delivery capability — provided it is scoped, verified and exercised as a technical and operational artefact, not just a signed contract clause.

Government procurement guidance and the Sourcing Playbook already identifies escrow as a mitigation option for critical software and technology assets. The NCSC’s resilience work and CAF v4.0 make supply‑chain and recoverability material to operational risk decisions. Boards approving supplier‑managed designs therefore confront a binary governance trade‑off: accept residual operational risk (with contingent legal remedies) or require an escrow package that can demonstrably restore function and data access within agreed tolerances.

Why escrow fails as a board control

Most escrow fails because it is treated as a legal relic: a sealed archive stored by a third party with release triggers defined using legal language. That model misses the delivery problem: restoring a complex cloud service requires more than source code — it requires build artefacts, infrastructure as code, configuration, deployment pipelines, database schema migration scripts, sampled data extracts, operational runbooks and signed verification evidence. Without those, a release is a dead letter.

Common operational gaps are (a) incomplete scope (only source code, no IaC or environment manifests), (b) no regular deposit or versioning, (c) no independent verification that deposits rebuild or restore the service, (d) legal release conditions that are impractical to execute under time pressure, and (e) no exercised recovery‑run to show staff and suppliers can perform the recovery in an acceptable timeframe.

What the board must require — precise artefacts and acceptance evidence

Escrow scope: require deposits to include (at minimum) source code (where applicable), infrastructure‑as‑code (IaC), deployment pipelines (CI/CD scripts), configuration manifests, database schemas and migration scripts, data extract procedures and a sample sealed dataset for verification, encryption keys (or escrowed key‑release plan), third‑party licences and supplier runbooks describing operational dependencies, cross‑supplier interfaces and emergency access procedures. For SaaS, require 'SaaS escrow' packages: environment build scripts, container images or manifest exports, and data export/migration tooling.

Verification artefacts: every deposit must be accompanied by an independent verification report from a neutral escrow verification provider. Verification must include a reproducible build/deploy using the supplied artefacts into an isolated test environment and a functional smoke test suite. The verification report must be machine‑readable (hashes, timestamps) and versioned against the supplier’s release cadence.

Release triggers and escalation: define operational release triggers (supplier winding‑up, prolonged SLI breaches beyond contractual cure periods, bankruptcy events, or unilateral cessation of support) but also define an 'operational release path' for time‑critical partial access (e.g. data exports or temporary admin credentials) with pre‑agreed legal and technical steps and SLAs for escrow agent response. Make release triggers executable (example: supplier enters an insolvency process AND has not restored SLI in 10 business days), not discretionary.

Recovery runbooks and playbooks: deposits must include a 'Rebuild and Failover Playbook' that maps roles, responsibilities, run commands and test data; the playbook must list supplier and sub‑supplier contacts, credentials release processes, configuration parameter values and mitigation actions for the most likely failure modes (database corruption, identity compromise, degraded third‑party APIs).

Operational tests: Boards should require live recovery exercises before final acceptance — at a minimum a verified 'rebuild in an isolated environment' and a full 'data export, restore and reconciliation' run for a representative production slice. These runs must be witnessed by the buyer’s delivery team, the escrow verifier and the supplier, and their results retained as evidence.

Data and code escrow as delivery‑grade mitigations: what boards must require before approving supplier‑managed services: editorial image for What the board must require — precise artefacts and acceptance evidence

Trace through the transformation lifecycle

Discovery and business case: explicitly cost and evaluate escrow options as part of option appraisal. The Green Book expects appraisals to show delivery risks and contingencies; escrow is a measurable mitigation and should be compared to insurance, redundancy, or in‑sourcing options.

Procurement & contracting: include escrow scope, verification frequency, verification acceptance criteria, escrow agent SLAs, release triggers and cost allocation in the procurement documents. The Sourcing Playbook flags escrow as an option; make it prescriptive for high‑risk suppliers rather than optional.

Build & delivery: require deposit automation in the CI/CD pipeline (every major release automatically deposits artefacts with the escrow agent). Require independent verification reports on every major release before it moves to production.

Cutover & acceptance: include rebuild and data‑restore runs as gates in the acceptance pack — boards should not accept live transition without passed escrow verification and an executed recovery run.

BAU & contract exit: escrow maintenance must be a live contract deliverable with scheduled re‑verification, version reconciliation checks and annual recovery rehearsals. The contract must name an alternative beneficiary (multi‑beneficiary escrow) if the contracting authority is part of federated services.

Data and code escrow as delivery‑grade mitigations: what boards must require before approving supplier‑managed services: editorial image for Trace through the transformation lifecycle

Failure modes and mitigation — what to watch for

Incomplete deposits: verify that the deposit contains the exact commit/version referenced in production and the complete dependency graph. Mitigation: insist on automated deposit metadata (hashes, tags) and verifier checks.

Escrow agent delays or lock‑in: some agents or jurisdictional arrangements can slow release. Mitigation: use reputable escrow agents with ISO27001/KYC, pre‑agreed expedited legal escalation and onshore storage where regulatory reasons require it.

Data privacy/legal constraints: some datasets contain personal or sensitive information that cannot be escrowed verbatim. Mitigation: deposit anonymised, transformable export procedures and procedural documents (how to generate a lawful extract), combined with verified tooling that can execute the export under controlled governance.

Vendor resistance on commercial grounds: suppliers may resist heavy escrow terms as a cost. Mitigation: evaluate escrow cost into the total cost of ownership; use risk‑based thresholds for escrow requirements and consider shared escrow across consortia to reduce per‑buyer cost.

Require escrow as a delivery artefact: scope it, automate deposits, verify builds and exercise recovery — or accept that legal remedies alone will not restore a failed public service.

Antares recommended actions

These are Antares's recommended first actions for organisations turning the issues in this article into practical governance and delivery.

  1. 1. Condition funding: require an escrow decision gate in the funding approval: either accept residual risk with explicit quantified tolerances, or mandate an escrow package scoped to the service criticality and verified by a neutral provider.
  2. 2. Define escrow scope by risk tiers: T1 (essential live services) = full SaaS escrow (code, IaC, CI/CD, config, migration scripts, sampleised data); T2 = configuration + data export tooling + runbooks; T3 = data export and documentation only. Use the Sourcing Playbook and Government Commercial Function risk appetite as a baseline.
  3. 3. Automate deposits: mandate that the supplier’s CI/CD pipeline performs automated deposits on every major release and that each deposit is accompanied by a machine‑readable verification report (hashes, test results). If automation is impossible, require a documented, dated manual process and stricter verification cadence.
  4. 4. Independent verification: require an accredited escrow verification provider to run a reproducible rebuild and smoke test for each deposit; accept only verified deposits into the official escrow ledger before awarding acceptance or approving go‑live.
  5. 5. Practical release triggers and operational release path: write release triggers as executable operational conditions with an emergency partial release path (data export, admin access) and legal escalation steps. Test the path in a witnessed scenario before contracting finalisation.
  6. 6. Exercise recovery: require at least one full 'rebuild + data restore' rehearsal with delivery teams and the escrow verifier before programme sign‑off. Include evidence in the acceptance pack provided to the board.
  7. 7. Cost and contract allocation: agree who pays for escrow, verification and recovery rehearsals up front. Where appropriate, use multi‑beneficiary escrow or cost‑sharing across buyers to lower per‑buyer cost while preserving recoverability.
  8. 8. Monitor and re‑verify: insist on scheduled re‑verification (quarterly for high‑risk services, annually for lower risk) and list re‑verification and recovery drills as Key Control Indicators (KCIs) in supplier performance reports.

Our insights are provided for general information only and reflect the position at the date of publication. They do not constitute legal, financial, regulatory, cybersecurity or other advice tailored to your circumstances and should not be relied upon as a substitute for appropriate professional advice.

While we take reasonable care over our content, we do not guarantee that it is complete, accurate or current. Reading our insights does not create a client relationship with Antares Consultancy. To the fullest extent permitted by law, we accept no liability for decisions made or losses arising from reliance on this content. External links are provided for convenience and do not imply endorsement.

If you'd like to discuss your own transformation, we'd be pleased to start the conversation.