How to review customer data platform routes in England: five routes, six gates and a fixture test

For a fictional England membership organisation, the decision begins with a defined sequence: an account source, an identity decision, then an approved audience sent to a controlled destination. The practical question is whether each available route can support that journey with dated evidence, not whether a provider has the broadest product description.

What exactly is the thing under review?

Review one named legal entity, exact UK service and edition, its function, evidence date, and the England journey. Naming these boundaries makes it possible to tell what is being tested and prevents a product label from standing in for the work actually proposed.

A CRM, warehouse, identity resolution, consent management, tags and device storage are not interchangeable; neither are analytics, activation and managed integration. A suite may combine several functions, but its label does not identify which legal entity performs each one. Check the possible contracting party against current first-party terms.

A free company record on GOV.UK can show the registered address, incorporation date, current and resigned officers, previous company names and insolvency information. It can help identify a possible contracting party, but it does not establish product ownership, availability in England or service quality.

Which routes merit the same comparison?

Compare the same defined journey across five routes: manual work, existing approved systems, a buyer-built route, a hosted edition, and managed integration. The route comparison lays out those options against common fields, so differences do not come from changing the scenario between candidates.

Record function, data roles, internal work, commercial units, change exposure, evidence, restore and exit for every candidate. Keep labels secondary to what the route actually does and who supplies each part. A selection checklist can help preserve the service, edition and journey boundaries while assembling the comparison.

What belongs in the candidate evidence file?

Use a dated dossier built from first-party records for the service, edition and UK availability. It should trace inputs and schema, identity logic, correction, objection, suppression and deletion states, alongside data roles, subprocessors, locations, security access and logs.

The ICO controllers and processors guidance describes a controller as the body deciding the purposes and means of processing, and a processor as one processing personal data on behalf of a controller. The page says it is under review. Contract wording alone cannot settle the roles, so describe the work each party performs.

Record where personal information is sent or made accessible outside the UK. People can lose the protection of UK data protection law in those circumstances, and each restricted transfer needs a mechanism such as UK adequacy regulations, appropriate safeguards or an exception. A location statement is one input to the buyer assessment, not a substitute for recording the transfer and its basis.

What can a fixture actually establish?

One fixture shows whether a declared setup produces its expected result on a particular date. It cannot establish production-scale performance or legal suitability. Use fictional data or records authorised for the exercise, and state expected outcomes and acceptance evidence before running it.

A predeclared run can cover identity separation and duplicate handling, stale and contested fields, correction and objection, marketing suppression, deletion, refused device tracking, revoked privileged access, a configuration log event, export and isolated restore. A fuller fixture and dossier method sets out how to capture the test evidence.

Keep the edition, configuration, version, output and reviewer together, so a later reader can see what was actually exercised. A pass applies only to the recorded setup and date. Do not treat it as evidence of production-scale performance or as a decision that the setup is lawful.

Can a pass cancel a failed gate?

No. A failed or unresolved gate must stay visible rather than being offset by strength elsewhere. Keep privacy and PECR, identity accuracy and rights, security, accessibility, contract, and export or exit as separate pass, fail or unresolved decisions.

The market-entry gates and quality checks describe these checks in greater depth, but retain separate records for each result. A functional-fit score may be defined by the buyer; it cannot turn a failed gate into a pass or make an unknown fact known.

Device technology needs a separate decision. The ICO storage and access technologies guidance was finalised on 29 April 2026 after two consultations. A consent field alone does not show that each collection and activation step is permitted; record a decision for every device-storage step in the journey.

How should unlike prices be recorded?

Keep commercial evidence on shared fields rather than reducing it to one averaged monthly amount. Profiles, events, destinations, storage and seats are unlike units, so a comparison needs to show the unit each price actually measures.

Record the source date, currency, VAT presentation, billable unit, measured quantity, billing period, minimum term, mandatory additions, support tier, overage rule, renewal and exit charge. If one field is absent, mark it unknown rather than estimating it, and keep units side by side using the pricing models. Ask a qualified accountant to confirm VAT treatment.

What supports an agency review?

Evidence of agency work needs a file for the exact legal entity and contracted service, including staff access, subcontractors, tools, client-authorised project records and permissioned references. A directory profile or company record does not establish delivery quality.

A case study remains a supplier claim unless the evidence supports it. If the reviewer did not observe the work, use the vendor due diligence dossier rather than presenting a pseudo-review. State which records are missing and withhold conclusions about expertise, service levels or outcomes where the material does not support them.

When should a verdict be revisited?

Date the verdict and name its reviewer, candidate, edition, journey, fixture version, failed and unresolved gates, conflicts, decision and expiry trigger. Revisit it when the evidence changes, rather than carrying forward a result with no current basis.

A supplier change, new subprocessor, changed price unit, security incident, new model version or changed buyer purpose can make the evidence out of date. Keep the trigger with the dated record so a later reviewer can see why the earlier decision may no longer apply.

Keep excluded candidates and the reason for each exclusion, so an evidence gap is not confused with a failed test. Keep production activation off until every gate passes, and consult delivery operations when recording that hold.

What should I do first?

Set the review boundaries and journey before requesting comparisons, then collect dated evidence and declare the fixture outcomes. These steps keep supplier gaps separate from tested results.