Customer data platform strategy
Definition
Customer data platform strategy
A customer data platform strategy is the plan for collecting, resolving and activating customer data in one place, with a lawful basis for every use. It covers identity, consent, ingestion and whether you build the platform or buy it off the shelf.
A customer data platform (CDP) is the store that holds one profile per person, stitched from web, app, billing and service records. The strategy is the set of choices around it: what goes in, who may use it, and why.
This is plumbing and legality, not conclusions. What you decide from the data belongs to segmentation and scoring work; the strategy only guarantees the data is present, correct and lawful to use.
Key takeaways
- A CDP strategy owns identity resolution, consent, ingestion and activation, not the analysis that follows.
- Lawful basis and purpose limitation are design inputs, not a compliance review bolted on at the end.
- Build versus buy turns on how many sources you must join and how quickly you need them joined.
- Without a named owner for data quality, any platform decays within two renewal cycles.
How it works
A CDP strategy works in four moves: ingest the sources, resolve them to one identity, govern what each use is allowed to do, then activate the profile in the channels that act on it.
| Layer | What it does | Where it usually fails |
|---|---|---|
| Ingestion | Pulls records from web, app, billing and support | Batch lag that nobody measures |
| Identity resolution | Merges records into a single profile | Merging two people who share a device |
| Governance | Records consent, purpose and retention | Consent captured but never enforced |
| Activation | Pushes segments into working channels | Profiles no downstream system can read |
Your customer relationship management (CRM) system is normally the anchor source, because it already holds the account structure everything else hangs from. A CDP widens that picture rather than replacing the system of record.
Settle build versus buy early, because it changes who you hire — a bought platform needs configuration and governance people, while a built one needs engineers you have to keep for years.
Identity resolution is the hard part. Matching on a verified email is safe. Matching on device and postal address is faster, and it quietly merges a whole household into one imagined buyer.
Governance is where plans fail silently. Article 5 of the GDPR requires personal data to be adequate, relevant and limited to what is necessary, which rules out ingesting every field a connector happens to offer.
The UK Information Commissioner’s Office goes further on process. Its accountability guidance covers impact assessments and contracts between controllers and processors, both of which belong in the design phase.
Consent is not a one time capture. It has to travel with the record into every downstream tool, and be revocable in all of them on the day the customer asks.
Measure the platform like any other system. Freshness, match rate, duplicate rate and the share of profiles carrying a recorded lawful basis are the four numbers worth reporting monthly.
Activation closes the loop. Data mining and modelling read from the profile, but the strategy’s job is making sure the result can reach a campaign, a dialler or an agent screen.
Almost every deployment is now cloud based, which moves the argument from servers to contracts — residency, retention, and what happens to your profiles on the day you leave.
A marketing operations manager usually runs activation day to day, while a separate data owner signs off what may be collected at all. Splitting those two roles is deliberate.
It also sits beside a collaboration platform strategy in most technology plans — one governs customer records, the other governs how internal teams work together.
Examples
The right customer data platform strategy depends on how many sources must be joined and how regulated the data is. A retailer with ten million profiles and a bank with strict residency rules will land in very different places.
European retail. A multichannel retailer joins till, loyalty and web data to recognise the same shopper in a store and a browser. The join is the project — the platform licence is the easy part of the budget.
Financial services. Banks and insurers usually buy a platform but keep identity resolution in house, because a wrong merge between two customers is a reportable incident rather than a marketing annoyance.
Outsourced service delivery. Offshore teams in the Philippines and Poland increasingly run the ingestion and data quality work on a client’s platform, under the client’s lawful basis, with access scoped to named records.
In all three, the written strategy is short. It is the discipline behind it, kept up over years, that decides whether the platform becomes an asset or just a licence you keep renewing.
Related terms
- Computer telephony integration (CTI): the call layer that consumes CDP profiles at the moment an agent answers.
- Information access: the broader question of who may see which records, which a CDP strategy must answer precisely.
- Customer relationship management (CRM): the system of record for accounts, usually the anchor source rather than the platform itself.
- Data mining: what analysts do with the profile once the platform has assembled it.
- Marketing operations manager: the role that most often owns day to day activation.
FAQ
Is a CDP the same as a CRM?
No. A CRM manages relationships and deals that people update by hand, while a CDP assembles behavioural records from many systems into one profile per person. Most firms run both, with the CRM as the system of record.
Should we build or buy?
Buy when you need five or six standard sources joined this year. Build when your sources are unusual, your residency rules are strict, or you already run a mature data warehouse.
What should the strategy settle first?
Lawful basis and purpose. Deciding what you are allowed to do with a field before you ingest it is cheaper than unpicking the field afterwards. Write the purpose down before any connector is switched on.
Who should own the platform?
Split it: a data owner approves collection and retention, and an operations lead runs activation. One person holding both jobs tends to approve whatever the campaign calendar wants.
How long does implementation take?
Six to twelve months is a realistic first phase for a mid market buyer. Anything faster has usually skipped identity resolution or consent enforcement.
Shortlist data capable delivery partners in the Outsource Accelerator provider directory before you sign a multiyear platform contract.







Independent




