Outlook
Part of How to read the customer data platform outlook from dated UK evidence and buyer scenarios
Customer data platforms market outlook: evidence limits
This customer data platforms market outlook sets official UK figures on data handling and AI use against the missing market size, plus buyer scenarios.
No authoritative source reviewed on 6 September 2026 measures the England customer data platform market as a defined revenue, supplier or buyer population. A defensible customer data platforms market outlook therefore starts with an evidence gap, not a calculated total.
The category must also stay narrow. CRM, warehouse or lake, identity resolution, consent or preference management, tags, analytics, activation, clean rooms and managed integration are different jobs. Adding their markets would double-count bundled services and still would not reveal what an England buyer needs.
What to take away
- No authoritative source measures the England customer data platform market as a defined revenue, supplier or buyer population.
- Official ONS indicators describe different populations and cannot be added into a market estimate.
- A market source should only enter the register if it passes a stated admission test.
- Three bounded buyer scenarios describe possible routes without stating likelihood, market share or growth.
- The research owner should report direct market size as not established and keep indicators separate.
What the market looks like without a number
The absence of a defensible total does not leave the market blank. Its shape can be described from the product categories buyers actually meet, even though no source measures their size.
- Traditional CDPsare all-in-one products that unify customer data from multiple sources within a single platform.
- Composable CDPsbuild and maintain customer profiles inside the organisation's existing enterprise data warehouse. Proponents such as Hightouch argue that customer data should stay in that warehouse rather than be copied into a separate platform.
- Infrastructure CDPswork more like data management platforms, supporting upstream operations such as event collection and identity resolution.
- Hybrid CDPscombine traditional and composable elements.
- Marketing cloudsare collections of cloud-based marketing tools, such as Salesforce Data 360 and Adobe Marketing Cloud, in which CDP functions are bundled with other jobs.
The direction of travel is also described, if not quantified. Industry analysts such as Gartner describe the category as moving from basic data collection towards advanced analytics and, in some cases, artificial intelligence. For an England buyer, that is the qualitative outlook that can be stated: more products bundle analytics and AI, and the composable route competes with the all-in-one suite. Which of these routes gains share is exactly what no reviewed source establishes.
What official indicators can establish
The ONS UK business: activity, size and location dataset publishes regional tables for VAT and PAYE based enterprises. It can define a bounded population of registered businesses for a stated year and region. It does not show whether any enterprise has customer data, uses a CDP, plans a purchase or has an activation journey.
The ONS analysis of AI in UK businesses for 2023 to 2026 is UK-wide, drawn from the voluntary Business Insights and Conditions Survey. It identifies exclusions, varying size-band treatment and a headline definition spanning several AI technologies. Its headline figure: self-reported AI use among UK businesses with 10 or more employees rose from around 12% to around 35% since late 2023. In June 2026, 12% of those businesses used data processing with machine learning.
It cannot be converted into an England CDP adoption rate. No England-specific customer-data or CDP adoption indicator is identified here. A buyer can ask whether AI appears in a supplier's documented workflow.
The government's UK Business Data Survey 2024 comes nearest to the pool of possible buyers. It found that 58% of UK businesses handled digitised personal data other than employee data, rising to 90% in finance and insurance. Among businesses with websites that handled personal data, 9% used cookies to acquire personal data. These figures describe who holds customer data, not who buys a platform to manage it.
These indicators are non-additive. They describe different populations and questions. Neither supplies the missing numerator, denominator, price unit or comparable period needed for a market estimate.
Build an admission test for new evidence
A market source should enter the register only if it states:
- The exact CDP function counted.
- The unit countedbuyer, supplier, contract or revenue.
- England as the measured geography, not an inferred share.
- The population and sampling frame.
- Collection dates and weighting.
- How bundled services are treated.
- The revenue basis and period.
- Uncertainty and funding.
CDP source admission test
- Exact CDP function counted
- Unitbuyer, supplier, contract, revenue
- England as measured geography
- Population and sampling frame
- Collection dates and weighting method
- Bundled services treatment
- Revenue basis and period
- Uncertainty and funding stated
If a field is missing, mark the claim unresolved. Do not repair the gap with a broad business count or a vendor forecast.
Three bounded buyer scenarios
Scenario: retain the current route. The existing process may remain preferable if it carries corrections and suppression reliably and supports export. Indicator: the documented problem stays within approved tolerances. Owner: journey owner. Disconfirming evidence: failures that current controls cannot remedy.
Three bounded buyer scenarios
Retain current route
- Owner
- Journey owner
- Trigger
- Problem within tolerances
- Stop signal
- Controls cannot remedy
- Evidence needed
- Corrections and export
Test one function
- Owner
- Data steward
- Trigger
- One defined failure
- Stop signal
- Cannot explain uncertainty
- Evidence needed
- Reproducible fixture
Managed integration
- Owner
- Procurement lead
- Trigger
- Service route enquiry
- Stop signal
- Responsibility unclear
- Evidence needed
- Scope, roles, exit
Scenario: test one specialised function. A buyer may examine identity resolution or preference propagation without buying a broad suite. Indicator: one defined failure has an owner, eligible population and reproducible fictional fixture. Owner: data steward. Stop if the candidate cannot explain uncertainty, preserve separate identities or return to the old version.
Scenario: examine managed integration. A buyer may investigate a service route. That is not demand until authority, scope, legal entity, data roles, security, accessibility, commercial unit, export and exit are evidenced. Owner: procurement lead. Stop if responsibility becomes unclear or records cannot be retrieved.
No scenario states likelihood, market share or growth.
Watch the legal and guidance context without forecasting it
The ICO's organisation guidance on the Data (Use and Access) Act 2025 (DUAA) says all its data-protection provisions were in force by 19 June 2026. The ICO's separate technology work programme lists guidance at drafting and consultation stages. Buyers should distinguish an in-force legal change from a planned publication.
Neither record forecasts product sales. They support a recheck trigger: reassess the actual journey when final guidance, legislation or relevant facts change.
The decision record
The research owner should report direct market size as "not established" and keep the official indicators in separate rows. Beside each row, record source date, geography, population, unit, limitations, affected buyer question and expiry. A future source changes the conclusion only if it passes the admission test.
The market map above also shows why any future size estimate needs a stated boundary. Each product type would be counted differently:
| Product type | What a size estimate would have to count | Double-counting risk |
|---|---|---|
| Traditional CDP | Licences or contracts for the all-in-one platform | Low, if bundled services are excluded |
| Composable CDP | Only the profile and activation layer, not the warehouse beneath it | High: warehouse spend is a separate market |
| Infrastructure CDP | Event collection and identity resolution services | High: overlaps data management platforms |
| Hybrid CDP | The share of a contract serving the CDP function | Medium: the split is rarely published |
| Marketing cloud | Only the CDP component of a wider suite | High: suites bundle many other jobs |
The useful next step is smaller than a market forecast. Describe one England membership journey, measure its current evidence gap and decide whether a reversible fictional-data test is worth authorising. Procurement remains on hold until reviewers are named for the statistics lead, commercial/procurement lead, privacy lead and legal counsel roles, and accept the source definitions and the buyer case.
Before you act
- Check whether the source states the exact CDP function counted.
- Confirm England is the measured geography, not an inferred share.
- Record source date, population, unit, limitations and expiry.
- Distinguish in-force legal change from planned guidance publication.
- Authorise a reversible fictional-data test only after review.
- Keep procurement on hold until named reviewers accept source definitions.
Common questions
Why can the ONS AI analysis not give an England CDP adoption rate?
The ONS analysis is UK-wide and based on the voluntary Business Insights and Conditions Survey. ONS identifies exclusions, varying size-band treatment and a headline definition spanning several AI technologies. It cannot be converted into an England CDP adoption rate, though it can prompt a buyer to ask whether AI appears in a supplier's documented workflow.
What must a market source state to pass the admission test?
It must state the exact CDP function counted and the unit as buyer, supplier, contract or revenue. It must also state England as the measured geography, the population and sampling frame, plus collection dates and weighting. It must cover bundled services, revenue basis and period, uncertainty and funding. If a field is missing, mark the claim unresolved.
What should the research owner report about direct market size?
The research owner should report direct market size as not established and keep the official indicators in separate rows. For each row, record the source date, geography, population and unit. Also record limitations, the affected buyer question and expiry. A future source changes the conclusion only if it passes the admission test.
