CDP ownership and operating model guide

Who should own a customer data platform?

A CDP needs one accountable business owner, supported by named technical, privacy, data-source, audience, media and measurement owners. Marketing should usually own the customer decision and value. Data or IT should steward the technical foundation. Agencies and specialist partners can operate parts of the capability, but the client should retain purpose, permission and commercial accountability.

Fractyl12 minute readPublished 16 November 2026

The short answer

Give the CDP one accountable business owner who can prioritise customer decisions, approve value and resolve cross-team trade-offs. That person is often in marketing, growth, digital, customer or CRM rather than technology alone.

Then assign named owners for source data, identity, privacy, audience rules, destination use, technical operation and measurement. A CDP is a shared capability, but shared contribution should not mean shared accountability.

Own the customer decision before the platform

The primary owner should be accountable for the decision the CDP improves: who is suppressed from acquisition, who enters a win-back audience, which lead stage changes media treatment or which customers form an acquisition seed.

That owner should approve the use-case priority, expected value, audience definition and decision to continue, fix, expand or retire it. Platform ownership without customer-decision ownership produces infrastructure rather than action.

Why marketing is often the accountable owner

Marketing, growth, digital or CRM teams usually control the audience strategy, media investment and customer treatment that the CDP is intended to improve. They can connect audience operation with an observable commercial outcome.

Marketing ownership does not mean marketing should administer cloud infrastructure, decide privacy questions alone or rewrite source systems. It means the team owns why the capability exists and whether it creates value.

Why data or IT should be the technical steward

Data and IT teams understand architecture, access, integration, security, reliability and upstream system change. They should help determine how approved sources connect, where transformations run, how failures are handled and how the implementation fits the wider data environment.

Technical stewardship is different from product accountability. A technically reliable customer layer can still be commercially unused if no one owns the audience decision.

Privacy and legal set the permitted boundaries

Privacy and legal owners should review the proposed purpose, collection context, notices, permissions, minimisation, retention, access, deletion and destination arrangements appropriate to the use case.

They should not be expected to invent the customer strategy. The business owner must describe the intended treatment clearly enough for the privacy and legal review to assess a specific use, not an abstract technology.

The source-system owner protects customer truth

CRM, billing, enrolment, membership, service and operational teams often maintain the fields that prove customer state. They need to define the values, update behaviour, exceptions and business meaning of those fields.

When a source changes, the owner should know which identity, audience or suppression rules depend on it. A CDP cannot compensate silently for an upstream field that has stopped being maintained.

The identity owner resolves record conflicts

Someone must own the rules that decide whether records belong to the same person, which source wins when values conflict and how likely false merges or missed matches are reviewed.

Fractyl uses visible identity and source-precedence rules. The identity owner may sit in data, CRM or a shared governance team, but the decision should be documented and accountable.

The audience owner defines entry and exit

Each audience needs one owner who can explain its purpose, inclusion rule, exclusions, timing, destinations and review date. That person approves changes and confirms when the audience is no longer useful.

Audience exit is a core responsibility. New customers should leave acquisition, reactivated customers should leave win-back, and ineligible or suppressed records should be handled under the approved rule.

The media owner controls platform use

The CDP can deliver a governed audience, but the media team controls how it is applied inside Google Ads, Meta, TikTok or DV360. It owns campaign configuration, exclusions, bidding, budgets, creative and platform-specific delivery decisions unless another arrangement is agreed.

A delivered audience creates no value if it is attached to the wrong campaign, used inconsistently or overridden by another setting. Customer-data and media teams need one shared operating view.

The measurement owner protects the commercial claim

Measurement ownership may sit in analytics, marketing effectiveness, data, finance or a shared team. It should define the baseline, comparison method, source of truth, outcome, cost and material limitations before activation.

Platform-reported conversion, match rate and audience size are supporting evidence. The measurement owner should prevent operational delivery from being presented as proof of incremental value.

The technical operator keeps the capability running

The operator implements approved identity, audience, destination and refresh rules, monitors synchronisation, investigates exceptions and maintains the agreed change process.

Fractyl can perform this role through a managed model, an agency or client team can license and operate the platform, or responsibilities can be shared. The operator should not approve its own purpose, permission or commercial-value claims without the relevant owners.

An agency can operate Fractyl without owning the client’s data purpose

An incumbent agency may own media strategy and execution. A specialist partner may manage the customer-data layer. D3 may operate Fractyl. Those roles can coexist when responsibilities and access are explicit.

The client should retain accountability for lawful collection, source accuracy, approved purpose, final audience use and commercial decisions. Fractyl does not require the client to move its media account to D3.

Six decision rights to assign before launch

  • Priority: who chooses the first use case and approves expansion?
  • Data: who certifies the meaning, quality and freshness of each source field?
  • Identity: who approves matching and source-precedence rules?
  • Permission: who approves purpose, notices, suppression and destination use?
  • Activation: who creates, approves, changes and applies the audience?
  • Value: who decides whether the evidence justifies continuing the use case?

A practical first-use-case ownership map

  • Accountable business owner: customer or media decision and commercial value.
  • Source owner: field definitions, quality, freshness and upstream changes.
  • Technical steward: architecture, access, reliability and integration controls.
  • Identity owner: match, merge, split and source-precedence rules.
  • Privacy or legal owner: appropriate review of purpose, permission, minimisation and contractual arrangements.
  • Audience owner: entry, exit, suppression, destination and review date.
  • Media owner: platform configuration and delivery treatment.
  • Measurement owner: baseline, comparison, outcome, cost and reporting limitations.
  • Fractyl operator: implementation, synchronisation, monitoring and approved changes.

Use a RACI only after the decision rights are clear

A responsibility matrix is useful when it records real authority. It should identify who is accountable, who performs the work, who must be consulted and who only needs visibility for each key decision.

Avoid a matrix with several accountable owners for the same decision. If no one can approve an audience change or stop an unsafe activation, the operating model is not ready.

Set a governance rhythm around decisions

  • Before launch: approve purpose, sources, identity rules, audience logic, destinations and measurement.
  • During testing: reconcile source populations, exclusions, delivery and platform application.
  • After launch: review failures, exceptions, audience movement and commercial evidence.
  • At an agreed decision date: continue, fix, expand or retire the use case.
  • When a source, rule, destination or owner changes: complete a documented impact review before relying on the new state.

Common ownership failures

  • Technology owns the platform, but no business team owns a customer decision.
  • Marketing owns the ambition, but no source owner can validate the data.
  • The agency receives an audience, but no one confirms how it is used in campaigns.
  • Privacy review happens after the data and destination have already been selected.
  • Audience rules change through informal requests with no approval trail.
  • Operational metrics are reported, but no owner is accountable for commercial value.
  • Several teams are consulted, but no one has authority to stop or retire the use case.

Choose the Fractyl operating model around capability

In a managed model, Fractyl can operate the infrastructure, audience rules and destinations while the client and its agency retain their business, privacy, source, media and measurement responsibilities.

In a licensed model, an agency or client team operates Fractyl with onboarding and support. In a shared model, the work is divided explicitly. The right model depends on internal capability, desired control, available time and the consequences of an operating error.

What Fractyl supports today

Fractyl currently creates and maintains persistent unified customer records, resolves identity, applies visible audience and suppression rules, and synchronises governed audiences to Google Ads, Meta, TikTok, DV360 and supported third-party data clean rooms.

CRM connectivity is live for HubSpot, Microsoft Dynamics and Salesforce. Google Cloud and BigQuery sit at the centre of the data foundation, including log-level GA4 data. Secure synchronisation can also be scoped for Snowflake and data environments in AWS or Azure.

Keep beta and roadmap ownership separate

Media mix modelling is in beta. It requires its own model owner, assumptions, validation and decision process. It should not be treated as automatic proof that an audience use case created value.

AI-assisted audience segments and attribution feedback are product direction rather than current live capability. Ownership of automated recommendations, approval and override will need to be defined before those capabilities are used operationally.

Questions to settle in the first workshop

  • Which customer decision is the first scope?
  • Who is accountable for its value?
  • Which fields prove audience entry and exit?
  • Who owns each source and can resolve exceptions?
  • Who approves identity and source-precedence rules?
  • Who confirms the permitted use and destinations?
  • Who applies the audience inside each media platform?
  • Who monitors synchronisation and investigates failure?
  • Who owns the baseline, outcome and cost?
  • Who has authority to pause, change or retire the use case?

The commercial decision

The strongest owner is not the team with the most technical access. It is the person who can connect customer treatment, operating responsibility and commercial value, then make a decision when the evidence is incomplete or teams disagree.

Give that person clear accountability. Surround them with named specialists and an operating model proportionate to the first use case. Expand only when the ownership works as reliably as the technology.

Bring the audience problem. We’ll pressure-test the fit.

Request a working demo