Real-Time CDP is where Adobe engagements become data programs. The work is schema design, identity resolution, governance policy and pipeline engineering — disciplines closer to a data-platform build than a marketing tool rollout — and the partner market has not fully caught up. Plenty of firms carrying strong AEM or Analytics credentials will happily sell CDP implementations they are structurally unequipped to deliver. Your evaluation has to separate data-engineering substance from Experience Cloud adjacency.
The four competencies that matter
Identity architecture. The entire value of a CDP is the unified profile, and the unified profile is only as good as its identity graph. Probe how candidates design identity namespaces, choose primary identifiers, handle the anonymous-to-known transition, and prevent profile collapse (the classic disaster: shared devices merging distinct customers). Ask for a war story about an identity design that went wrong — anyone who has shipped real CDP work has one.
Schema and source discipline. Experience Data Model (XDM) schema design decides whether your CDP stays maintainable. Strong partners run structured workshops mapping your sources — web behavior via the Web SDK, CRM, transactions, service systems — into field groups with clear ownership, and they push back on ingesting data nobody will activate. "Land everything, decide later" is how CDPs become expensive swamps.
Governance as design, not paperwork. Data-usage labels, consent enforcement, purpose-based marketing policies — Real-Time CDP has genuine machinery for privacy governance, and regulated industries buy the platform for it. A partner who treats governance as a final-week checklist item disqualifies themselves; the labels shape the schema, so they belong in week one.
Activation economics. Profiles that never reach a destination are pure cost. Insist the engagement be scoped around named activation use cases with owners and success metrics: suppression of converted customers from prospecting (usually the fastest payback), cart-abandonment journeys, high-value lookalikes, churn-risk save campaigns. Three shipped use cases beat thirty ingested sources.
Sequencing that survives contact with reality
The implementations that work run in vertical slices: one source, one identity join, one audience, one destination, one measured outcome — then widen. The implementations that stall run horizontally: ingest all sources first, model everything, activate someday. Ask every candidate to sketch their week-by-week plan for a first slice; if profile-to-activation takes longer than a quarter, the plan is horizontal no matter what they call it.
Costs and the license conversation
Implementation engagements for a focused first phase (three to five sources, core identity design, two or three activation use cases) typically run $80,000–200,000 over three to five months. Enterprise programs with complex identity landscapes and governance requirements reach $250,000–500,000+. Expect an ongoing data-operations need — schema evolution, pipeline monitoring, audience management — whether staffed internally or as a $5,000–20,000 monthly retainer. A good partner will also pressure-test license sizing: profile counts and segment-evaluation frequency drive Adobe's pricing, and over-licensed CDPs are common.
Interview probes
Ask how they would handle a 40% anonymous traffic share in identity design. Ask what they exclude from profile enablement and why (profile-enabled data has cost and governance weight). Ask which destination they would activate first for your business and what metric proves it worked. And ask who on the named team has written an XDM schema in the last quarter — the honest answer sorts implementers from supervisors quickly.
The pattern across all of it: buy outcomes per quarter, not platform ambition. Shortlist firms with documented Real-Time CDP delivery, make them scope a first vertical slice with a measurable result, and let the quality of that plan decide.
Frequently Asked Questions
How much does a Real-Time CDP implementation cost?
A focused first phase — a handful of sources, identity design, two or three live activation use cases — typically costs $80,000–200,000 over three to five months. Enterprise programs with complex identity graphs and governance scope run $250,000–500,000+, plus ongoing data operations.
What skills should a Real-Time CDP partner have?
Data engineering first: XDM schema design, identity-graph architecture, pipeline and API work, and privacy governance design — alongside activation strategy that connects profiles to journeys and ad platforms. Experience Cloud familiarity alone, without data-platform depth, is insufficient for CDP work.
How long until a CDP shows business value?
A well-sequenced program activates its first measurable use case — commonly suppression of existing customers from acquisition media, or cart-abandonment journeys — within 10–16 weeks. If a proposed plan shows no live activation inside a quarter, it is sequenced for platform completeness rather than value.
What causes CDP implementations to fail?
The recurring causes: identity designs that collapse or fragment profiles, ingesting sources without activation owners, governance retrofitted after schemas are built, and horizontal sequencing that spends quarters on ingestion before any use case ships. Partner evaluation should probe each of these directly.