Canny

Every personalization promise an agency makes collapses if the client's customer data sits in six disconnected systems, because segmentation, journey logic, and reporting all inherit that fragmentation. The CDP decision is therefore not a tooling choice but a margin choice: heavyweight suites and composable warehouse-native layers carry different cost curves, and picking the wrong weight class burns retainer hours on plumbing instead of campaign work.

Profile Persistence Threshold is the point at which a client's unified customer record stays accurate long enough to justify real-time activation. Below it, segments decay faster than campaigns ship: a retainer built on weekly batch syncs cannot support same-day lifecycle triggers, so personalization claims outrun delivery. Above it, identity resolution holds across devices and channels, and every downstream channel inherits the same truth. The framework forces an agency to ask one question before scoping: how long must a profile survive to make the promised journey work? A composable route (Hightouch, DinMo, RudderStack) keeps profiles in the client's warehouse, so persistence is bounded by the client's own data model. A full-suite route (Braze, Insider One, Twilio Segment) owns persistence but adds migration cost. Forrester's September 2026 finding that private AI deployments outperform shared public models applies directly: profile depth is the differentiation clients cannot rent from a competitor.

Activation Surface Ratio measures how many distinct destinations a unified customer profile actually reaches, not how many the platform claims to support. A CDP that unifies 40 million profiles but activates into two channels is a reporting tool; one that pushes the same profile into ad platforms, email, CRM, and warehouse reverse-ETL is revenue infrastructure. For agencies, this ratio determines whether personalization promises survive contact with delivery: a retainer built on lifecycle segmentation collapses if the client's stack only exposes email. The ratio also predicts governance load, since every added destination is another place consent and identity resolution must hold. Composable tools such as Hightouch and DinMo push warehouse segments into 300+ and ad-platform destinations respectively, while full-suite platforms like Braze and Insider One bundle activation inside their own journey engines. Audit the ratio before signing scope, because the gap between unified and activated is where agency margin quietly disappears.

Warehouse Gravity Well is the pull a client's existing data warehouse exerts on every CDP decision. When the warehouse already holds clean identity and event data, composable tools that read from it (Hightouch, DinMo, Jitsu) activate segments in days, while full-suite platforms (Braze, Insider One) require re-ingesting that data into a second store. The framework asks one question before any demo: where does the client's authoritative customer record already live? If it lives in Snowflake or BigQuery, a composable layer wins on speed and governance; if it lives nowhere, a full-suite platform earns its premium by supplying the profile store itself. For agencies, this decides retainer scope: composable work is often a fixed build plus a smaller monthly activation fee, while full-suite work carries a larger recurring license the client will scrutinize. Misreading the gravity well means selling a second warehouse the client never needed, or under-delivering personalization because no profile store existed to begin with.

Choose the CDP by the client's data ownership model and journey complexity first, then by feature list, because a composable layer and a full-suite platform solve different problems and rarely substitute for each other.

Audit where the client's cleanest customer record already lives, then buy the activation layer that reads it rather than a platform that duplicates it.

IF a client already runs a governed warehouse (Snowflake, BigQuery, Redshift) and the primary need is pushing segments into ad platforms and CRM tools, THEN a composable CDP layer is the lower-cost path because it activates data where it already lives. IF the client needs multi-channel lifecycle orchestration (email, SMS, push, WhatsApp) owned by a marketing team without engineering support, THEN a full-suite engagement platform is the correct build despite higher seat and volume costs. The wrong pick shows up as either paying for journey orchestration nobody uses or under-delivering on personalization the retainer promised.

A fixed-scope engagement that stands up a warehouse-native customer profile and activates two live segments into paid and owned channels, giving agencies a repeatable entry point into the customer data platform category without a full platform migration.