Most CDP evaluations start with features. They should start with a number: how many weeks until your first activated segment generates attributable pipeline?
I've watched three PE-backed marketing teams blow six figures on customer data platforms that never made it past the "unified profile" stage. The profiles existed. The segments existed. The activation to anything that moved revenue? Still waiting. The CFO eventually asked the question nobody wanted to answer: "What did we buy, exactly?"
The 2026 CDP market has split into two architectures, and picking the wrong one for your operating model will cost you a year of implementation time and a mid-six-figure write-off. Bloomreach's analysis frames this as "Marketing CDPs vs. Infrastructure CDPs," and that distinction matters more than any feature comparison you'll read.
Two Architectures, Two Operating Models
Marketing CDPs build data unification directly into the execution layer. Profiles, segments, and campaign orchestration live in one system. Data flows from ingestion to activation without moving between tools. Bloomreach, Klaviyo, and Insider One take this approach. The benefit is speed: a marketer can build a segment and trigger a campaign without filing a ticket with data engineering.
Infrastructure CDPs focus on the data layer itself: collecting, cleaning, transforming, and routing customer data to separate activation tools. Segment, Hightouch, and Tealium operate here. The benefit is flexibility: you're not locked into one vendor's execution stack.
The CFO question is different for each. For marketing CDPs, it's: "Can your team actually use the native channels, or will you still need three other tools?" For infrastructure CDPs, it's: "What's the total cost of the CDP plus every downstream system it feeds?"
Improvado's B2B evaluation puts implementation timelines at 2-4 weeks for warehouse-native tools like Hightouch, versus 3-12 months for suite-embedded platforms like Adobe Real-Time CDP. That's not a feature difference. That's a fiscal-year difference.
The Identity Resolution Tax
Every CDP promises identity resolution. Few buyers ask what happens when it fails.
Modern Data 101's analysis calls identity resolution "the magic trick" that stitches anonymous cookies, device IDs, email addresses, and loyalty numbers into a single profile. Get it wrong, and your personalization strategy targets a fictional person.
The practical failure mode looks like this: your CDP creates duplicate profiles for the same account because the deterministic match (email) didn't fire before the probabilistic match (device fingerprint) created a separate record. Now your ABM campaign is sending conflicting messages to the same buying committee. Sales notices. They stop trusting marketing data. The CRM becomes the source of truth again, and your CDP becomes an expensive data warehouse with a nice UI.
Coworker AI's enterprise guide notes that platforms optimizing identity matching can shrink end-to-end data processing time by up to 50%. But the inverse is also true: poor identity resolution creates reconciliation work that consumes half your ops team's day.
Before you sign, ask the vendor: "Show me your merge logic when deterministic and probabilistic signals conflict. Show me the audit trail when a profile splits or merges. Show me how a marketer identifies and fixes a bad match without filing a support ticket."
The Governance Constraint Nobody Budgets For
Decentriq's comparison surfaces a question most CDP evaluations skip: "What happens when the data you need lives somewhere else?"
Retail media, regulated industries, and publisher monetization are forcing this to the surface. Your CDP unifies your first-party data beautifully. But the enrichment data, the partner audiences, the clean room collaborations? Those require governance controls your CDP may not have.
Tealium leads on compliance certifications (HIPAA BAA, server-side tracking). Adobe has added "RTCDP Collaboration" features for joint audience analysis with partners. But most mid-market CDPs assume all your data lives inside your own walls. If your growth model depends on second-party data partnerships, that assumption will break your activation timeline.
The CFO question: "What's our liability exposure if a partner's data ends up in a segment we activate to a non-compliant channel?" If your CDP can't answer that with an audit trail, you're carrying risk that doesn't show up on the implementation SOW.
Pricing Models That Hide Total Cost
G2's 2026 review lists Salesforce Data 360 at $240 for profile-based pricing, Customer.io at $100/month for 5K profiles, and "pricing available on request" for most enterprise options. That opacity is intentional.
The hidden costs live in three places:
- Overage charges on monthly tracked users (MTUs) or events. Segment's MTU-based pricing can spike unpredictably if a product launch drives traffic.
- Professional services for implementation. Adobe's 3-12 month timeline isn't free.
- Downstream tools your infrastructure CDP feeds. If you're paying for Segment plus a separate email platform plus a separate personalization engine plus a separate analytics tool, your "CDP cost" is actually four line items.
Model the total cost of ownership across a 24-month horizon. Include implementation, overages at 150% of projected volume, and every downstream tool the CDP requires. Then calculate CAC payback on the first campaign you'll activate. If payback exceeds 18 months, the business case doesn't close.
The Two-Week Pilot That Actually Proves Value
Enterprise CDP research shows 60% of businesses achieve ROI within the first year of adoption. But that stat hides the teams that never got past implementation.
Run a focused pilot before you sign an annual contract. Instrument three metrics:
- Time from data ingestion to activated segment
- Match rate on your highest-value account list
- One downstream conversion event (demo request, MQL, opportunity created)
Four to eight weeks is enough to see whether the platform's identity resolution actually works on your data, not the vendor's demo data.
If the pilot can't produce an activated segment that influences pipeline within the pilot window, the platform won't produce one at scale. Kill the evaluation and move to the next vendor.
The Decision Framework
For Salesforce-native organizations with CRM as the system of record: Salesforce Data 360 eliminates sync lag but locks you deeper into the ecosystem.
For warehouse-first teams with strong data engineering: Hightouch or Segment offer flexibility at lower implementation cost, but you're buying a data layer, not an activation layer.
For Adobe Experience Cloud shops: Adobe Real-Time CDP is the path of least resistance, but budget for a 6-month implementation and dedicated admin headcount.
For mid-market teams that need speed: Klaviyo (if you're ecommerce) or Zeotap (if you're B2B) offer faster time-to-value with less IT dependency.
The board doesn't care which CDP you picked. They care whether marketing can prove its contribution to pipeline with data the CFO trusts. Pick the architecture that gets you there fastest, run the pilot that proves it works, and kill everything else.
