Capture every event — and actually keep it.
Most CDPs will collect your data. Far fewer let you hold it — raw, complete, fast, and free of vendor tolls. The gap only shows up later: when a model needs the request headers that got stripped, when an attribution audit hits an hourly batch sync, or when a renewal quote reveals the raw export was a paid add-on all along.
The Data.C4 Toolkit gives you a structured, evidence-based framework to evaluate how well a packaged or composable CDP delivers first-party data collection — across custody, payload fidelity, warehouse latency, and cost of access. Use it to decide which architecture fits your requirements before you sign.
Who Is This For
- Founders and CTOs at high-growth D2C, subscription, or commerce businesses choosing their first CDP
- Marketing Ops and Data leaders who need raw event data for ML, attribution, and bespoke segmentation
- Teams moving to a warehouse-centred stack and weighing packaged versus composable collection
- Consultants and agencies advising mid-market clients on customer data ownership and lock-in
What Is Inside
Data.C4 Workbook (PDF) — a structured framework that walks you through:
- The four dimensions that decide whether you own your collection layer: custody (leverage on exit), payload fidelity, latency to availability, and cost to access
- How packaged CDPs (Segment, mParticle, Tealium) and composable stacks (Snowplow, Snowflake, dbt, Hightouch) handle each, with concrete vendor behaviours
- The seven knowledge areas vendors won’t volunteer — schema standardisation, dropped headers, batch-vs-streaming reality, retention caps, and the cost model for raw access
- Six configurable parameters that turn the generic capability into your requirements
- An importance-weighting step so the score reflects your priorities, not a generic average
- Six due-diligence questions with side-by-side packaged vs composable investigation paths
- Binary, weighted scoring that produces a clear, defensible verdict — worked through a full sample scenario
How the Toolkit Works
The toolkit follows Datawhistl’s four-step framework, applied to first-party data collection.
Step 1 — Understand the capability. Sections 1–3 define what raw, complete, low-latency collection actually means, set out the four assessment dimensions, and arm you with the knowledge areas vendors rely on you not knowing. This removes the ambiguity sales conversations depend on.
Step 2 — Configure your requirements. Section 4 turns the capability into six parameters. You select a value for each — that’s what gets tested against the architecture.
Step 3 — Weight your requirements. Section 5 has you assign an importance percentage to each requirement, totalling 100%. This makes the scorecard reflect your business priorities, independent of how strict your selections are.
Step 4 — Run due diligence and score. Section 6 gives you six questions with documented investigation paths for each architecture. Section 7 applies binary, weighted scoring to produce a verdict you can take into a board meeting or a vendor negotiation.
Benefits and Outcomes
- Prove who owns the data — confirm in writing whether raw history lives in your storage or the vendor’s, before the renewal leverage matters
- Protect the signal your models need — surface dropped headers, truncated payloads, and property caps before they break attribution and ML
- De-risk vendor selection — get written commitments on export rights, warehouse latency, and access fees instead of demo theatre
- Avoid hidden tolls — expose raw-export add-ons, faster-sync surcharges, and retention caps early
- Build stakeholder alignment — present a scored, requirement-backed recommendation, not a vendor pitch