Turn messy customer journeys into a single, readable flow

Cross-channel attribution has a visibility problem: your performance data is usually spread across platforms (CTV, display, streaming audio, social, retargeting, search) and stitched together in spreadsheets that hide the real story. A Sankey diagram solves that by showing your customer journey as “flows” between touchpoints, with thicker lines representing more volume (users, conversions, revenue, or weighted credit).

For marketing managers and agency teams, Sankey diagrams are a practical way to explain how channels work together—especially now that many advertisers rely on data-driven attribution (DDA) in GA4 and other environments, and several legacy rule-based models have been retired from Google’s UI. (searchengineland.com)

Why attribution gets confusing (and why Sankey helps)

Most cross-channel reporting breaks down at the moment you ask a simple question: “What sequence of touchpoints is driving conversions?” Standard dashboards often answer with isolated channel metrics (impressions, clicks, CPA) but not the path.

Sankey diagrams work because they are built for path analysis:

They show sequences (CTV → site visit → retargeting → conversion) instead of isolated channels.
They make “assist behavior” visible, so awareness channels aren’t dismissed just because they rarely get last-click credit.
They reduce reporting friction for stakeholders: one visual can replace multiple slides of tables.

Step-by-step: building an attribution Sankey that’s actually useful

1) Choose the “unit of flow” (be explicit)

Pick one primary measure per diagram:
  • Users (best for awareness and reach questions)
  • Conversions (best for performance and funnel questions)
  • Revenue (best for eCommerce)
  • Attributed credit (best when you’re comparing models or weighting touchpoints)
2) Normalize your channel taxonomy

Sankeys get noisy when you have inconsistent naming (e.g., “CTV”, “OTT”, “Connected TV”, “Streaming TV” as separate nodes). Decide on a single taxonomy:
Example nodes: CTV/OTT, Online Video, Display, Streaming Audio, Paid Social, Search, Site Retargeting, Email, Organic, Direct
Tip: Keep it to 8–12 nodes for executive-facing visuals. Create a second “ops Sankey” if you need vendor-level detail.
3) Decide what counts as a “step” in the journey

Common step definitions:
  • Touchpoints (impressions/clicks across channels)
  • Sessions (site visits grouped by source/medium)
  • Events (e.g., video view → landing page view → form start → form submit)
If you’re using GA4, remember that attribution interpretation depends on the model you select (DDA vs last-click variants), and Google has retired several legacy models in both Ads and Analytics—so your “step logic” should match what your stakeholders see in reports. (searchengineland.com)
4) Limit path length and group the “long tail”

Without guardrails, Sankey diagrams become spaghetti. Two practical constraints:
  • Path depth: show 2–4 steps (e.g., First touch → Assist → Last touch → Conversion).
  • Thresholding: group any path under ~1–3% volume into “Other”.

A practical comparison table: Sankey vs common attribution views

View Best for What it misses Where Sankey helps
Last-click report Short cycles, conversion capture Assist channels look “unprofitable” Shows the common assists that lead to last-click wins
DDA / model comparison Complex journeys, fractional credit Hard to explain to non-analysts Turns weighting into an intuitive flow of influence
Channel CPA/ROAS table Budget guardrails No sequencing or synergy Adds “why” behind efficiency shifts across channels
MMM (marketing mix modeling) Strategic budget allocation over time Less granular user-path detail Sankey complements MMM by visualizing journey-level paths
Note: MMM and MTA answer different questions and can produce different results; many organizations use both for planning and performance measurement. (en.wikipedia.org)

Common mistakes to avoid (so your Sankey doesn’t mislead)

  • Mixing measurement types in one diagram. If one link is “users” and another is “attributed conversions,” the visual becomes persuasive but wrong.
  • Ignoring attribution model context. GA4’s DDA is the default in many setups and uses a proprietary approach; last-click variants (e.g., “Google Paid Channels last click”) can tell a different story. Align the Sankey with the model you’re using for decision-making. (insightland.org)
  • Letting “Direct” dominate because of tracking gaps. A sudden spike in Direct often signals tagging issues, cross-domain problems, or missing click IDs—fix instrumentation first, then visualize paths.
  • Over-granular nodes. “Facebook Prospecting / 18–34 / Video View 25%” is great for ops, terrible for an exec Sankey. Build two layers: executive (simple) and diagnostic (detailed).

Local angle: cross-channel attribution in the United States (privacy, platforms, and fragmentation)

In the U.S., most teams are dealing with a fragmented measurement reality: multiple ad platforms, multiple devices, and evolving privacy constraints. That makes stakeholder communication just as important as the math.

A clean Sankey diagram becomes a shared language for teams that don’t live in analytics tools—especially when you’re coordinating campaigns across CTV/OTT, streaming audio, display, and retargeting, then reconciling it with site analytics and lead outcomes.

For agencies and multi-location brands, Sankeys are also a practical way to show regional journey differences (e.g., metro vs suburban paths) without exposing sensitive platform-level tactics.

Pro tip for U.S. reporting decks: Include a one-line “definition strip” under your Sankey (e.g., “Flow = conversions, Model = DDA, Lookback = 30 days, Paths grouped under 2% = Other”). This reduces back-and-forth and prevents misinterpretation.

How ConsulTV supports attribution clarity across channels

When your media mix includes multiple channels, the biggest operational bottleneck is often not buying—it’s unified visibility. ConsulTV’s full-stack programmatic approach is built to coordinate targeting and optimization across digital channels while keeping reporting clear for internal teams and agency clients.

If you’re building Sankey-style journey visuals for client-facing reporting, ConsulTV also supports Sales Aides & Agency Partner Solutions for white-label enablement and scalable workflows.

Want cleaner attribution visuals across CTV, display, audio, and retargeting?

Share your current channel mix and reporting goals—ConsulTV can help you structure cross-channel measurement so it’s easier to act on (and easier to explain).
Talk to ConsulTV

Response-oriented, not pushy—bring your questions.

FAQ: Sankey diagrams & cross-channel attribution

Are Sankey diagrams an attribution model?
No. Sankey diagrams are a visualization method. They can display last-click paths, data-driven attribution credit flows, or even event sequences—depending on what you feed into them.
What’s the best attribution model to pair with a Sankey?
Use the model your business actually uses for decisions. Many teams pair Sankeys with DDA for realism and with last-click variants for operational clarity (especially for lead-gen). Google has retired several legacy models, so model availability may be limited depending on your setup. (searchengineland.com)
How do I keep a Sankey diagram from becoming unreadable?
Limit nodes (8–12), cap path depth (2–4 steps), and group low-volume paths into “Other.” If you need more detail, create a second version for ad ops that expands vendors, creatives, or audience segments.
Can a Sankey show cross-device journeys (CTV to mobile to desktop)?
It can, as long as your underlying identity and measurement setup supports cross-device stitching. If cross-device visibility is partial, label the Sankey clearly (e.g., “observed paths”) and avoid over-claiming precision.
How is MMM different from multi-touch attribution (MTA), and where does Sankey fit?
MMM is typically used for high-level budget planning and long-term impact; MTA is more granular and journey-oriented. Sankey diagrams are most naturally aligned with MTA-style thinking, but they can complement MMM by explaining the “journey narrative” behind channel interactions. (en.wikipedia.org)

Glossary (quick definitions)

Sankey diagram: A flow diagram where the width of each link represents volume (users, conversions, revenue, or weighted attribution credit).
Cross-channel attribution: Measurement that assigns credit for conversions across multiple channels (e.g., CTV, audio, display, social, search, retargeting).
Data-driven attribution (DDA): An attribution approach (often ML-based) that assigns fractional credit based on observed contribution across converting and non-converting paths; commonly used as the default in GA4. (insightland.org)
Lookback window: The time period used to consider prior touchpoints eligible for attribution credit (e.g., 7, 30, or 90 days).
MTA (multi-touch attribution): Journey-level attribution that distributes credit across touchpoints rather than assigning all credit to one interaction.
MMM (marketing mix modeling): A statistical approach typically used for strategic budget allocation that estimates channel impact in aggregate over time, often complementing MTA. (en.wikipedia.org)