What This Chart Shows
This chart is leaderboard of causal pathways (recurring cause→effect chains in your model). It shows you which pathways show up most often, which have the strongest causal impact, and which are fastest.
Each bar is one pathway. You choose how to rank them: by Impact (average strength of the causal signal, via Transfer Entropy), by Reliability (how often the pattern appears, support), or by Speed (how quickly the path plays out, using average lag). So you see which pathways drive the most impact, are most frequent, or are fastest, and you can click a bar to drill into that pattern.
Key Questions This Chart Helps Answer
- Which causal pathways (chains of touchpoints) drive the most impact?
- Which pathways are most reliable?
- Which pathways are fastest?
- What is the sequence of touchpoints in each top pathway?
- Where should we focus for conversion and journey optimization?
Axes, Metrics, and Units
Element | Description |
|---|---|
| X-axis | Pattern label (pattern ID and/or formatted name, e.g. with support). Truncated for display; full path in hover/side drawer. |
| Y-axis | Depends on Rank by selection. Impact: Average Transfer Entropy (TE). Support: Support (frequency). Speed: Average Lag (lower = faster) All values are modeled. |
| Bar colors | One bar per pattern; height = the Y-axis metric for that pattern. Color by relative intensity (e.g. green = high, yellow = medium, light blue = low). |
| Bar text | Value on each bar (e.g. 3 significant figures). |
| Hover | Pattern #, Support, Avg TE, Avg Lag, and the path as "Series A -> Series B -> ..." (when edges are available). |
| Title | "Critical Causal Pathways (ranked by impact)" or "ranked by support" or "ranked by speed" to match the control. |
These are modeled pattern-level metrics, not raw event counts.
Control Options Reference
Control | Meaning |
|---|---|
| Rank by | Reliability (Support): Sort by support (frequency) Impact (Avg TE): Sort by average Transfer Entropy Speed (Avg Lag): Sort by speed (lower lag = faster) Default is Impact. |
There are no event/funnel filters or grouping mode on this chart.
How to Interpret the Results
- Impact mode: Taller bar = stronger average causal signal (TE) along that pathway. Use to find pathways that best predict or explain the outcome.
- Support mode: Taller bar = pattern appears more often. Use to find common journeys; high support can mean more robust inference.
- Speed mode: Shorter bar = faster pathway (lower average lag). Use to find quick paths to conversion; sort order is “fastest first.”
- Path in hover: "Series A -> Series B -> ..." is the ordered sequence of touchpoints (series) in the pattern. "Pattern path unavailable" means edge data for that pattern was missing.
- Bar order: Top 15 patterns by the chosen metric. Green bars are relatively strong/fast/frequent; gray are lower on the scale.
Practical Applications for Marketers
Application | How to use this chart |
|---|---|
| Causal storytelling | Use top pathways by impact or support to explain "these chains drive results" in exec or stakeholder reviews. |
| Prioritization | Focus analysis and optimization on high-impact or high-support pathways; use speed to separate fast vs slow chains. |
| Drill-down | Use as the entry point for macro-to-micro: click a pathway to see details, edges, or timelines. |
| Strategy | Combine impact and speed to find "strong and fast" pathways to invest in or replicate. |
| Validation | Check that top pathways match domain intuition (e.g. known funnel steps or channel sequences). |
Common Mistakes & Misinterpretations
Mistake | Why it is a problem | How to avoid |
|---|---|---|
| Treating pathways as guaranteed causation | Pathways are model-discovered patterns; they indicate structure in the graph, not proven cause-effect in the real world. | Use for hypothesis and prioritization; validate with tests or domain knowledge where it matters. |
| Mixing up the three modes | Impact, support, and speed are different metrics; the same pathway can rank differently in each. | Check the title and Y-axis: "ranked by impact" vs "support" vs "speed" and interpret bars accordingly. |
| Ignoring the path order | "A → B → C" is directed; order matters for cause and effect. | Read the path in hover and use it to interpret direction (e.g. channel A → conversion B). |
| Assuming all pathways are shown | Only the top 15 by the selected metric are shown. | Use for "top" pathways; other views or exports for the full set. |
| Comparing impact and support directly | Impact (TE) and support (count) are on different scales and meanings. | Compare within one mode at a time; don't mix "tall in impact" with "tall in support" without switching the view. |
Caveats & Considerations
- Edge Bridge / TCEG: Pathways come from the temporal causal edge graph (TCEG) and pattern discovery; they depend on pipeline configuration and data quality. No TCEG/pattern run → no chart data.
- Top 15: Only the top 15 patterns by the chosen ranking metric are shown; the rest are not in the view.
- Path label: The chain "A → B → C" is built from series hashes in the edge family; if labels are missing or hashes don't resolve, you may see "Pattern path unavailable" or truncated IDs.
- No filters: The chart does not accept event category, funnel stage, or outcome filters in the code; it shows all patterns produced by the Edge Bridge. Filtering would require product or pipeline changes.
- Transfer entropy: "Impact" is average transfer entropy along the pathway—a measure of information flow/causal strength in the model, not revenue or conversions unless the pipeline maps it that way.