What This Chart Shows
This chart shows which journeys or causal pathways actually moved the KPI, and by how much.
The left panel rolls similar journeys into a few big shapes (for example, Paid Search → Sales Pipeline → Closed Won). The right panel shows the individual journeys behind those shapes. By default, bar length is modeled dollar lift for the selected KPI.
KPI Lift Path ranks journeys by lift of the KPI whereas the Critical Pathways chart ranks the same journeys by statistical strength (how often they appear and how strongly they transfer information). A journey can be common but have low lift, or rare but high lift. This chart is built to surface the second case so you can fund, scale, or protect the paths that actually drive results.
Key Questions This Chart Helps Answer
- Which journeys created the most lift against the KPI I selected?
- Which high-level shapes (channel path, category path, or source lever) carry most of that value?
- Are my biggest-lift journeys also frequent, or are they rare but high-impact?
- How fast do the valuable journeys typically convert (average lag in days)?
Axes, Metrics, and Units
| Visual element | What it represents | Units / notes |
|---|---|---|
| Y-axis (both panels) | Journey / chain label | Categories, not time. In sales-pipeline runs, the last node may read Closed Won ($), Closed Won (speed-weighted $), or Closed Won (#). |
| X-axis (Rank by Lift) | Attributed lift for that chain | KPI unit dependent. |
| X-axis (Rank by Reliability) | How often the journey appears | Support count (frequency). Rollup uses the sum of support across underlying paths. |
| X-axis (Rank by Impact) | Average transfer entropy | Model strength, not dollars. Rollup uses a support-weighted average. |
| X-axis (Rank by Speed) | Average lag | Days. Shorter is faster. Ranking uses faster-first; the axis still shows days. |
| Bar color | Relative size within the panel | Green = top third of the panel. Yellow = middle third. Light blue = smaller bars. Color is relative to the other bars on screen, not a traffic-light score. |
| Bar label | Exact value, plus volume context | Rollup adds (n=…) = number of underlying journeys in that bucket. Single-pattern labels add (n=…) when support is greater than 1. |
| Hover | Extra context | Lift, support, average transfer entropy, average lag, unique source-lever count (rollup), and an example specific chain. Click a single-pattern bar to open that journey in the detail drawer. |
Control Options Reference
| Control | What it does | What each option means |
|---|---|---|
| Rank by | Changes what the bars measure and how they are ordered. Default is KPI Lift. | Dollar Lift ($): attributed KPI contribution. Reliability (Support): how often the journey is observed. Impact (Avg TE): causal-strength score from transfer entropy. Speed (Avg Lag): typical time from start of the path to the KPI (lower is faster). |
| Grouping | Changes how hops are labeled and how the left panel rolls journeys together. Only Lever, Series, Channel, and Event Category apply to this chart. | Lever: collapse each journey to the marketing source that started it. Series: keep the exact series path. Channel: project each hop to its channel (Paid Search, Email, Retail). Event Category: project each hop to its event category. Consecutive duplicate hops collapse (Retail → Retail → Email becomes Retail → Email). |
| KPI / drilldown | Scopes both lift dollars and which journeys qualify | Only journeys that end at the selected KPI are kept when possible. Lift is also scoped to that KPI (or the selected industries / segments). If too few journeys end at that KPI, the amber Scope widened badge appears. |
| Date range | Windows the lift dollars | Attribution is summed over the selected dates. Journey shapes still come from the discovered pattern set; the dollars on those shapes follow the date window. |
| Channel, event category, funnel, taxonomy, and search filters | Keep or drop journeys by the source lever | Include/exclude modes apply to the starting marketing lever. Covariate and external (non-marketing) sources are always removed. |
| TTE mode (when a sales-funnel KPI is selected) | Changes the outcome the dollars represent | Amount: closed-won dollars. Discounted: speed-weighted dollars. Count: conversion counts. The title shows which pass you are on. Pipeline hops display as Sales Pipeline instead of Unlabelled. |
How to Interpret the Results
- Start on the left, then confirm on the right. The rollup tells you which shape of journey is valuable. The single-pattern panel tells you whether that value is concentrated in a few specific chains or spread across many variants.
- Bar length is the story. In Dollar Lift mode, a longer bar means more modeled contribution to the selected KPI. Compare bars in the same panel, same Rank By setting, and same KPI/date scope.
- Magnitude is attributed, not “the whole KPI.” Each lever’s total lift is split across the journeys that start from it, using both causal strength and how often the journey is seen. So a bar is that journey’s share of its source lever, not an independent experiment result. The left panel sums those shares, so rollup totals do not double-count the same lever across similar paths.
- Average lag (days) tells you how long that journey typically takes. Speed ranking reorders the same journeys so the fastest rise to the top; it does not change the underlying lift math.
- Read n= as concentration, not spend. On the rollup, n= is how many distinct discovered paths sit inside that macro shape. A large dollar bar with small n= is a concentrated bet. A large dollar bar with large n= is a broad shape.
- Color is relative, not a grade. The green bar is simply one of the largest in this panel. It is not “good” vs “bad” on its own.
Practical Applications for Marketers
| Decision | How to use this chart |
|---|---|
| Budget allocation and reallocation | Fund the macro shapes with the largest Dollar Lift, then check the right panel so you are not scaling one noisy variant. Compare with Critical Pathways: high lift + high reliability is the strongest scale case. High lift + low support is a high-upside path that may need more volume or a test before a large shift. |
| Channel or tactic optimization | Use Channel or Event Category grouping to see which hop sequences carry value. If Paid Search → Sales Pipeline dominates, protect that handoff. If a channel is frequent on Critical Pathways but small here, it may be generating activity without KPI lift. |
| Campaign evaluation | Filter to the campaign’s channel or search terms, keep Rank by Dollar Lift, and read whether its journeys appear in the top rollup. Use lag in hover to judge whether the campaign should be judged on a short window or a longer conversion cycle. |
| Experiment design or validation | Use this chart to pick which path to test, not as the test result. A high-lift, low-support chain is a good candidate for a holdout or geo test. After the test, compare the experimental lift to this modeled ranking; agreement builds confidence, disagreement means investigate scope, lag, or a confounding hop. |
| Strategic planning and forecasting | Use Amount vs Count TTE titles to separate dollar quality from conversion volume. Discounted (speed-weighted) dollars highlight journeys that win revenue sooner. Do not treat the visible top 12 / top 15 as a complete forecast of total KPI change—unshown tails, levers with no discovered path, and non-marketing lanes are outside these bars. |
Common Mistakes & Misinterpretations
| Mistake | Why it is a problem | How to avoid it |
|---|---|---|
| Treating this as the same ranking as Critical Pathways | That chart ranks statistical reliability / transfer entropy. This one ranks attributed dollars. The leaders often differ. | Use both: reliability for “is this real and repeatable,” lift path for “is this worth the money.” |
| Adding the visible bars and calling that “total KPI lift” | Only the top 15 singles and top 12 rollups are drawn. Other levers, journeys that never made it into the pattern set, seasonality, and external factors are not in these bars. | Use this chart to rank paths. Use the Attribution Waterfall or Quarter Attribution for a full add-to-total decomposition. |
| Comparing a rollup bar to the sum of the visible single-pattern bars | The panels are different grains. The right panel is a top-15 slice. The left panel can include many paths that are not shown on the right. | Read left for the shape, right for examples. Do not expect the two columns to add to each other. |
| Using Speed or Support mode to make budget decisions | Those modes reorder the same journeys by frequency or lag. The bar is no longer dollars. | Switch back to Dollar Lift before allocating budget. Use Speed/Support as diagnosis, not as the money ranking. |
Caveats & Considerations
- Only marketing-rooted journeys are shown. Chains that start from covariates (for example weather) or external shocks (for example competitor or rate series) are removed so the leaderboard stays actionable for media and CRM. In sales-funnel (TTE) runs, purely internal pipeline chains (pipeline age buckets → Closed Won with no marketing start) are also removed. Intermediate pipeline hops on marketing-rooted chains are labeled Sales Pipeline.
- The selected KPI and date window matter. If no journeys terminate at your KPI, the chart does not silently invent them. It widens scope and tells you. Lift dollars follow the same KPI blocks as the waterfall, including multi-segment drilldowns.
- No confidence bands are drawn on these bars. Uncertainty is not a shaded interval here. Do not over-read small gaps between similarly sized bars.