Key takeaways

In this module, you learned:

• Dimension breakdowns let you progressively narrow your analysis. Adding a breakdown takes a top-level dimension row and expands it with another dimension underneath — for example, breaking "call category" down by "call reason," and then breaking a specific reason down by "city." Each level adds specificity, helping you move from identifying a broad pattern all the way to pinpointing where and why something is happening.
• Adding a filter as a row item in a freeform table is different from adding it to the filter drop zone. Dropping a filter into the drop zone at the top applies it globally to the entire panel. Adding a filter as a dimension row instead places it as a line item in the table, enabling side-by-side comparison of metrics across multiple segments — like web, mobile app, and in-store — all within a single table without building separate panels.
• Flow visualizations can run forward or backward depending on where you place the dimension. Placing a dimension in the "starts with" box builds the journey forward — showing what happened after that touchpoint. Placing it in the "ends with" box builds the journey backward — showing what path led up to that point. This makes flow useful for both "what do people do next?" and "how did people get here?" questions.
• Fallout visualizations can be segmented by channel to reveal where each channel loses people. By adding channel filters (web, mobile, in-store) as the first step of a fallout funnel before the key conversion steps, you create a segmented view of the same funnel for each channel. This lets you see whether drop-off is concentrated in one channel or distributed across all of them — and is more actionable than a single combined funnel.
• A panel-level filter is required when you need to connect behaviors across channels to the same person. If you want to analyze people who purchased on the web and then returned via the mobile app, metric-level filters alone will show all web orders and all mobile returns separately — including people who only did one of those things. The panel-level filter restricts the entire panel to only those people who performed both behaviors, making the cross-channel connection explicit.