Set Up Funnel Analysis
Using Event Segments
on Lumenview

What is Funnel Analysis?

Funnel analysis is a method of measuring how visitors move through a sequence of steps toward a goal. Each step — landing on a page, starting a signup, reaching checkout — is a stage, and the share of visitors who make it from one stage to the next is the conversion rate for that step. Instead of a single number for the whole site, a funnel shows you exactly where people continue and where they drop off, so you can focus your effort on the step that actually loses the most visitors.

When Should You Track Funnels?

Track a funnel whenever a goal takes more than one action to complete. Typical examples are onboarding flows, checkout processes, newsletter signups behind a confirmation step, or a docs page that should lead into a trial. Funnels are most useful when you suspect friction somewhere in the middle of a flow but can't tell from pageviews alone which step is responsible. Setting one up on Lumenview takes a few minutes and requires no changes to your backend.

Defining an Experiment Step by Step

Before wiring anything up, define a hypothesis. It is usually tied to a goal, such as increasing completed signups. For example, you might hypothesize that the plan selection screen is where most visitors stall, because it presents too many choices before they have seen the product.

To test this, you decide which events mark each stage of the flow, attach a segment to the visitors who fire them, and then compare the stages in Lumenview. Meaningful results need some traffic and a little patience — with only a handful of sessions the percentages will swing wildly.

Event Segments

Using event segments, building a funnel on Lumenview takes just minutes. Segments can be used for all sorts of things, including funnels, but you can also use them to group visitors or attach metadata to page views. They are always tied to the page view itself, like the path or the page title. Here is an example of how to define segments using the JavaScript snippet integration. The API also supports appending segments server-side.

<script defer src="https://cdn.lumenview.io/lv-events.js"
    id="lumenviewjs"
    data-site="your_site_code"
    data-segment-flow="signup"
    data-segment-cohort="spring-beta"></script>

In the example above, data-segment-flow and data-segment-cohort are appended as segment keys to the page view with the values set to signup and spring-beta. Segments are always prefixed with data-segment-. You can also set segments without specific values by adding the data-segment attribute with a comma-separated list of keys (e.g. data-segment="flowA,flowB").

Once attached, segments are listed in the dashboard and can be used to filter results.

For the steps in the middle of the flow, fire a custom event when the visitor reaches each stage. The call is one line and accepts optional metadata that ends up in the same segment view.

lumenview.event("checkout_viewed", {
    segment: "spring-beta",
    step: 3
});

Each stage of your funnel is now either a page view carrying a segment or a named event. That is all the instrumentation the funnel needs — everything else happens in the dashboard.

Reading the Results

Visiting the dashboard now shows the funnel taking shape. Filter by the flow segment and the stat row updates to the visitors inside your experiment, while the chart plots how many of them reached each stage over the last two weeks.

You can see that visitors carrying the spring-beta cohort complete the funnel noticeably more often than the dotted baseline. Clicking any segment key and value in the details dialog filters the whole dashboard down to that slice, so comparing cohorts is a single click rather than a query.

With more data, you can begin to compare stages side by side and see which step recovers the most visitors. The comparison mode (since Lumenview version 3.1) is very useful in this case.

Conclusion

Funnel analysis is a powerful way to find out where a flow leaks visitors and to verify that your fixes actually work. I hope you found this article useful. If you have any questions, don't hesitate to contact us via email, Twitter, Mastodon, or on the public support forum.