Skip to content
Robin Schreiner

UX Management · Product Leadership

← All notes

Think End to End · 3 min read

One journey, one success metric, several levers.

A clear outcome gives a journey direction. The measures around it show what we can change and what we must not make worse.

This is part two of Think End to End. Start with the first article if you want the whole thread.

Ask five people whether an application journey works and you may get five answers. One looks at completion. Another looks at time on task. Someone else points to support tickets. All of them may be right about a part of the experience. We still need to agree on what success means for the person using it.

For this journey, I would start with: What share of people who begin an application submit a complete one within seven days? It is a possible primary metric, not a universal template. We would need to define what “begin” and “complete” mean, and separate people who learn they are ineligible from people who wanted to apply but could not finish.

The metric gives the team a direction. It does not explain how to get there.

Find the levers along the journey

Working backwards, we could look at whether people understand the eligibility rules, know which documents they need, get through the form, recover from an error, and understand the confirmation. Each step offers a lever we may be able to change.

For example, clearer document guidance might increase the share of people who reach the upload step prepared. Better error messages might reduce repeated failed submissions. Neither is the goal by itself. We care because it may help more people complete the application successfully.

This is the useful part of the North Star framework: a meaningful outcome with a small set of actionable inputs. A product may have one overall North Star; a team can also choose a primary outcome for the journey it owns. I would not call every journey metric the product’s North Star. The measures should connect, not compete for the same name.

Keep a few guardrails

If completion rises because we stop checking applications properly, we have not improved the service. If support contacts fall because people cannot find support, that is no victory either.

So I would watch a small number of guardrails alongside the primary metric: applications requiring rework, avoidable support contacts, and whether different groups can finish. The exact set depends on the service. The point is to notice when a seemingly good number hides a worse experience.

Segments matter here. An overall completion rate can improve while first-time applicants, people using assistive technology, or people on mobile are struggling more. A good average is not permission to stop looking.

A metric is a question we keep asking

Before changing anything, I would record the current rate and check whether the tracking is trustworthy. Are events fired at the right point? Can we distinguish a saved draft from a submitted application? Does the definition survive a change to the form? If the answer is no, fix the measurement first.

Then we can make a claim we are willing to test: “If we explain document requirements before the form starts, more first-time applicants will submit a complete application within seven days, without increasing rework.”

After release, we check the outcome and the guardrails. A rise in one input metric does not prove that the whole journey improved. And if the outcome does not move, we have learned something about our assumption.

One metric gives us focus. The levers show where to act. Research helps us decide which lever is worth touching in the first place.

Previous: Stop collecting UX artefacts. Start connecting the evidence. Next: When does a signal become an insight?.

Keep reading

When does a signal become an insight?