> ## Documentation Index
> Fetch the complete documentation index at: https://docs.arcdemo.io/llms.txt
> Use this file to discover all available pages before exploring further.

# Learn from engagement

> Read the metrics and step funnel, then decide what to improve.

Open the demo in the editor and select **Analyze** to compare engagement with the outcome you intended. A sales leave-behind and an onboarding guide may have different definitions of success.

## Read the overview

| Metric | What it tells you |
| - | - |
| Views | Sessions in which the demo was opened. One person can open the demo more than once. |
| Unique viewers | Distinct viewers in the selected period. |
| Engagement | The share of viewers who clicked a hotspot. |
| Completion rate | Completed views divided by views. |
| Avg session | The average duration of completed views; it is not the average for everyone who opened the demo. |

Choose a useful date range before comparing results. Internal and QA traffic is excluded from analytics. Some breakdowns require a higher plan; the interface identifies features that need an upgrade.

## Find the step to improve

The **Step funnel** shows views for each step and, when available, hotspot clicks, CTA clicks, and average time. Look for a meaningful change between steps, then view the demo yourself from the preceding step. Check whether the next action is visible, the instruction matches the screen, and the content answers the question that brought the viewer there.

A drop is a clue, not a diagnosis: someone may leave because the next action is unclear, or because they already found what they needed. Branching paths and different entry points can also affect the shape of a funnel.

Change one thing, publish it, and compare a later period with the earlier one. Treat small samples carefully and combine analytics with feedback from people who use or share the demo.

## Example: investigate a drop

Suppose a fictional demo has 100 views and 40 completed views in a period. Its completion rate is 40%. The funnel shows fewer viewers reaching the step after a filter selection. That tells you where to investigate; it does not tell you why people left.

Open the published demo and try the preceding step on the devices your audience uses. Is the filter visible? Does the guide use the same label as the screen? Does the next step show the promised result? Ask a colleague to try it without coaching. If the instruction is unclear, revise that instruction and publish the change.

Compare a later period using the same metric and a comparable audience. If traffic came from a different campaign, or only a few people viewed the demo, a change in completion may have other causes. Completion is also not the only outcome: an onboarding guide can be useful even when someone leaves to perform the task in their own account.

## Choose the next experiment

Write down the question, what you observed, and one change to try. For example:

| Question | Observation | Change to test |
| - | - | - |
| Can viewers find the filter? | Fewer views reach the result after the filter step. | Make the guide identify the filter by its visible label and move it away from the control. |
| Does the ending support a follow-up? | Viewers reach the ending, but few select its CTA. | Check that the CTA describes the destination and that the destination works. |
| Can a new user repeat the task? | Reviewers finish the demo but ask where to start in their own account. | Add the prerequisite and starting location to the surrounding article. |

These observations suggest what to investigate; they do not prove that the proposed change will work. Record the published change date, compare an appropriate later period, and include feedback from actual reviewers. Keep the audience and distribution channel in mind: a website visitor and a buyer receiving a follow-up may behave differently.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.