Goals and results

A goal is the action you want to see increase: a click on “Sign up”, a visit to the thank-you page, a purchase. An experiment measures, for each goal it tracks, how many visitors of each variation reach it. In Highwind, you create a click goal simply by clicking the element in the preview.

The site’s goals

The Kameleoon panel’s Goals tab lists the site’s goals, newest first (archived goals don’t show up). Each row shows the goal’s name and its type: Click, Page, Custom, Scroll, Page views, Time spent or Retention. An inactive goal has a greyed-out icon.

The panel creates click and page visited goals. Other types are created in Kameleoon; they then show up in the list and can be linked to your experiments.

Create a click goal from the preview

1. In the Goals tab, click New goal. The Click on an element type is selected.
2. Click Pick in the preview. The button shows “Click an element in the preview…”.
3. In the preview, click the element to track: a button, a link, a pricing card…
4. The form fills in:
   • CSS selector: the selector that finds the element, editable;
   • On page (optional): the path of the page you clicked on (empty for the home page);
   • Goal name: suggested from the element’s text, for example “Click “Sign up””.
5. Check the selector, adjust the name, then click Create goal.

The message “Goal “Click “Sign up”” created in Kameleoon.” confirms it, and the goal shows up in the list.

With On page, the click only counts on pages whose address contains that path (for example /pricing). Leave the field empty to count the click anywhere on the site.

A stable selector

The selector must find the element on the published site, not just in the preview. Prefer an id or a data-… attribute over a chain of tags or generated classes. The easiest way: ask the agent “Add data-goal="signup" on the sign-up button”, publish, then use the selector [data-goal="signup"].

Create a page visited goal

To count visitors who reach a page (thank-you, order confirmation…):

1. Click New goal, then choose Page visited.
2. Page path: for example /thank-you. The goal is reached when the visited address contains this path.
3. Give a Goal name, then click Create goal.

Note

For a conversion that’s neither a click nor a page (a sign-up confirmed by the server, a successful payment…), ask the agent to track it in code: it calls Kameleoon at the right moment with your goal’s id. See The agent and Kameleoon.

Link goals to an experiment

In an experiment’s detail, the Tracked goals section lists the goals measured by the test.

• Add: choose a site goal in the Add a goal… list, then click Add. If the site has no goals yet, the panel invites you to create one in the Goals tab.
• Main goal: it carries the main badge. To pick another one, hover it and click the star (Make it the main goal). The first goal you add automatically becomes the main goal.
• Remove: hover the goal and click the cross (Remove from the experiment). If you remove the main goal, the first remaining goal takes its place.

Tip

The main goal is the one that decides the test: choose the action that really matters to you (sign-up, purchase), and keep the others as secondary indicators.

Read the results

1. Open the experiment’s detail.
2. In the Results section, click Compute results. The button shows “Computing…”.
3. Kameleoon computes the results on demand, which can take a few seconds. If the computation isn’t done, the panel shows “Results are still being computed: try again in a moment.”: click again a little later.

Results are shown per goal, in a table with one row per variation:

Column	What it shows
Conversion	The variation’s conversion rate for this goal.
Improvement	The difference from the original, green when positive, red when negative.
Reliability	The statistical confidence that this difference isn’t due to chance.

The Original is the reference: its Improvement and Reliability columns stay empty.

The winning variation

A variation gets the Winner badge (and its row turns green) when, for a goal:

• its reliability is at least 95%;
• and its improvement is positive.

Don’t conclude too fast

A high reliability after only a few visits can still change. Let the test run long enough (at least one full weekly cycle, for example) before deciding, and rely on the main goal first.

No data yet

As long as the experiment hasn’t received visits, the panel shows “No data yet: results show up once the experiment gets visits.”. Check that the experiment is Running, that the tag is published and that the tested page’s address is really your site’s.

Tip

You can also ask the agent for an analysis: “Analyze the results of experiment 123 and tell me if a variation wins.” It reads the same numbers and can then build the winning variation into your app.

See also

Experiments and variationsCreate the experiment and code its variations.

Select an elementThe preview's Select mode, used for click goals.

The agent and KameleoonTrack goals in code and analyze results.

The Kameleoon panelTabs, refreshing and read-only mode.