Goal types
Pageview, click, custom and revenue goals, and the settings that decide how they are counted.
When you create a goal, you first choose its type. The type decides how ABlyft recognizes the event. Three more settings decide how the events are counted and evaluated: the counting method, the winning direction and, for revenue goals, the value filter.
Choose the type
| Type | Use it when | How it is triggered |
|---|---|---|
| Pageview | Visiting a page is the success, for example a thank-you page. | Automatically, when the linked page becomes active. |
| Click | Clicking an element is the success, for example an "Add to cart" button. | Automatically, when a visitor clicks an element that matches your CSS selector. |
| Custom | Something else is the success, for example a form submit or a video played. | With a line of JavaScript on your website or in your tag manager. |
| Revenue | The success has a monetary value, for example an order. | With a line of JavaScript that passes the amount. |
Pageview goals
A pageview goal counts a conversion when its Page becomes active for a visitor. Select the page in the Page field; the page defines which URLs count (see Pages). No code is needed beyond the installed snippet.
Click goals
A click goal counts a conversion when a visitor clicks an element that matches your CSS selector, on a page where the goal's Page is active.
| Field | Description |
|---|---|
| Page | The page on which clicks are tracked. Required. |
| CSS selector | Selector of the element, for example .my-button. Several selectors can be separated by commas. Required. |
How pageview goals and click goals behave in detail, and what to watch for with links to other pages, is described in Click & pageview goals.
Custom goals
A custom goal is triggered by your own code. In the Definition section you set:
| Field | Description |
|---|---|
| API Name | The name you use in your code. Letters, numbers, dashes and underscores. Unique in the project. |
| Measurement type | Conversion: calculate based on conversion rate. Value: calculate based on a numeric value (metrical). |
The API Code box ("To track this event, add the following code to your page or tag manager.") shows the code for your goal. Click it to copy. How to send the event is described in Custom goals.
Revenue goals
A revenue goal measures money. Selecting Revenue sets the measurement type to value automatically.
| Field | Description |
|---|---|
| API Name | The name you use in your code. |
| Currency sign | The sign shown next to revenue values in the results. Default: €. Required. |
| Value filter type | None, Absolute or Relative. See Value filters. |
| Value filter value | Threshold of the filter. Required unless the filter type is None. |
The API Code box generates the call with eventType: 'revenue' and an eventValue line. How to send the revenue is
described in Revenue tracking.
Shopify shops can send purchases to ABlyft by other means than this code, see Shopify tracking.
Counting method
The counting method decides what happens if the same visitor triggers a goal several times.
| Option | Meaning | Typical use |
|---|---|---|
| One | One event per participant is counted (unique). | "Did the visitor convert at all?" Newsletter sign-up, purchase as yes/no. |
| Every | Every event per participant is counted (total). | "How often?" Page views, add-to-cart clicks, orders per visitor. |
With One, ABlyft sends the goal only once per participant and variation, even if your code calls it repeatedly. For revenue goals, Every is usually what you want for revenue per visitor, while One only counts the first order. With One, the number of conversions can never exceed the number of participants, so the conversion rate stays between 0 and 100%. With Every, conversions can exceed participants and the conversion rate can be above 100%.
With One, ABlyft remembers in the visitor's browser which variations already received the goal. If a visitor clears their browser storage, the goal can be counted again.
Winning direction
| Option | Meaning |
|---|---|
| Increase | More is better. A variation wins if the metric goes up. |
| Decrease | Less is better. A variation wins if the metric goes down, for example for a bounce or error event. |
The direction decides whether a significant change is shown as a winner or a loser in the results.
Value filters
Revenue goals can limit the influence of extreme values, such as one very large order, on the result. Values above the limit are not removed; they are capped at the limit.
| Filter | Limit | Example |
|---|---|---|
| None | No limit. | Every value is used as it is. |
| Absolute | The number in Value filter value is the maximum value. | Value 500: an order of 1,200 counts as 500. |
| Relative | The limit is the average value plus the given percentage (average + average / 100 × percentage), calculated separately for each variation. | Average order 100, value 50: the limit is 150. |
Use Absolute if you know a clear maximum that makes sense for your business. Use Relative if your data varies and you want the limit to adapt; it does not compare against a fixed or global maximum.
When to use a value filter
- A few transactions have a disproportionate influence on the result.
- Revenue values differ strongly between visitors.
- You want to evaluate typical visitor behavior instead of edge cases.
What the filter changes
The filter affects the calculation and display of the results, for example the average revenue per visitor, the total revenue and the uplift. It does not delete or overwrite any data, and it does not change tracking. You can change or remove the filter at any time, the results are recalculated and you see the effect right away.
To set it, open the revenue goal of your experiment, choose the Value filter type, enter the Value filter value and save the goal.
Value filters change what the results show. Decide on a filter before you start the experiment, not after you have seen the numbers.
Next steps
- Verify measurement before you start an experiment
- Metrics: what conversions, value and improvement mean
- Statistical methods: how binary and value goals are evaluated