Mida allows you to control who sees an A/B test using segment filtering and event filtering. These filters ensure that experiments are only shown to relevant users.
Segment Filtering
Segment filtering targets users based on historical user attributes. An A/B test will only be shown to users who have previously met the defined conditions.
Segment filters operate on attributes that were set during earlier interactions and stored in Mida.
Segment Matching Conditions
1. Existence
Value | Label | Description |
| Exists | The attribute exists for the user |
| Does not exist | The attribute does not exist for the user |
2. String
Value | Label | Description |
| Is equal to | The attribute equals the given value |
| Is not equal to | The attribute does not equal the given value |
| Contains | The attribute contains the given value |
| Does not contain | The attribute does not contain the given value |
3. Boolean (attribute-based)
Value | Label | Description |
| Is true | The attribute value is true |
| Is false | The attribute value is false |
4. Number (attribute-based)
Value | Label | Description |
| Greater than | The attribute value is greater than the given value |
| Greater than or equal to | The attribute value is greater than or equal to the given value |
| Less than | The attribute value is less than the given value |
| Less than or equal to | The attribute value is less than or equal to the given value |
5. Date (days ago)
Value | Label | Description |
| More than (days ago) | The attribute occurred more than X days ago |
| At least (days ago) | The attribute occurred at least X days ago |
| Less than (days ago) | The attribute occurred less than X days ago |
| At most (days ago) | The attribute occurred at most X days ago |
6. Time-based (day of week / hour of day)
Value | Label | Description |
| Is equal to | Matches the specified time value |
| Is not equal to | Does not match the specified time value |
| Is after | Occurs after the specified time |
| Is after or on | Occurs after or on the specified time |
| Is before | Occurs before the specified time |
| Is before or on | Occurs before or on the specified time |
User Identification (Required for Segments)
Segment filtering requires users to be identified so attributes can be stored and reused.
Identify calls support queue-based usage and should be sent as early as possible.
<script>
window.mdq = window.mdq || [];
window.mdq.push([ "identify", {
id: 123,
plan_id: "premium",
company: "Example Company"
}]);
</script>
Event Filtering
Event filtering allows you to target users based on historical events they have triggered.
Event filtering is split into two distinct concepts:
Which event the user has triggered (targeting)
Conditions on event properties (filtering)
Event Rules (Which Event Was Triggered)
Event rules are used to decide whether a user is eligible to see a test based on event history.
Only two matching conditions are supported.
Value | Label | Meaning |
| Equal | The user has triggered the specified event |
| Not Equal | The user has not triggered the specified event |
Event Attribute Conditions (Event Properties)
Event attribute conditions are used when filtering events or conversions by properties attached to the event.
Value | Label |
| Equals |
| Not equals |
| Contains |
| Not contains |
| Greater than |
| Greater than or equal to |
| Less than |
| Less than or equal to |
Event Tracking
Events must be tracked in order to be used for event filtering.
Queue-based tracking is recommended.
<script>
window.mdq = window.mdq || [];
window.mdq.push(["track", "subscribe"]);
</script>
Tracking events with properties:
<script>
window.mdq = window.mdq || [];
window.mdq.push(["track", "purchase", {"price": 120, "currency": "USD"}]);
</script>
HTML-based tracking:
<button mida-event="click-cta">Create a free account</button>
Summary
Segment filtering targets users based on stored user attributes.
Event filtering targets users based on historical events.
Event rules determine whether a user has triggered an event, while event attribute conditions filter based on event properties.
Using both together ensures experiments are shown only to relevant users.