How do you measure interaction rate?
When comparing two variations against each other, I would find it useful to know what percentage of users interacted with the widget/recommendations/element served.
Often, the element placed on screen would require more than one click, yet the CTR shown in reports counts each click meaning that CTR is not always a good measure of interaction.
Example test and hypothesis
My sale clothing category is heavily fragmented in terms of sizing. Driving visitors to this category has a low success rate (low CR, high bounce, high exit) as customers quickly click on items that are unavailable in their size and navigate away. I create an interactive widget to AB test against my simple banner which asks users to select their size first, and hypothesise that these users will find items more easily in their size and add to basket.
Variation 1 (click to browse)
- 10 users see a simple banner inviting them to browse the sale category
- 7 click it
- CTR is 70% (7 clicks total, 7 users interacted)
- Revenue is low because most users got frustrated and left
Variation 2 (choose size, and then browse)
- 10 users see an interactive banner which asks them to choose their size, and then invites them to browse the sale category
- 4 users click to select their size, 2 of these users click to browse the category
- CTR is 60% (6 clicks total, 4 users interacted)
- Revenue is high because these 2 users both found something straight away and checked out
With DY default reports, I would see various metrics such as revenue, AOV, CR, but I would have no idea how many users or what percentage of users interacted with what was served. Indeed, a badly designed widget which users had to click on a lot would register a high CTR (most often a good sign) whilst not leading to an uplift in other KPIs.
In the above example, revenue would be up for Variation 2 despite a lower CTR but I would have no idea how many users used it, how often on average each user was clicking it, etc.
Interested to hear whether others have wanted to measure rate of interaction and how they've managed to do so.
-
Hi Mark-
Great question!
I actually just did something similar for another one of my clients here in the US. Our goal was to measure how many users not only Saw the DY-built quiz, but also Completed It (measured by clicking "get my results").
What ended up doing was creating a custom code campaign to fire an Event ON CLICK. We then mapped this event fire as a Goal (Assets > Goals) and use it as our main KPI.
In a similar situation here, you can fire an event when a user clicks on the element (i.e. their size).
Once created, you can map it as a goal and use it as your Main KPI for your reporting.Additional to note - all events will automatically appear in the CSV export of your campaign, regardless if they are mapped as Goals and/or used as main KPIs.
Hope this helps!
jake
-
Hi Mark,
Another way to more accurately represent interaction, would be to use events with handlers that would only execute once. If you have access to a front end developer, they could help you set something like the following:
<script>
let button = document.querySelector("button");
function once() {
DY.API('event', {name:'Button_Clicked_Once', properties:{}});
button.removeEventListener("click", once);
}
button.addEventListener("click", once);
</script>The snippet above, would add a handler that would listen for a click on the given element ("button"), trigger the event on click, and the then remove the handler so the event is not fired again and again on subsequent clicks.
This would help you have a metric representing using clicks on the element indication the rate of interaction.
I hope this helps with what you are trying to achieve!Anas
-
Thanks both for the great suggestions.
Knowing how both of these variations are built, I think creating an event listener which then gets removed will be an easy addition.
I presume if I use a standard name for this and set it up as a goal, I can reuse this across several experiments where I think measuring interaction will be useful.
Appreciated!
Please sign in to leave a comment.
Comments
3 comments