This article offers best practices for setting the Basic predictive routing configuration sliders. Before you read this article, you should be familiar with the information in the Predictive Routing article. If you are using the Advanced configuration tool instead, see Predictive Routing - Advanced Use Cases.
This article contains the following sections:
After creating a new predictive routing setting, you should experiment with the default values for a few days and then see what works and what needs to be adjusted. In order to help you understand when to change from default, this article breaks each field down with use cases.
Tip: Whenever you change the default settings, you should make small changes and observe the results for a few days. Alternatively, you can split your traffic to see how different Predictive Routing configurations behave for your campaigns. You can learn more about how to split traffic in the Splitter node article.
Basic Configuration
When you leave the Use Revenue switch toggled off and the Advanced switch also toggled off, you see the basic configuration tools. These sliders offer a user-friendly and intuitive interface to experimenting with your configurations.
The settings in this area include the following sliders:
- New Target Priority
- Underperforming Target Priority
- Training Requirement
The following sections describe how to use the sliders.
New Target Priority
New targets may not have enough data to receive a fair number of calls. Use this slider to give a bonus to new targets in the routing plan until they have enough data to not need the bonus anymore.
- If you notice that new targets are getting too many calls, move the cursor a bit to the LEFT from the Default.
- If your new targets are getting less calls than you wish, move the cursor a bit to the RIGHT from the Default.
Underperforming Target Priority
Take special care when changing this slider from the default. If you move it all the way to the left, it won't route to underperforming targets at all. If you move all the way to the right, you'll give underperforming targets a lot of bonus and punish your most successful targets by sending fewer calls.
That's why the default value is set to give a small bonus to underperforming targets. If you adjust this slider, you should do it in small increments to observe the results.
Training Requirement
This slider determines how many of the most recent calls Ringba uses in its calculations.
- Less Calls: If you move the slider toward Less Calls from the default position, Ringba uses fewer calls for the calculations.
- More Calls: If you move the slider toward More Calls from the default, Ringba considers more calls for calculation, causing it to take longer to react.
Predictive Routing Use Cases FAQ
How does Ringba choose between targets when the predictive routing calculation results in equal values?
If multiple targets have identical winning revenue-per-call values, Ringba chooses randomly among the targets with the equal values. Each target has an equal chance to getting the call each time this occurs.
I removed a target from a campaign with predictive routing, then re-added the target; does Ringba reset the target's call performance?
Predictive routing keeps of Track Conversions by analytics for all clients at all times, regardless of your settings: if you remove a target from a campaign and re-add it later, the predictive routing algorithm still has access to all the target's data from before you removed it.
In fact, even if you remove predictive routing entirely from the campaign and re-add it later, the system does not lose the data.
However, removing a target can still affect the predictive routing calculations. For example, if you removed a target from a routing plan for a significant number of hours during the day but also have your Minimum Number of Hours set high, the hours the target missed could cause the New Target Bonus to apply.
Should I change the Predictive routing settings every day?
As the calculation relies on data analysis, you shouldn't change it constantly or adding to your routing plan on a daily basis. You should observe the results for a few days before making more changes.
I modified my settings today; why didn't I see my changes get applied immediately?
As described in the previous question, the calculations depend on data analysis, especially if you position the Training Requirement slider closer to the More Calls end. When the system needs more calls to do the calculations, changes take longer to occur. If you want to see faster changes, you should set your Training Requirement slider closer to Less Calls. Regardless of where you position the slider, though, you won't see changes instantly; data analysis takes time.
What if I want to test different scenarios, for example being more reactive or more stable?
Ringba has an amazing feature for doing A/B testing that is called Splitter node, find more info here: Splitter Node.
In the following example, you see three different routing plans, each with a different setting: one is the default, one is reactive, and one is stable, so you can see how each works. The traffic can be split however you like. In this case, the calls were divided equally, but you can change the percentage of calls for each dial as you like.