Transcript
In this video, we’ll explore design considerations for multidimensional health. Major topics include dimension hygiene, weight allocation, and health score ranges.
In a previous video we created a new health profile from scratch, using the Account Health model. Although many of those same concepts apply to the Multidimensional model–such as segmentation, profile order, and health recalculation–there are additional considerations when you create a multidimensional profile.
For starters, multidimensional–as the name implies–relies on dimensions (or groups) of metrics. A best practice is to ensure your Data Modeler is up-to-date so that the attributes you want to use belong in the correct dimensions.
As an example, let’s imagine that we want to create a new multidimensional health profile for our Tier 3 (or Enterprise) accounts that factors in the number of enabled users. You may remember this scenario from a previous video.
Within the Health Designer–under the Data Management Settings menu–this time I’m going to ensure I’m on the Multidimensional Health tab.
Again, we’ll copy the default profile. And then we’ll edit the profile to give it a name. And of course, we’ll set the segment criteria to Tier 3 accounts. This time, when we add a new metric for our enabled users, the dimension–Utilization–is automatically added to the definition.
Unlike in our account health definition, metrics within multidimensional profiles must have conditions set for both Good and Poor columns.
Now let’s adjust the weights, which is unique to multidimensional health. All metrics within a dimension must sum up to 100%. Since we only have one, it’s automatically set to 100%. If we scroll up and find a dimension with more than one, you can see how the weights still total 100%.
Next, all dimensions in the profile must also total 100%. Since I just added this one, I’ve already allocated 100% among my other dimensions, which is why it shows in red. I need to re-adjust allocations in other dimensions accordingly. Or, I can remove a dimension.
When I’m done editing, I can click save. Just like we saw in a previous video, the order of profiles will determine which accounts apply. So if we move this one to the top, we can see the number of accounts that apply AND the average health score according to each dimension.
Another setting that is unique to multidimensional health is the option to make changes to the ranges that classify Good, Average, and Poor. Drag the sliders left or right, but keep in mind that Poor can be a maximum of 50, and good can start from a minimum of 55.
The other setting applies to accounts with multiple levels or hierarchies. You can either calculate the parent level account health based on an aggregation of the health of its child (or sub) OR you can calculate parent health using a set of rules you define. As an example, if you choose sub accounts, you can tell Totango to make parent health poor when more than 25% of its children are in poor health.
Again, any changes you make will be applied nightly, assuming this is the active health model. You can also apply the health model retroactively, according to the date you specify.
Once the model is applied, you can see the new profile at the account level–including a numeric score–and you can expand each dimension for more details.
