Dimensions and "True Blanks"

Like many other customers, you likely have a need to examine data by organizations, layers, or other dimensions within your company. For instance, you may want to examine Headcount (EOP) by Supervisory Chain to better understand how your headcount is distributed across levels of management.

One Model can successfully build dimensions to help you slice the data in such a way. However, when you filter your view, you may find some "Unknown" values aren't able to be filtered out. On the surface, it's easy to be confused about why this is happening, as genuine "Unknown" values can be filtered out. The values which can't be filtered out, while labeled as "Unknown", actually represent true blanks, or blank values created when individual employee records are pivoted by multiple levels of the supervisory chain and similar dimensions. In this article, however, we'll focus on Supervisory Chain to help explain the concept.

The first step is understanding what genuine "Unknown" values are, or how data fields traditionally get mapped to "Unknown". For example, below you'll see a table with employee location data:

As you can see, Employee 3 is missing location data. In this particular case, Employee 3 is missing data under "City/Town", causing "?" values to carry all the way through to the top level. Since location dimensions (like many other dimensions) in One Model are mapped from the bottom up, a missing value at the bottom will cause "Unknown" values all the way through to the top level for an employee. In this case, the "?" values would be read in One Model as "Unknown" values, which you can easily filter out within a query. However, if a proper "City/Town" value is inputted for Employee 3, the employee will map to a value at all levels in the dimension and you will no longer see "Unknown" values.

The second step is explaining the logic behind true blanks. To understand how true blanks manifest in One Model, and differ from genuine "Unknowns", let's start by visualizing a supervisory chain. Below, we've provided a simple example of what this may look like:

Now, let's use this example to see where a sample of employees may end up along this chain. In One Model, and in various HRIS systems, the data is structured in the following way:

As you can see above, the breakdown of where employees stop in the chain is:

If you focus on the Executive column, you can count along the rows to logically see how many employees roll up to the Executive level. Doing this, you can see that there are a total of 6 employees.

Maintaining this approach on the Mid Level column, you can count how many employees roll up to the Mid Level, which equals a total of 5 employees.

However, notice the blank cell in the Mid Level column for Employee A. While you may understand this reflects how Employee A doesn't proceed along the chain beyond the Executive Level, One Model will identify this empty cell as a unique and inherent piece of data, or a true blank.

From this perspective, if you visualized this on this level in a bar chart in One Model, you would see the true blank created by Employee A's record in the "Unknown" category like in the bar chart below. In this view, essentially, you're seeing how Employee A exists within the company, but doesn't roll up to the level you have filtered to.

If you maintained this approach of counting each row with data along the chain, you would find that, the further down the chain you go, the more true blanks appear. Your true blanks will always consist of employees who sit above the level you're examining.

Visualizing the data on this level in a chart would look like:

Now, it's important to emphasize that you are selecting an entire level for each of the charts above. If you were, conversely, to select specific nodes in a level, you would not see true blanks appearing in your chart. Please refer to the illustration below:

As you can see, by selecting the node "Front Line", the 3 true blanks in Level 3 do not appear. The reason why this happens lies in what you're effectively telling One Model to do when you make your selections. When you select the entire Level 3, you're giving a general command and telling One Model to visualize all of the records (true blanks and all) that fall within that level. On the other hand, when you select specific nodes in a level, you're telling One Model to only visualize records within Level 3 that specifically have the string "Front Line" in them.

To conclude, without intervention, true blanks will be categorized as "Unknown" when they do appear, though they cannot be filtered out in the same way as genuine "Unknown" values when levels are selected. Consequently, this is why you may see that some "Unknown" values cannot be filtered out while others can. Our current solution is to help end users label true blanks separately from genuine "Unknowns". For example, our data engineers can label these true blanks as "Above Selected Level", thus providing clarity in what end users are actually seeing.

If you're interested in creating a new label for true blanks, or have any questions on this concept, please reach out to your Customer Success Lead.