Getting an overview of your data isn’t always as simple as it seems. Anyone dealing with data, from marketers to analysts, needs to beware of particular metrics that can’t be added or summed — or risk the credibility of their data. These kinds of data points are called non-aggregatable metrics, or nonags (or even nags!)
In this blog, we’ll take a look at what non-aggregatable metrics are, why it’s important for data analysts as well as marketers to understand them, and what happens when you try to aggregate a non-aggregatable metric.
A non-aggregatable (nonag or nag) metric is a quantitative value that can’t be summed or averaged. This is because this value depends on a more granular level of data that is not given.
A few examples of nonag metrics include:
As a result, such metrics can only be displayed at their given granularity level, otherwise, discrepancies will be generated.
Adding or averaging nonag metrics will give false results, so it’s important that marketers understand which metrics can’t be aggregated. There are three common types of nonag metrics:
1. Running totals
Examples: Inventory, follower count, subscribers
Adding up the daily follower count across a week will give marketers an inflated total. Instead of quantifying the followers over a period of time, you’ll be registering your total number of followers multiple times to end up at the inflated number of 5,651. Likewise, taking an average by dividing this number won’t tell you much either.
| Day | Followers |
| Monday | 789 |
| Tuesday | 792 |
| Wednesday | 803 |
| Thursday | 809 |
| Friday | 815 |
| Saturday | 820 |
| Sunday | 823 |
| Total | ? |
2. Unique metrics
Examples: Reach, unique impressions, cookie impressions, ga:users
Reach is an estimation of how many people saw your ad. But if we add the numbers generated for reach across two days together, then we may end up inflating the numbers beyond their true value.
| Day 1 | Day 2 | Day 1 + 2 |
| Maise | Anna | Maise |
| Marcus | Maisie | Marcus |
| Charlie | Matt | Charlie |
| James | Anna | |
| Charlie | Matt | |
| James | ||
| Reach = 3 | Reach = 5 | Reach = 6 |
This is because we can’t know for sure how much overlap to account for. It’s possible that some of the people who saw the ad on the first day are the same people who saw the ad on the second day. Because most advertising platforms do not share user IDs, there’s no clear way to get an accurate number when aggregating reach results from two separate time periods.
3. Calculated KPIs
Examples: CPC (Cost per click), CPM (Cost per mille), Conversion rate, CTR (Click through rate)
Scenario: What happens when you try to aggregate CPC?
Because we’re working with weighted averages, adding them up will not give an accurate representation. The CPC depends on the context of the number of clicks. If we average CPC for a week by adding the CPC across seven days and then dividing by seven, the value won’t account for this.
The table below shows CPC across 7 days. If we were to average the CPC across 7 days by adding the CPC per day and dividing by seven, we’d get the inflated average CPC of $0.91
Instead, we need to look further beyond the CPC, into its basic components. If we are to aggregate CPC over a week accurately, we must take the total spend for that week and divide it by the total number of clicks that week to get the average CPC, which is $0.80.
| Mon | Tues | Weds | Thurs | Fri | Sat | Sun | Total | |
| Spend | 15.00 | 15.00 | 10.00 | 10.00 | 17.00 | 15.00 | 10.00 | 92.00 |
| Clicks | 17 | 12 | 21 | 15 | 30 | 10 | 10 | 115 |
| CPC | 0.88 | 1.25 | 0.48 | 0.67 | 0.57 | 1.50 | 1.00 | ? |
In other words, we need to apply the same formula on the overall period, which is why calculated KPIs must always be executed in the BI tool of choice, with clear naming conventions.
If marketers aren’t aware of the types of metrics that can't be aggregated, the quality of their data is at risk. This means that the data-based decisions they’re making might be doing more harm than good, and ultimately, damage may be done to the reputation of the marketing team.
Investing in a strategy based on false numbers is a major risk for revenue and one that marketers can’t afford to take. But it’s not just about optimizing the budget, the reputation of marketing teams is also at stake.
A good example of this is a use case where an agency tried to aggregate metrics for reach for a campaign that ran in Belgium. The data was incredibly granular, looking at reach per ad per device, across different regions, and so on. Because there was so much overlap within these categories, the total reach calculated came out to more than the total population of Belgium.
Agencies and marketing teams need accurate data to build their business case for the impact of marketing — but if the data is inaccurate, then rebuilding trust in a data strategy can be incredibly difficult.