What Are Unique Metrics?

• Also known as non-aggregable metrics
• Refer to data points that cannot be summed up or averaged across different time frames or dimensions without losing accuracy
• Used when tracking individual user interactions, such as unique impressions or unique visitors. For example:
  • If a user visits a webpage multiple times in a month, they will be counted as multiple impressions but only as one unique visitor.
• If unique metrics are incorrectly summed over multiple days, the results will be inflated and inaccurate. The table below illustrates this concept:

2 unique impressions is the correct metric. However, if we try to calculate the unique impressions for the month by adding them all together, we would get 5 unique impressions, which is incorrect. It’s more than twice the correct number of unique impressions.
When aggregating the data together, or rolling it up, we don’t have the data necessary to be able to dedupe the users to be unique.

How Unique Metrics Function in NinjaCat

Unique metrics require specific handling in NinjaCat to maintain accuracy. Here are key characteristics of how they function:

• Data Storage & Refreshing: Unique metric data is only stored temporarily for two days if it includes recent data. After that, it is discarded and retrieved fresh when needed.
• Date Range Considerations: The system caches unique metric data for reports and dashboards. However, any modification to the date range results in a new API call, as the data cannot simply be recalculated from previous requests.
• Filters & Aggregation Issues: Applying filters to unique metrics can lead to inaccuracies because it forces segmentation and subsequent re-aggregation, which NinjaCat cannot accurately perform.
• Use in Datasets: Unique metrics should not be used in datasets, as they will be aggregated incorrectly. This can cause significant discrepancies in reporting.

Challenges with Unique Metrics

• Filtering Impact: Filtering by campaign, ad set, or keyword can make unique metrics inaccurate. The only exception is Facebook’s “reach” metric, where filtering by campaign and ad set is natively supported.
• Data Caching Limitations: Dashboards cache the initially loaded date range for unique metrics, but modifying this range forces a fresh API call.
• Performance vs. Accuracy Trade-offs: While unique metrics ensure precision, they may result in slower data retrieval due to the necessity of real-time API calls.

So what now?: Best Practices for Working with Unique Metrics

Assuming you are trying to work with a unique metric, there are a few ways to report on unique metrics in NinjaCat accurately. 

1. See if there is a "unique" version of the metric in our Metrics list. You'll notice the word "unique" in parenthesis if this option is available. These options calculate the unique metric differently than standard metrics and can result in a more accurate total.

2. When presenting a unique metric, try to avoid using scorecards. A table or graph allows you to select a dimension to measure, which will give you some more control. This allows you to select a dimension that is specific to your data source as opposed to an aggregate data source. A network specific data source will calculate your metrics similar to how you see your metrics in your data source. You can tell a dimension is network specific if it is listed underneath your data source, as opposed to under the Aggregate options in the metrics dropdown.

• Avoid Summing Unique Metrics: Always retrieve unique metrics at the level of granularity required, rather than summing across multiple periods.
• Minimize Filters When Using Unique Metrics: Filtering by dimensions that are not explicitly included in the report can lead to incorrect results.
• Use Dashboards & Reports Wisely: Ensure that dashboards reflect a stable, predefined date range for unique metrics to improve performance.
• Be Cautious with Datasets: Avoid including unique metrics in datasets, as they do not support real-time API calls and will lead to incorrect aggregations.

If none of these options help you, or you have more questions on unique metrics, you can always contact our Customer Advocacy Team who can work with you to see if there is any way to get the data closer to what you are seeing from the source. Just click on the bot icon in the bottom corner of your NinjaCat account to request a call, or submit a ticket.