Analytics Strategy 101: Choosing Your Metrics

Reading Time: 5 Minutes


A metric is simply a unit of measure which helps you quantify something. When it comes to marketing and communications there are 5 different categories of metrics that serves different purposes and which can help your organize your KPIs (see How to Categorize Your Marketing Metrics). However, in order to ensure you choose the right metrics to quantify performance you also need to understand the different types of metrics, what they tell you, and when it's appropriate to choose one metric type over another.

Overall there are three types of metrics; 1) volume, 2) ratio and 3) meta:

Volume Metrics

Volume Metric Examples

Volume metrics represent an absolute measurement of something. They consist of a single unit of measure and are generally represented as a COUNT or SUM (not a percentage). Overall, volume metrics tell you things like how many times did something occur or what was the total sum of something. 


Volume metrics are helpful for providing context when evaluating performance. For example, if you are comparing the conversion rate between 2 websites, understanding the differences in traffic volume between the two domains might provide helpful context in terms of determining why the conversion might be different. In this case, looking at volume metrics like visits, unique visitors or pageviews would be useful.


Volume metrics (e.g. impressions, visits) can sometimes be used for the sake of vanity as they often produce large numbers. Also, volume metrics may not be helpful when it comes to comparing some data sets that are not relatively equal in shape or size. For example, if you wanted to compare performance between 2 different websites you would generally look at visits (a volume metric) as many external factors can affect traffic volume (e.g. population, market share, etc). In this case, non-volume based metrics like Bounce Rate, Time on Site or Average Pages Per Visit would be better suited for comparison.

Ratio Metrics

Ratio Metric Examples

Also known as weighted metrics, these are relative metrics which are based on 2 disparate units of measure where one unit is divided into another. Ratio metrics are commonly represented as a percentage (e.g. Bounce Rate = 64%), but can sometimes also be represented as a decimal (Average Pageviews Per Visit = 1.45).


Ratio metrics are generally useful for comparison as they provide relative values which normalize your data. For example, bounce rate (a ratio metric) is a measure that can be easily compared between two websites and which tells you about visitor quality. On the other hand, comparing a volume metric (such as visits) between two sites doesn't give you the same level of insight. 

As another example, imagine you have an eCommerce website which operates globally. It may not be helpful to compare a volume metric for conversion (e.g. total transactions) as a means to understand performance across multiple markets as many external factors can affect sales volume (e.g. population, market share, etc). As a result, comparing a rate of conversion (e.g. total visits / total transactions) does give you a number that can be easily and meaningfully compared across markets.


In some cases ratio metrics lack sufficient context and may not tell you the whole story. For example, imagine you were interested in understanding the difference in visitor quality between 2 websites by comparing bounce rate, and you found that website A had a bounce rate of 50% while website B had a bounce rate of 35%. In this case if you only looked at bounce rate you might be inclined to conclude that website B is performing better. But what if you then considered traffic volume and discovered that website A had generated 20,000 visits while website B only generated 1000. Indeed, website B had a lower bounce rate but context is everything, and website A drove 20 times more traffic with an acceptable bounce rate. 

I mentioned earlier that volume metrics can be problematic as they are sometimes used for vanity and/or are not always ideal for comparison. However, they can provide useful context when comparing ratio metrics which is why I always recommend never looking at any metric in isolation. Therefore, it's good to always have a mix of both volume and ratio metrics in your KPI mix.

Custom Metrics

Example Bespoke Metrics

Meta metrics are customized metrics which consist of 2 or more disparate units of measure. These are the most complex of the 3 metric types and are generally used to simplify and consolidate multiple metrics into a single unit of measure. Meta metrics usually require a framework as well as consideration into how you normalize, weight and calculate the end value.


Meta metrics can be very useful when it comes to Executive/C-suite reporting as they can help you simplify multiple metrics into one place. For example, say you have 3 different measures for looking at visitor quality on your website (e.g. bounce rate, pages per visit, average time on site), but you don't want to report on all 3 metrics to your executive. In this case, you could create a meta metric which factors all 3 measures together into a single Website Engagement Score.


Meta metrics can be very difficult to develop and maintain. They generally require an underlying framework which guides things like how your meta metric weights different units of measure and how you normalize the data. Furthermore, there is generally no one right way to develop a meta metric but there are certainly many wrong ways, and if not done properly your meta metrics might end up of being based on faulty calculations or arbitrary weightings. As such, you should always consult a professional when it comes to building meta metrics.

Summary & Helpful Tips

  • Make sure you understand the difference between volume and ratio metrics, and know when to use them.
  • Try to have a mix of both volume and ratio metrics included in your core Key Performance Indicators (KPIs)
  • If you want to develop a custom metric, consult a professional before putting it in practice

Giving a presentation? Check out the Slideshare version of this post