Proxy Metrics: Use, but with Caution

There are strong cases to be made for and against proxy metrics.

By proxy metrics, we mean numbers outside of our KPIs that are predictive of down-funnel performance. An example would be using the “add to cart” metric to predict purchases. Good proxy metrics follow the same goal as B2B lead qualification – it’s not revenue yet, but the data can help advertisers understand how certain indicators will translate into bottom-funnel activity.

Proxy metrics are also useful for newer accounts that need to track a broader picture of demand generation and brand awareness. They’re essentially micro-signals (think LinkedIn page follows or direct site visits) that can be tracked to show brand growth over a long period of time. 

So all of that makes sense, and there are a million ways to establish these correlations and incorporate the numbers into your reporting/bidding and media mix. But they can be gamed. 

First, if you’re taking the metrics at face value without considering the channel, your understanding of the metrics’ actual impact will be foggy. We like to use proxy analysis early on in a program where branded organic is the lion’s share of the traffic. We establish an engagement-to-conversion ratio and then apply that logic to something like non-brand paid search or, if you want to be extreme, something like display prospecting. You will see very quickly that the value of similar proxy engagements varies widely by channel.

Second, if you’re using a proxy metric in certain ways, your conversion rate may be at risk. Let’s say you’re doing Facebook prospecting campaigns, and because of low conversion data density, you move the target optimization to add to cart to give the bidding algo more data to work with. In theory, that’s fine, but when you make “add to cart” the goal, the algorithm goes only after add to cart engagements, and the quality of the engagement drops.

Used well, proxy metrics are good for marketing. But used poorly, they can make certain channels look more effective than they are, and they can dilute the quality of engagements you get for your marketing budget.