Measuring Digital Advertisement Performance


Impressions, viewable impressions, engagements, clicks, conversions, site sales, new customers, lead generation, brand recognition, etc. etc. There are so many ways to measure the performance of your digital advertisement campaigns. In an effort to differentiate their offerings and communicate their value to their clients, online marketing vendors have created a cluster of performance metrics. I will take a stab at making some sense of this complicated world of KPI’s.

What is the ultimate goal of a marketer?

The answer to this question, as is most things in life, is “it depends”. Some marketers may be focused on getting new customers to sign up and some may be tasked with achieving volume even at the expense of margin. Some marketers are looking for stringent ROI and some want to increase brand awareness. The one common theme among marketers is that they want to influence their audience. Whether the ad triggers a consumer to click through and buy a product on the spot, or the ad enters the consumer’s subconscious and two years later ends up in a purchase, what is important is that the ad is causing some (positive) change to the consumer’s psyche.

All measurements are flawed

Yes, all of them. Take for example, the device you are using to read this. Have you ever seen an ad for the product or the brand? Did the ad have any effect on you that caused you to ultimately purchase the device? Your likely answer is yes I have seen an ad for my phone or tablet or PC or the brand but no, I can’t tell you whether it had an effect on me. This is because our purchase decision is nonlinear and often illogical, and you just can’t quantify influence. However, in aggregate, marketing works and advertisement works. After all, you need to be aware of the existence of Apple, the iPad, and the iPhone to buy them, and it seems like a bunch of people are purchasing the products. This means, as responsible adults, we have to take a crack at measuring influence in order to figure out a way to influence consumers effectively and efficiently.

Measuring Influence

Impressions, clicks, conversions, etc. These are all proxies for influence, and they are all flawed. The graphic below describes the upside and downside of these metrics.

False Positives

Let’s take a look at our good friend Impressions. Yes we can easily count the number of impressions an ad delivered. However, 56% of online banner ads are never seen (according to Onscroll). So 56% of impressions have exactly 0% chance of influencing the user.

This is where the question of viewability comes into play. Unless the impression is viewable, it serves zero purpose for a marketer. Well great, we should only be looking at viewable impressions then. OK but most viewable impressions didn’t do jack to the consumer’s brain. It is way more likely that the consumer did not even notice the ad because she was busy consuming the content which was why she was there in the first place. When was the last time you saw an ad that actually made an impression on you (pun not intended)?

So measuring our campaigns by impressions is flawed for two reasons. 1, there are a bunch of impressions that are not viewable (which can be solved for) and 2, there are a bunch of impressions that are not influencing the consumer. By casting a wide net, we end up with a measurement that is full of false positive signals.

False Negatives

The flip side of the false positive issue is the false negative issue. Let’s look at clicks. Have you ever accidentally clicked on ad? Sure. Does it happen often? Not that often. So if you clicked on an ad, you were likely influenced by it. Something has happened in your brain.  However, when we look at ads, even if they influence us, it does not mean we will click on them. The more likely scenario is, you see an ad, register it in your brain, and move on with life.

Engagement is a measurement that alleviates some issues with click without solving the bigger issue. The definition of what constitutes an engagement is up for debate and I’m not here to settle any debate, but I look at engagement in two layers.

One is the mouse over. You point your mouse over an ad for a couple of seconds and we assume your eyes followed, so not only was this ad viewable, it was viewed. Have you ever unintentionally left your mouse over an ad? Yes, and it happens more often than an accidental click. When you look at an ad, do you put your mouse on it? Um, maybe? Nah. This measurement has both false positive and false negative issues.

The second layer of engagement is when a user plays around with the ad unit. Let’s call this “interaction” to make the distinction from the mouse over. Some ads are dynamic with buttons to scroll through different products, or zoom in features, or some interactive functions. I would never click on an ad, but I may play around with an ad just to see. The experience is not as disruptive as a click through because the user does not get sent to another page when she may or may not be done consuming the contents of the current page. These interactions are definitely influencing the consumer, even though the user is not clicking through. This measurement solves for the false positive issues, but will never cover all influential touch points, so the false negative issue remains.

In (Cheesy) Summary

Some metrics are definitively better than others, like viewable impression is a better proxy for influence than regular impression, but there are no end all be all metric. Influence is merely a concept and cannot be measured quantitatively, at least not precisely. If I can be poetic and cheesy for a second, influence is kinda like love. You know it’s there but how do you measure it? There are proxies to quantify love like the number of text messages sent, your heart rate when in front of the one, or the size of the rock on a ring, but none are accurate. The goal of measuring influence is not to be accurate. What is more important is to be actionable, because at the end of the day, the marketer’s goal is not to measure, but is to influence.


Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>