FouAnalytics Attentiveness versus Attention

From Viewability to Attention to Attentiveness

There's been a lot of attention around "attention" in digital marketing recently. That's good, because it is a step up from "viewability" which occupied the minds of advertisers and media buyers for years. Viewability refers to whether the ad itself had the "opportunity to be seen" -- e.g. 50% of the pixels of a display ad in the viewport for 1 second, or 50% of the pixels of a video ad in the viewport for 1 second. But, the ad having the opportunity to be seen is not the same as a user looking at the ad, or paying "attention" to the ad, as it were. So attention metrics were born, eye tracking studies were done, and media agencies are all out selling "attention metrics" to their clients as the latest shiny object.

I have no objection to this, because directionally, buying ads based on "attention" does marginally help you avoid buying on the shadiest of fraudulent sites. But complications exist. For example, MFA sites tend to exhibit far higher attention metrics in the ads because the humans that do read boredpanda do actively scroll down the page, click to the next article, etc. And fraudulent sites might also be using bots that easily fake mousemove, page scrolling, clicks, and touch events. Buying on attention may also expose you to more of these types of fraudulent sites if you don't have the analytics in place to "see Fou yourself."

Viewability - using FouAnalytics in-ad measurement

If you have FouAnalytics in-ad tags measuring your ad impressions, you can use the following fields to check the viewability of your ads, compared to what the platforms, legacy viewability vendors, and your media agencies are reporting to you. You might be surprised to learn they have been over-reporting viewability for years (e.g. in the 60 - 90% range).

In the examples on the left above, you can see viewability across different campaigns range from 10% to 47% (viewable:1). You can filter for "viewable" or "viewable-MRC" in FouAnalytics. Viewable-MRC adds the 1 second requirement. Note in the unload_ms (milliseconds) data grid above, if the user left the page in under 1,000 milliseconds (1 second), viewable-MRC = 0 (the ad did not satisfy the 1 second requirement to be marked as "viewable" according to the MRC standard). Ask your legacy verification vendor to provide you with this level of detail so you can confirm for yourself. (hint: they wont, because they can't).

Attention

Everyone is obsessed with "attention" right now. But did you know none of the attention vendors can actually measure whether someone was looking at the screen (and paying attention to the ad)? That's right. Javascript tags in ads do not have camera permissions to turn on the camera on your smartphone or laptop to detect whether someone was in front of the device and looking at the screen. If the javascript attempted to access the camera, it would generate a user notification and permissions prompt. Attention vendors did conduct eye-tracking studies in a laboratory environment, and yes, larger ads get more attention than smaller ads. And they use crawlers, page layout and algorithms to approximate the attention you will get from certain ad units. That is different from whether someone was actually paying attention. That is a problem.

Engagement (legacy definitions)

Before we move on to attentiveness, which can be directly measured when campaigns are running, we should address the legacy concept of "engagement." Too often, marketers conflate clicks or click through rates ("CTR") with engagement. On the surface, they say "users engaged with our ads when they clicked on them." So they use clicks and CTRs as a proxy for engagement. But note that bots click on ads too, to trick advertisers into thinking those ads were performing well. Further, "engagement" on websites usually mean time on site or pageviews per visit. For example, advertisers assume that more time on site or more pages per visit means more engagement. But did you realize that someone could be spending more time on your site because they couldn't find what they were looking for? And they left your site unhappy? So more time on site is not necessarily better. Also, less time on site is not necessarily worse because the person could have found what they needed right away, and left happy. So legacy definitions of engagement are due for a re-think. And metrics such as clicks, click through rates, time on site, or pages per visit should also be questioned as measures of engagement. These metrics also don't correlate well to effectiveness and outcomes either. So what is a marketer to do? Consider the attentiveness of the users on the site/landing pages. Read on.

Attentiveness - using FouAnalytics on-site measurement

Attentiveness means that the users that arrived on the landing pages did something else after they got there, like move the mouse, scroll the page, click something, etc. Before adding FouAnalytics to their sites and landing pages, many advertisers have told me that they observed high bounce rates and low time on site in Google Analytics. This implied that the visits they got from various paid media channels were not of high quality. They even hypothesized (correctly) that the clicks were from bots, because bots have no incentive to waste time and energy faking interactions on the landing page). But beyond knowing that these visitors were not of high quality, there was nothing much else they could do with the data from Google Analytics or Adobe Analytics.

Using FouAnalytics in the ads allowed practitioners to optimize away from fraudulent sites and apps. Doing so meant that fewer clicks coming to the site are obvious bots and low value visitors (the ones with high bounce and low time). Once you have done the above, you are ready to move on to more advanced steps -- optimizing for humans that do something, take some sort of action, after arriving on your landing pages -- i.e. attentive visitors.

Let me now show you what I and other FouAnalytics practitioners have looked at over the years to further optimize campaigns towards attentive visitors. You're already familiar with the concept that dark blue means humans. The slide above shows a lot of dark blue, measured on-site. In the small insets below the time series chart, you see "clicks on desktop" and "touch events on mobile." In FouAnalytics, if you go to the clicks tab you will see three sections -- clicks, mousemove, and touch events -- grouped by screen resolution. 1920x1080, 1366x768, etc are desktop monitors (in landscape) which show clicks. 390x844, 414x896, etc are smartphone screens (in portrait) which show touch events.

FouAnalytics attentiveness

When humans deliberately click an ad, they want to get more information when they arrive on the landing page/website. So we expect them to do something when they arrive on your site -- we expect to see mouse movements, page scrolling, clicks, and touch events.

The data grids below show this. "mousemove-exists:1" shows the percentage of pageviews that have mousemove. In the example below, 52.7% have mousemoves. There are valid-clicks in 46.8% of the pageviews, and valid-scrolling:1 in 32.4% of the pageviews. In mobile:1 environments (33.9%), we expect to see touch events. Touch-exists in 23.2% of the pageviews. That's about 2/3 of the mobile pageviews (23.2% / 33.9%). The percentages of pageviews that have mousemove, clicks, pagescroll, and touch events gives you a sense of the attentiveness of the users. More attentive users are better, obviously. That means when they arrived on your landing page, they actually did something.

For a contrasting example, see the set of data grids below. It shows mouse moves in only 11.4% of the pageviews, clicks in only 9.9%, and 0% pagescroll. Touch events are similarly very low. These visitors are not very "attentive" and are therefore less valuable to you. These are what you previously saw as "high bounce" or "low-time-on-site" visitors in Google Analytics. These users, some bots, didn't do anything after they arrived on your site.

If you copy and paste the following parameters into the Filter bar in FouAnalytics -- valid-clicked mousemove-exists valid-scrolling touch-exists mobile -- and click [FILTER], you will get the 5 grids shown in the examples above. This way, you can check the attentiveness of the users on your site.

The summary data below shows 5 major advertisers using FouAnalytics to measure the attentiveness of the users that arrive on their landing pages from various forms of paid media. These are compared to the direct traffic (users that came to the site directly, not by clicking on an ad of some sort -- yellow highlight). By comparing the rate of clicks, mouse move, scrolling, and touch events of the visitors coming from each of the paid media sources against these characteristics of direct visitors, you can see if the visitors coming from paid media were more attentive or not. You can also see the relative ratios of bots versus humans. Note in all of the examples below where Facebook was the paid media source, it showed the highest human visitors, which also exhibited the highest "attentiveness" -- i.e. most of them did something when they got to the landing page. This bodes well for conversions, since humans have to take action on the site before they can complete the purchase. Attentiveness has thus been a far more effective indicator of success than other vanity and quantity metrics and also metrics like attention (to the ad).

If you want to use FouAnalytics on your site to measure attentiveness of the visitors, please message me. If anyone has questions, please screen shot something from the dashboard and email me. If I can be of further assistance on the topic of "attentiveness" please let me know.

Further reading: https://www.dhirubhai.net/today/author/augustinefou

And yes, I have been looking at this kind of data in FouAnalytics for clients for many years. In the slide below from 2018, you see the ratios of human (dark blue) versus not human (dark red) across three steps -- "arrived" on the landing page, "clicked" something on the page, and "converted" (e.g. made an ecommerce purchase). Note that humans converted, bots didn't.