The Self-Defeating Nature of Facebook Ads

The nature of the Facebook Ads system is such that numerous factors will inevitably lead to click-through rate decline. Unlike Google’s Quality Score, where CTR is a defined metric relating directly to impressions, Facebook does not give an indication of CTR’s effect on impressions – but we’ve seen its effects. It’s unclear whether or not bid costs go up, but certainly impressions go down, as they would; no point in showing ads people don’t click on.


The advantage Facebook ads are meant to have over Google’s content network is the ability to highly target your ads using demographic information. In setting up an ad, you can target location to city granularity, age to the year, gender, relationship status, language, education, workplace, and a general keyword matched interest. Indeed, without a Quality Score connecting a search term to ad copy to a landing page, demographics are all you’ve got.

The thinking goes, by hitting an extremely targeted group of people, you will have the best possible click-through rate and ad performance. But the more specific you get, the smaller the group, and the greater your odds will be of a decreasing click through rate, and thus your impressions, and your total clicks. But if the targeting is specific, wouldn’t the opposite be expected?

Facebook Fatigue

The same ads are being shown again and again to the same people, selected based on demographics-driven targeting. This multiple display can bring about numerous situations:

– The user hasn’t clicked yet, but after enough hammering away at them, they eventually cave and check it out.
– The user hasn’t clicked, and never will.
– The user has clicked, and keeps getting shown the ad anyway.

Situation 1 is the only one worth paying for, and has to represent the minority of cases. Even then, it can turn into case 3.

The problem is, as a Facebook advertiser, you have no indication of when this saturation point has occurred, other than a dropping CTR. You have an idea at the beginning of roughly how many people you’re targeting with “estimated reach”, but you don’t really know who you’ve reached and how many times.

Out of 1000 that fit the demographic, what if 999 never even log in to Facebook, and the remaining person is a heavy user? This is an exaggerated case, but it makes the point; demographic reach gives no indication of the user behaviour of this “reached” audience. This information is extremely relevant to a CPC setup, even more to the CPM setup.

Google’s Content Contrast

Contrast this with keyword-matched search behaviour, where being on a particular page is indicative of intent, to the extent to which Google has done a effective job of pairing Google ads with the content.

Google’s content network isn’t flawless, and sometimes, in the effort to match ads to content, the similarity is strained. But this is a worst case scenario, where the best case is relevant for the given user and as such the click-through rate has a good chance of being reasonable. If the same ad is being shown to the same surfer again, the fact is that they are again surfing related content, and there’s reasonable demonstrated intent all over again.

Real-Time Solutions

“Likes” and current demographic data can potentially loosely match, but still have no mechanism of determining time-sensitive intent.

24: Jack (Kiefer Sutherland, R) and the team at CTU set a trap in order to capture the unsecured nuclear rods in the “1:00 – 2:00 AM” episode of 24 that aired Monday, March 1 (9:00-10:00 PM ET/PT) on FOX. ©2010 Fox Broadcasting Co. Cr: Kelsey McNeal/FOX

To creepify things a little, the best potential for time-sensitive demographic application and ad serving is based on real-time analysis of user statements. This is something Facebook has begun testing with a “small” pool of 6 million users. Such a method would be more effective than the current setup, but still not as representative of proactive, actionable intent, and not as much as content browsing.

The Core of the Problem

Back to fatigue. Once the pool is saturated, recognized only because your CTR has dropped, you have some options. You can try to re-make and jazz up the ad to something different, and we recommend doing so. You can also widen the demographic target – get less targeted – which will potentially increase your total volume of clicks and counteract fatigue, but the decrease in the specific targeting will result in, of course, lower click-through rates.

An alternative is to start wide and get more specific, so that you can benefit from the increase in CTR as you get more targeted. But there’s something rather counter-intuitive about having an ad system like Facebook’s and deliberately not being as targeted as possible!

More CTR Murder

For seasonal businesses, ads served by demographics alone are bound to provide a worse CTR during the low season compared to the high. While that may seem obvious, the fact is that keyword-driven navigation won’t have this problem, or nearly to the same extent.

With keyword searches, you serve up the ad with the matching content, and the targeting remains, even if at lower volumes. If the surfer is browsing content relating to the low season keyword, they get the relevant ad. If they’re not, they don’t. So, the while impressions and clicks may drop to match lower interest, the click-through rate will should not and you should suffer no penalty.

Remaining Questions

This has been a very complaint-filled post, raising questions and only sometimes providing answers. In many cases, perhaps the seasonal businesses case, there is no easy answer for Facebook because it is a consequence of the nature of the platform. It’s not their responsibility to be an optimal platform in all cases, and they should be able to just play to their strengths. If Google has them beat in some areas, it’s because Google Content Network ads are simply a different animal, and it’s up to the advertiser to be savvy enough to know the difference and act appropriately.

That concession made, there’s no excuse for the lack of tools provided for advertisers to help predict and avoid fatigue, including potentially refining “reach” to factor in or account for the varying degrees of use per user. Also useful would be some way of having an idea of how often your impressions are being seen by the same eyes, so you can make informed decisions. This mystery is likely good for Facebook’s bottom line, but it certainly doesn’t help yours.

Also missing is any real clarification of what Facebook’s equivalent of a Quality Score is or would be, and CTR’s role therein.

Or maybe I’m demanding too much out of my ad platform?

Leave a Reply

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