July 24, 2012

Dead Air More Effective Than Facebook Ads

The broadcast industry has a term called "dead air." It occurs when there's a mistake or a technical glitch that results in no audio on radio, or no picture on a TV screen. A blank TV screen is "dead air."

In an absolutely astounding experiment, the banner advertising equivalent of dead air -- a blank display ad -- performed better than the average Facebook ad; twice as good as the average "branding" display ad; and only one click in ten thousand worse than the average of all display ads.

Here are the details.

AdAge this week has a piece called How Blank Display Ads Managed to Tot Up Some Impressive NumbersThe article was written by Ted McConnell, exec VP-digital for the Advertising Research Foundation.

Ted and a few friends (an astrophysicist from an online analytics firm, a measurement expert from the Advertising Research Foundation, and an ad-platform wizard from a buying and optimization company) decided to do an experiment. The experiment was designed to discover how much clicking of banner advertising was actual engagement with the ad, and how much was just noise -- people clicking for no reason.

To do this they created a unique ad -- an ad with no message. A blank. According to McConnell...
"We created six blank ads in three IAB standard sizes, and two colors, white and orange. We trafficked the ads via a demand-side platform (DSP) with a low bid. We started with run of exchange, and in another phase trafficked to "named publishers" that would accept unaudited copy." 
Here are the results:
  • The click-through rate on the blank ads was .08%. According to published reports, the click-through rate on the average Facebook ad is about .05%. The blank ad performed 60% better.
  • The click through rate for the blank ad was about double the average click-through rate for a "branding" display ad (an ad without an offer.)
  • The click-through rate on the average banner ad is .09%. This means the blank ad drew one click in ten thousand fewer than an average banner ad.
  • About .04% of the clicks were mistakes. Since the average click-through rate for display ads is .09%, this indicates that it is possible that as much as 44% of banner ad clicks are mistakes.
The astounding thing is that with all the data Facebook is collecting, all the geniuses we have analyzing display ad results, all the space-age targeting we are constantly being beaten over the head with, and all the young creative prodigies lecturing us on the magic of online advertising, empty ads outperformed our online geniuses.

You simply cannot make this shit up.


Thanks to Tom Donald for alerting me to this story.

15 comments:

  1. An empty ad is an interesting one - it's not the usual bland crap you get in facebook ads.  Why is it blank? What's going on? Maybe it's something cool and edgy, trying to get your attention in a novel way, so you click.  Better test might have been just a bland uninteresting picture of a pair of people shaking hands, with a company name.  This would have had really NO possible interest, and I bet the score would have been lower...

    ReplyDelete
  2. maybe im wrong but i dont think you got the point. facebook ads are targeted using the data they collect from everyone. A blank ad is not directly targeted at anyone and if it performed equally then whats the point of wasting time and money collecting data if it really doesnt help get more clicks.

    ReplyDelete
  3. On your last bullet, I think you mean that 50% of the clicks were mistakes. 0.04% of people who were served the blank ads clicked by mistake.

    ReplyDelete
  4. Well, have you seen FB ads? a 110x80 with a limited 90 character copy is not something you would click on unless there is some really interesting thing displaying. Come on, you cannot be serious comparing those

    ReplyDelete
  5. Did the orange ad beat the white ad?  Just curious

    ReplyDelete
  6. You're right Paul

    ReplyDelete
  7. I agree. Blank ads are different because they are blank. Any "different" or new ads in market perform better for the first period of time they are in rotation, compared to months after they have been in rotation (depending on frequency and reach). So after a while people would hear about these blank ads - and know they are clicking through to nowhere. So performance would go down.... But after a long while: as John says, they could come across as edgy and weird. An ad that doesn't try to sell you something?

    Now THATS weird

    ReplyDelete
  8. Sadly though, not as edgy or weird as contextual advertising, Tom

    ReplyDelete
  9. I agree that a high portion of banner clicks are mistakes. But as I've said before, advertising is a game of what you catch, not what you spill.

    1. Facebook ads are purchased on a cost per click, so CTRs don't matter one whit. If 0.01% of users click on the ad, we only pay if they click. Bids on Facebook are about $2 per click. So the question is, what happens after you pay $2 to get one user to click on the ad?

    2. If conversion rates are even modest, say 2%, a $2 cost per click on Facebook at 2% conversion to sale = a $100 cost per sale. This doesn't work for selling chewing gum, but for many products, say, a college MBA program, home roofing, or a new car, a $100 cost per customer acquisition is a bargain.

    3. While most impressions are ignored, there's also value in the advertising itself. I'm not clicking on your book for "The Ad Contrarian" at above right, but I'm intrigued, and I may chase it down later. We've done pre- and post-exposure studies to consumers that show that after being exposed to digital advertising, intent to purchase a product rises 24%.

    Banner blindess is a huge issue, and marketers are fighting it with testing either performance-based buys (cost per click, cost per action) or high-impact units (that break out like, say, magazine ads and are seen by more people).

    Yes, digital advertising has huge waste. So does all advertising. That's what makes making it work so challenging.

    ReplyDelete
    Replies
    1. Thanks Ben. Buy the book, then we'll talk.

      Delete
  10. OK.... accidental clicks are 4 in 10,000 and an effective brand display ad is 8 in 10,000 clicks

    So, 50% of Display Advertising really is wasted... and we still don't know which 50%

    ReplyDelete
  11. Forget the exact maths (yes we have an s) - some people see digital ads and some of those people (o.05 to 0.09%) click it, of those people up to 44% made a mistake. 

    As noted the clicks are paid for.

    Ben - I think you'll struggle to convince many people that an on-line ad caused someone to buy a motor or of all things a roofer, you have to be kidding me.

    Sign up for a free packet of chewing gum seems much more likely, which cost the advertiser $100 to give away a packet. 

    ReplyDelete
  12. This experiment should have been done by comparing targeted ads to random ads, rather than a blank ad, which as others have pointed out is in itself interesting due to its novelty.

    ReplyDelete
  13. Brand ads simply don't require a click-through, as the objective is raising awareness and should be quantified with other metrics such as hovering, eye-tracking, and purchase intent. 

    ReplyDelete
  14. If we're seriously discussing how attractive and interesting a blank space is, civilisation is doomed.

    ReplyDelete