24 January 2021

How Short Should Ads Be?

Ads keep getting shorter, but not subliminal.

Way back in TV advertising history, there were 60-second ads. Some people still remember those; there was even an article in The Atlantic looking back on them, wistfully.

Ads used to be longer

Like most GenXers, I grew up with :30s and :15s on TV. Conventional wisdom, after I joined the industry but before bandwidth permitted online video, was to run :30s until awareness reached some level when :15s could take over as reminders or reinforcement.

Miller High Life ran 1-second ads in
the 2009 Super Bowl, featuring
the late, great Windell Middlebrooks
During my international career, I became familiar with :20s and even :10s. In Argentina, where I lived and worked for three years, ads could be any length client and agency wanted, because media time was bought and sold on a second-by-second basis. So, we made :17s, :36s, :52s, etc.

Maybe that Argentine flexibility is going global. 30-second units are still common on linear TV, but audiences can be reached on other platforms that allow for all sorts of possibilities. Bandwidth has improved and the shift to mobile devices and mobile-friendly formats, like YouTube and TikTok, permits shorter ads and new rules for what makes effective communication. The six-second format is common.

But does the six-second format work?

Shorter ads can work, but...

Magna Global, the IPG media research hub, has studied ad lengths across different video platforms. Their 2015 study found that even 5-second ads could build awareness, but it took :15s or :30s to drive brand favorability and purchase intent.

A lot changed in Magna's more recent study, just published in the last few weeks. This time, they found :06s and :15s to be similar in their ability to drive search intent, brand preference, and purchase intent. Why would that be?

One reason may be the platforms on which they ran the test: Snapchat, a video aggregator (i.e., YouTube) and a Full Episode Player (FEP, perhaps a streaming app like Hulu). Audiences are already accustomed to short ad lengths in these environments, and there were no :30s tested for comparison.

The Snapchat part of the test was interesting because more people watched :15s all the way through, but they were all placed mid-roll, about ten seconds into the content, so perhaps viewers were really staying for the content, which might explain the other finding that these ads were slightly less convincing.

YouTube was different. Viewers didn’t like the :15s, which were skippable after six seconds, but keep in mind that all of these were pre-roll ads, meaning that they were a barrier to the selected content. On the FEPs, :15s were better-received, but may also have been more expected during a 20-minute TV program.

None of these findings should surprise us, especially when you consider:
  • Linear TV wasn’t part of the test and neither were 30-second ads. It would be instructive to have these points of comparison.
  • The three viewing platforms in the test give individualized watching experiences, which may improve the ability of short copy to get across its messages, and also lead to less patience for longer ads.
  • We don’t know the quality of the ad creative shown. There were four brands included (Clinique, Mini, Lego and a “major CPG brand”) but we can only assume their ads were adequately memorable and persuasive.

On that last point, not only do we not know the quality of the ad creative, we don't know if it took full advantage of the format. Traditionally, :15s were (mostly) just shorter versions of :30s, both seen on linear TV. Newer formats, like a six-second pre-roll on YouTube, are seen by an individual person watching a very small screen. That calls for a different creative approach, and opens creative avenues instead of closing them.

It's always nice to have more time to get your message across, and :30s will continue to run on TV. But newer formats may prove to be a useful piece of your overall plan.

15 January 2021

Artificial Intelligence vs. Genuine Creativity

Will Artificial Intelligence replace Creativity?

No, but it might help it, in a roundabout way.

What is AI? What can it do?

Nearly all AI work today is based on successes in machine learning. Think of machine learning as having enough data and enough processing power to think through analyses that would take humans too long to do. Imagine a mountainous, time-consuming task that needed to be done, however slowly.

Here’s one example. For half a century, scientists have been mapping the three-dimensional shapes of proteins that are responsible for diseases like cancer and Covid-19. They refer to this mapping as “unfolding,” and doing it for just one protein takes a long time and a lot of money. Up to now they’ve “unfolded” only a fraction of the 200 million known proteins. The work done so far was recently fed to an AI program called “AlphaFold” which used it to do decades of work all at once. The results have been published online for review by the scientific community.

Did AI cure any diseases? No, but it advanced the work of scientists trying to do so.

Can AI advance the work of creativity?

How to approach AI

Some AI experts will tell you to approach AI with three questions in mind:

1) Is the task genuinely data driven?
2) Do you have the data needed?
3) Do you need the scale that automation provides?

On that last question: If you have a decision that needs to be made more than once per minute, then yes, you need the scale; if you have a decision that needs to be made only once per year, then probably not.

Does creativity answer “yes” to all three questions? What kind of creativity are we talking about? A painting, a sculpture, a novel? An advertisement? Let’s focus on advertising for the moment.

Advertising, Big Data and AI

We can’t say the task of creating ads is “genuinely data driven.” Sure, advertising ideas for a particular client or project may entail data or feature a data point, but even that isn’t a matter of computation. Nor is the task so routine that we create ads at a rate of more than one-per-minute. (OK, it feels that way sometimes.) Variations on an ad, however, might drive that kind of scale. Personalization of ads, for example, might be accomplished with AI that considers not only the recipient’s name but their past purchase history and other data. That’s already happening in most online marketplaces, and don’t forget that direct mail is personalized. But these are all variations on an ad created by humans.
It's a protein,
not a creative brief

There have been attempts to create at least one kind of advertising with AI: movie trailers. The first experiment was back in 2016: someone wrote a program, based on consumer reactions to movie trailers, that could lift scenes from a movie and sequence them in 30 seconds that would effectively convince people to see the movie. Judge the results for yourself and see here a more recent experiment from 2018. More recently, Netflix invested in technology to automate trailers for their content, while adding personalization for its subscribers, which makes sense for an individualized setting like your Netflix account.

Setting aside the irony that movie trailers are already quite formulaic, we see that AI made an ad. But does anyone really think that the studio marketing head won’t ask the machine for revisions? What about the movie itself? Could AI create a full-length, cinematic feature?
This may depend on one’s world view.

Keep AI in perspective

AI can certainly enable a human being to see new possibilities. For example, large amounts of data may help us predict future changes in consumer behavior. Knowing these possibilities may lead to a new insight on how to position a product or service. There’s great value in AI when it comes to aiding our thinking process. It can give us insight that inspires creativity. But that creativity is human, not artificial.

At an AI conference, a very intelligent professor of computer science said, “There’s no aspect of human cognition that can’t be modeled on a machine.” At the next break, I sought him out to learn more. He explained his world view that humanity – the human mind – is essentially physical, part of the physical world, and therefore can be modeled. Yes, he explained, machines will gain the ability to make cinematic features when AI develops enough to mimic every function of the brain. I asked, does it follow that humans are essentially machines? Incredibly, he said, “Yes, that’s a fair statement of how I see it.”

I see it differently. Creativity takes judgment, and human judgment comes from each person’s uniqueness, and their interaction with other people’s uniquenesses, to create something with passion and imagination. Perhaps, like me, you believe that we are more than machines. We have a spirit, a soul if you will, that animates us and gives us the ability to create sculptures, novels, choreography, and advertising. No machine can ever replicate that.