![Erez Levin](https://430281.fs1.hubspotusercontent-na1.net/hub/430281/hubfs/2024_03_Interview-34_Blog_Banner_Erez-Levin.jpg?width=2880&height=720&name=2024_03_Interview-34_Blog_Banner_Erez-Levin.jpg)
The Human Element Necessary for AI-Powered Attention
![Keith O'Brien](https://f.hubspotusercontent30.net/hub/430281/hubfs/Author_Keith_O-Brien.png?width=100&name=Author_Keith_O-Brien.png)
Keith O'Brien
Erez Levin spent thirteen years at Google, most recently serving as US Agency Senior Data Transformation Lead. He's now Principal at Emet Advisory, where he helps businesses in the advertising & marketing industry preparing for a rapidly changing environment.
He talked to Realeyes about the importance of experimentation, why an industry standard is unlikely, and why the human element won’t go away with AI adoption.
Why is attention important? And why is there so much visibility and interest in it in 2024?
I guess I can answer the first question in two parts. The obvious one is that an ad needs some level of attention to be effective. Now, the level of attention that's needed varies due to a lot of different factors, and we can't measure and activate off of these perfectly at scale. But buyers need to account for them [to inform] their buying and pricing decisions wherever they can.
Attention and quality measures really help assign better value to media. |
The other reason is [related to] the broader category of media quality and that not all impressions are created equally. We sort of commoditized media and the landscape using vanity metrics. In the process, we've seen a lot of waste and missed opportunities. Attention and quality measures really help assign better value to media. We're seeing more and more adoption, more interested buyers are proving the value of attention and the need for it, which is fantastic.
On the other hand, I'm not sure that we've gotten far enough, as an industry, to get the consensus that we need. I'm not too bullish on getting a standard or a specific definition. I don't think those are on the horizon, but even sort of the broad contours of what attention measurement is and some of the principles behind it are really important for us to get behind so that attention can become at least a best practice, even if it's not a standard.
What are your thoughts on media attention and creative attention? Are those things that should be viewed together or separately? And what's the best way to maximize both of them?
As the digital landscape sort of proliferated, we saw the media ecosystem and the ad ecosystem fragment into so many different placements. There are hundreds, at least, of different ad placements and ad experiences that people are exposed to, and so attention measurement or media quality measurement really helps understand the relative value of those different types of media exposures. Creative attention is about understanding what the right creative is that fits in each one of those different ad experiences. And both of these are really critical. I can give one obvious example: in-stream and audible video ads inventory is finite and scarce.
Now you do need the media attention measures to understand what those different ad experiences are like and price them accordingly. |
At the same time, there is an abundance of out-stream, muted ads all over the place. Now you do need the media attention measures to understand what those different ad experiences are like and price them accordingly. But most marketers in most campaigns need to decide because they might want to run a different creative for each of those different ad experiences. So obviously, both of them are really critical. And I do think it's sort of the Holy Grail when you can combine them. But even doing one or the other makes a huge difference in effectiveness.
What are some things that you know towards the back half of the year that people should be paying attention to, such as there's more conversation around this and more brands and agencies are looking to embrace attention?
I connect this to a broader sort of trend we're moving towards, which is quality of all types is going to be much more commonly used, especially by enterprise buyers. There are three pillars of quality: media quality, creative quality, and data or audience quality. All are part of this quality era that we're entering.
There are three pillars of quality: media, creative, and data or audience quality. |
Attention is going to play a really critical role in the first two, but I think it will also play a pretty important supporting role in the daily data and audience piece. We will see these concepts and the solutions spread where it becomes easier for buyers to measure it with these different measures of quality, much more seamlessly than they can today.
What’s the best advice for advertisers that are maybe a little behind in adopting attention to get acclimated or start catching up?
I have three pieces of advice because I've spoken to so many different marketers in different situations. The first one is to experiment – to really get buy-in on changing measurement and moving towards these attention-based solutions. The marketer really needs to get buy-in within their organization and aligned with their agencies and other partners. It's really important to run experiments and show that added value impacts the business in real incremental terms. The second one is to look for the low-hanging fruit. Experimentation is easy to say, and sometimes harder to get prioritized.
It's really important to run experiments and show that added value impacts the business in real incremental terms. |
For those buyers that want to do this, it’s going to take them some time due to other organizational priorities. There are other low-hanging-fruit opportunities that advertisers can use those same principles of attention or media replacement quality to at least stop overpaying for some of the low-quality media that they're currently buying based on those vanity metrics and shift their spend towards higher-quality media. It's really easy to just take a look and make sure you're not overpaying for those muted out-stream video ads that are [incorrectly labeled] as in-stream.
And then the last recommendation I have is don't let the pursuit of perfection hinder your progress. We won’t get a perfect standard. There is no such thing. I do think that it's too subjective what attention or quality means. But there's enough inefficiency in the market right now. Even some of those simple and imperfect applications of attention or quality. Whether you do it yourself or work with a vendor, it's going to give you a competitive edge over your peer set. As more buyers start to incorporate attention into their strategies, you don't want to be at the back of the pack. So, what you really want to do is just make sure you're staying ahead of some of your peer set.
I think that [human element] is really critical to maintain growth and make AI work really well for this industry. |
How do you think advances in AI are already impacting attention measurement and will continue to do so?
Machine learning right now already underpins so much of the attention measurement technologies on both the creative and media sides. Those technologies will get more powerful and faster. But ultimately, I believe the more important advancements to come are just the solutions that make it easier for buyers to measure, plan, and activate off of the attention and other quality measures.
I hope that we, as an industry, don't put too much faith into AI. There will be plenty of promising AI solutions for marketing and marketing effectiveness. Some of them will be a black box, and some of them will offer buyers more control over their data and with data science expertise. There will be some value in all of those.
But measuring the effectiveness and impact of marketing precisely is pretty much impossible to do in most cases. Marketers have to play a critical role in customizing and refining their uses of AI constantly. I always think of the George Box quote: "All models are wrong. Some are useful.” We need to remember that and not put too much weight into these models and too much trust in them to always deliver without constantly using our human insights – like everything that we know about the brands. I think that [human element] is really critical to maintain growth and make AI work really well for this industry.