Dynamic creativity, viewability, and the use of ad tech for brand building are topics at the forefront of the current conversation. Quantcast’s Konrad Feldman sat down with WARC’s David Tiltman to talk about the state of the industry.

What do you make of the criticism of the ad tech ecosystem around issues such as viewability, fraud, or just generally the amount of money vanishing in the supply chain?

There's a lot that's fair in criticism. Over the last few years, tens of billions of pounds have been spent on digital marketing and technology-related solutions. There's been an enormous investment, and yet, if you look at the biggest brands, they're not growing. So there's a disconnect. Something's not working as well as everyone might like, and I think there's a lot of pieces that you can pick out.

Our ambition should be what we can do now with technology, the ability to reach consumers and to create relevance. Most of the industry has not been that sophisticated with the ‘plumbing’ that has been created and I think the vision has to be to make all the advertising that we're exposed to much more relevant and much more useful. We can have less of it and still generate the same returns overall for the industry.

Viewability and fraud are a part of that, but the fundamental challenge is how to connect with the subset of consumers that are going to find a message relevant and useful, and to do that systematically. I think technology comes along and people focus on applying the technology to make some process more efficient. And they haven't really thought about how technology could fundamentally change the way we do things to make it more effective.

Computers are much better at making low-level, tactical, high-frequency decisions. People are much better at connecting the dots between different disciplines, so if you think about creativity, machines will take low-level technical work and free up people to do those sorts of things.

We're only at the beginning of that journey. The drive for efficiency can only take us so far. We've got to focus on effectiveness.

What are your views on dynamic creativity? Is it being done well yet?

There's good stuff being done. It's most commonly used in the retargeting space. I think, more broadly, it has a lot of opportunity. You see people varying creative based on where their office is, for example. I think there's a little way to go to get the full benefit of what's possible, because the creative offering tools need to make it easier.

Today, if you need 20,000 different creative treatments, that's very expensive. It's going to take time until broad-use creative offering tools are available.

There's also still work to be done to help the creative teams understand how they can use these sorts of technology – how is the machine going to inspire the human learning?

One of the classic areas that we work on is developing insights for creative. And when you actually look at the set of customers that are responding to a product, you often find that's not the target group. It's actually a different group. For example, a pharma company had a product designed for older men. That's the target persona. But when you look at the people engaging with their content, you could see that a large minority of the audience was younger women. They're caregivers. So that gave them new thoughts – what's the content required for those people? What are the brochures that we send them when they request more information?

I think there's a lot of opportunity for how insights can drive creativity there. And over time, I expect to have the opportunity to connect that directly to elements of the creative. But I still think there’s work to be done to make that really stand out.

A lot of the things you’re talking about are what might be called activation or lower-funnel marketing.

That’s certainly where we started. But ‘brand’ advertising has been a rapidly growing area for us.

And in that world, it's about managing reach and frequency consistently against the audience.

Except, people talk about an average frequency. When you actually look at the distribution of frequency, an average is a very misleading number. Most people saw one ad. Most people weren't in the effective range, wasting a lot of budget. So we’ve had a lot of success with systems managing frequency.

Is programmatic frequency management something that is widely being used?

The predictive modelling that you have to do upfront is quite hard. You need the right number of opportunities in the right environments to show people more ads, to get to the frequency you want.  

If there’s a person I'd like to reach, and I've got the opportunity to reach them now, I can reach them now with one ad. Do I feel confident that I'm going to have enough opportunities spaced far enough apart that I'm going to deliver the optimal frequency to this person? You need pretty sophisticated frequency models for that. You have to know how often you'd like to see people in the appropriate environments.

Are there any trends in the kinds of data clients are bringing in to campaigns now?

The ability to de-average your customers when you have customer data, and can say you value customers differently – rather than saying your cost per acquisition is ten pounds, you de-average and say it's worth 50 pounds to prime customers and five pounds, or nothing, to other types of customers.

Once you can do that, then you can align the technology and the goals you're looking to drive. Machine learning, fundamentally, is a goal. There's lots of different types of algorithms, but they're all goal-driven. They're trying to maximize an output or minimize an error. And AI is far less tolerant of ambiguous goals than any human you've ever met. So the ability to provide it with a precise goal of what you really value is actually very powerful.

That is a really exciting use of some of the customer information systems people have put together – the ability to say you're not going to optimize and consider everybody the same.

What about location data?

We're working with partners that specialize in that space. You are able to look at the impact that advertising exposure had on driving differentiated outcomes in terms of where people go. That's a really interesting use of the data. It's a nice way of measuring the impact that your advertising has had.

Ultimately, TV or radio advertising might work on this model if it can tell who is watching or who's listening. How quickly do you think that sort of transition's going to take place?

From a technology perspective, I think it could happen fairly quickly. But it's not just about the technology, it's about how the business operates, how the business is funded, how people want it to operate, and what's more convenient for the different parties in the ecosystem. It takes time for those things to work through. And it probably occurs at different speeds in different parts of the market.

We're definitely seeing the TV space grow. We're going to see more opportunities for addressability in television. I think it will make television more compelling and may reduce the ad load. In the US, the ad load is very high on traditional television.

If you could show different things to different people, you'd actually create more relevance, markets are going to pay more for it, and the publisher networks will be able to generate the revenue they need to produce the content consumers want with fewer ads. So everyone, really, can win. It's not a zero-sum game.

You’ve talked a lot about relevance. There is a counter-argument that some of the value in advertising is in the waste. So for example, you may not have any children, so baby care advertising is not relevant to you, but you will have children one day, at which point brands that have built their brands across a number of years will benefit.

That's not really waste. That's just not being able to measure it, in terms of the metrics that are available, in the required timeframe. If you decide that actually you want to reach people and create some awareness or some level of frequency way ahead of them doing it, then you should deliberately do that.

But I think that people should be deliberate in their experimentation. If you've got a defined experiment, you have a hypothesis and you run and test it – then you're actually learning, as opposed to ‘we did this and some of it worked and some of it didn't, we don't really know what’.

If your goal is to get a new set of people, maybe they're not parents yet, maybe they haven't even thought about that, but you want to create awareness with them, do it deliberately and measure it. Over time, we'll connect the dots. We've got trusted proxies in terms of awareness and favorability and things like that, and they'll be used for a long time. People know they work.

Today there is a disconnect between planning and buying. It's like throwing things against the wall and hoping something sticks. We can get much more systematic. And the companies that I believe are going to be successful are companies that are systematic in their experimentation, that are always trying to learn.