While CTV is rising up brands’ media plans across the board, digital native direct-to-consumer brands are showing a particular interest in the space. CTV provides the reach and premium content associated with TV, while also giving some of the detailed performance metrics those brands are accustomed to from the social platforms.
TVDataNow, launched at the start of last year, seeks to cater to these digital natives, helping performance marketers to measure outcomes in CTV environments. VAN spoke with John Hamilton, CEO and founder of TVDataNow, to hear why DTC brands are investing in the space, and where they’re finding value in CTV.
TVDataNow works primarily with digital-first advertisers, why are those companies investing in CTV?
There are a few reasons why digital marketers are looking to test connected TV right now. One is because they might have already maxed out their spend on Google and Facebook, not because they don’t have enough budget, but because they can’t spend more and get the same effectiveness. Or it might be because they’re a sophisticated marketer, and they know that ROI on their digital campaigns is going to fall in 18 months because of the changes we’ll see with third-party cookies and IDFA.
So we’re finding a lot of traction right now with digital-first marketers wanting to learn more about connected TV, on both the brand and agency side.
We work with those clients a few ways. Some just want to use our product, which might be our data, our dashboard, or the attention score. But we’re also acting as an agency, because a lot of these DTC companies don’t work with agencies for their digital spend, and they don’t have linear spend so they don’t work with a TV agency either. So they need someone to go and do the media buys for them, and we’ve been doing that. We don’t take the risk on the media, the brands sign the IOs, but we go and build the media plan and coordinate the buys.
What do digital native brands want out of CTV advertising – are they as focussed on performance as they are on the social platforms?
It’s still mostly about performance. Our clients judge performance on a cost-per acquisition (CPA) basis, or on some sort of conversion metric. But they don’t generally hold CTV to the same metrics as they would with Facebook and Google. So their Facebook and Google CPAs might be $25 while CTV might be $50, but they’re more likely to accept those higher costs for CTV.
That’s because digital has been much more finely optimised, which brings those CPAs down. Marketers are more confident on digital that the right creative is being shown to the right individual at the right time. But that’s based on trillions of data points, and we don’t have that data in CTV yet. So there’s no way we should expect CPAs should be as low on CTV as they are on social platforms.
But brands know CTV will become huge and they’ve got to figure it out, so that’s why they’re investing now.
What sorts of data does TVDataNow’s platform provide, and how do your clients use that data?
There are really three parts to what we do. Firstly we’ve got our dashboard, which basically provides three pieces of data. There are exposure metrics, so that’s things like impression, reach, frequency, completion percentage. Then it’s conversion metrics, so conversion rate, site visits, CPAs, and return on ad spend (ROAS). And finally there’s our attention score, which I’ll explain later. Those three things encompass about five to ten percent of all the data that our pixel captures.
The second component of our product is our technology and data, that enables us to capture about 100 pieces of data. And the amount of data we’re able to capture is only going to grow over the next 18 months. We look at that as a product in and of itself. So some of our clients will just buy that data for their own uses.
Then the third component is the attention score in and of itself, which people can buy by itself. It tracks when users pause, fast-forward, change volume, mute, or change the size of the ad on their screen, as well as data on where in the pod an ad appears, and other data like that. And all of those things factor in, to help us estimate the amount of attention an individual is paying to an ad.
For brands who just pay for access to those 100 data points you mentioned, what sorts of data can they access, and how do they use it?
I’ll give some examples. Let’s say I’m not looking at conversion data at all, I’m running a very basic reach campaign. And let’s say I see that on Tubi, one percent of people paused the ad, while on Hulu, 0.25 percent of people paused the ad. It could be a good thing that those one percent of viewers paused the ad. So I might want to segment that one percent and retarget them, which you can do with that more granular data.
Or let’s say that a conversion goes against a campaign which is running on Tubi, Pluto and Hulu, with the consumer last seeing the ad on Hulu. And let’s say that on Hulu muted the ad, on Tubi they paused the ad, and on Pluto they did nothing. Should Hulu get credit for that ad? Probably not since the ad was muted, and this granular data helps you recognise these things and interrogate how your campaigns are really performing.
So if an advertiser or agency has their own team of data scientists, they might want this raw data so they can come up with their own insights. It’s still early days for CTV, we don’t have best practices yet in terms of what works and what doesn’t work, in terms of creative too. And some brands want access to all of this data to start figuring out those best practices.
How well do you think brand CMOs understand the CTV space?
I think that they broadly and admittedly don’t know very much. In part that’s because if you’re a 45 year old CMO, you’re just not the target demographic for a company like Pluto right now, so you just don’t know about it. That’s changing a bit as companies like Pluto and Tubi have been bought by bigger media companies, but those are all relatively recent. So most don’t really know what’s going on, and they admit that.
That’s part of the reason why we’re acting as a media agency. CMOs often know they want to run CTV campaigns, but they don’t know which companies to talk to and how to run those campaigns.
But there is a high variance in knowledge. You’ll talk to some buyers who have been running CTV campaigns for three or four years now, and they actually know quite a bit about the space. But they’re the top five percent. And I liken it to my seven year old son who plays Little League Baseball. You’ll see some kids who can barely throw or catch a ball, and then you’ll have others who are playing at the level I was playing in fifth grade! You see that same kind of variance in CTV.
Do you think there’s much value in the long tail of CTV?
I think there’s value in the long tail, but not in the extreme long tail. In the US you’ve got Hulu, Tubi and Pluto, which are the three biggest CTV publishers. Behind those you have a bunch of smaller players which have around 10 million monthly unique users for advertising. But after the top 100-150 publishers, that’s where you start running into spam and fraud.
So we advocate to our clients that they work directly with 5-10 big publishers, and then work with Roku and their DSP to reach the other top 100 publishers. But you just shouldn’t go beyond that top 100, because that’s where you’ll run into problems.
What are the challenges with outcome measurement on CTV?
The biggest challenge is connecting the back end conversions to the exposure data. There are a couple ways you can do that today. One very simple way is to use IP addresses, but those don’t work very well for a number of reasons. The second way is to use an identity graph provider, which we do with LiveRamp.
But with identity graphs, there are challenges with accuracy and match rate. LiveRamp’s best known graph is deterministic, which means they’re fully confident in identifying users at different touchpoints. But their CTV graph is probabilistic, which means it’s less accurate. And there just isn’t as much data in CTV as there is in digital. So identity graphs don’t have as high match rates, they might be as low as 25 to 30 percent rather than 75 or 80 percent which you might see on digital.
We’re using our attention score as an interim solution. Our attention score doesn’t prove conversions, but it indicates how much attention an individual paid. So you can optimise against that. It doesn’t solve the core challenge for performance marketers around comparing campaigns on a CPA basis. But it works as a proxy while we wait for CTV identity graphs to get to where they need to be.