Connected television (CTV) is poised to become one of the biggest performance advertising channels, but is there an effective attribution model to support it.
The lack of transparency into CTV attribution, with some models lumping CTV in with other inventory, has left performance advertisers wondering how much of the sales can truly be attributed to CTV ads alone.
Similar to other advertising units, CTV ads should
only take credit where it’s due. To do that, the playbook for linear TV won’t work, according to MNTN Vice President of Engineering Richard Girges.
Girges, in a candid discussion with Data & Programmatic Insider (D&PI), outlines what advertisers should look for to ensure they get an accurate picture of what their CTV campaigns are responsible for, such as view-through and cross-device attribution, and how they can give CTV too much credit. How always-on incremental measurement that isolates key variables can increase transparency, and linear TV attribution methods that fall flat on CTV.
MNTN engineers are working on an artificial intelligence (AI) assembly tool. It leverages generative AI and allows the user to create a “living room compliant” video ad using generative audio voiceovers and images based on the brand’s assets such as design principles. The tool can be used to speed the process of video editing for a television ad.
About 60% of MNTN’s clients have never advertised on TV before, so the biggest barrier to entry has been getting a high-quality TV ad made.
In the next few months, MNTN will release performance enhancements to its technology, along with new analytics projects being integrated with Google Analytics.
Data & Programmatic Insider: What are the myths in CTV attribution?
Richard Girges: You can’t drive accurate attribution. Many of the attribution solutions that most performance marketers rely on in CTV today are outdated. First-touch attribution and a superset of view-through attribution are the most common. Many models are based on guesswork or inherently inaccurate.
There are myths around transparency and myths around the halo-effect. Some believe CTV cannot drive a halo-effect, and doesn’t have much of a place in the multichannel marketing mix. They treat is as a short-term gain. Their attribution models look for short-term spikes in web traffic or purchases.
It goes directly against the nature of CTV. If I view an ad and immediately buy the product, it’s a rarity, but usually I will view the ad a couple of times or view it in different channels before I’m influenced. Many think CTV media is a short-term strategy. That’s not the way TV works.
D&PI: How does CTV work?
Girges: Most people that watch TV and see an ad remember the ad for weeks, even a month. The attribution model must take that into account. MNTN built an attribution model called Verified Visits. It works when a user sees a MNTN connected TV ad and visits the brand’s website. The technology confirms the person’s visit to the site wasn’t prompted by any other media. And then the visit gets recorded as a verified visit.
A person may not visit or buy a product for a long while, so you want to make sure you can capture the lag time and residual effect.
D&PI: How does the MNTN CTV model work?
Girges: MNTN receives events that tell us how far into the ad the viewer got. When they make it into the fourth quartile, we count it as a viewed ad. If it’s not a viewed ad, we don’t attribute it later on when the user visits the advertiser's website within a window of time defined by the brand. The technology then validates the source that drove the visit. If the goal is to drive traffic to the website, then the attribution ends there. Most brands want to prove return on ad spend, though, which means we also tie conversions and revenue back to the campaign.
D&PI: What are the murky waters in attribution?
Girges: Dated attribution models. We live in a programmatic world. CTV now runs on programmatic technology, but the attribution models are more in line with old-school liner TV. They accept inaccuracy. We don’t need to accept that. We also need to give clients control.
Attribution models should be configurable. Advertisers should be able to decide conversion windows, visit windows, and black list and white list. Some clients prefer that the performance marketing channel they use doesn’t get discounted when a user visits through a paid search channel. People sometimes don’t type in the actual URL into the browser. So, you end up in questionable situations like paid search may get credit when CTV actually drive the visit.