As any marketer who’s run streaming television campaigns knows, IP addresses are still widely used when trying to identify and target individuals with relevant ads in this still relatively new medium.
In fact, this information is so crucial to many streaming campaigns that content publishers leverage it to justify charging premium prices.
But there’s a major catch. “The IP address alone is typically worthless,” notes Aaron Lange, senior vice president technology at performance television agency Marketing Architects. “A campaign needs to be able to map the IP address to a consumer profile built with a combination of first-party and third-party data, both of which come with additional costs and room for error.”
The costs and effort involved can be worthwhile when it works — when an advertiser’s ideal customers are hit with a fantastic ad reminding them how great that product or service is, and they immediately go out and buy," he notes. "But all too often, IP-based targeting falls far short of that goal, resulting in wasted ad spend."
We asked Lange to elaborate on the streaming targeting scenario, and share some advice.
Specifically how does IP targeting go wrong?
Lange: One big culprit is bad third-party data. I’ll use my own household and viewing experiences to illustrate. While watching Hulu recently, I noticed that over a three-week span, around 30% of the ads I saw for at least three different brands were 100% Spanish-language. I don’t speak Spanish, nor does anyone in my household. In a few cases, I couldn’t even figure out what type of product was being advertised.
Now, I suppose it’s possible that these brands are intentionally choosing not to target. But it’s also possible that I did something like accidentally visit a Spanish-language site at some point and was grouped into a segment based on this one data point. These brands then received that misleading data and paid a premium to target me with ads in a language I don’t speak.
It’s hardly news that third-party data accuracy is a challenge, and that it’s growing as privacy regulations limit the availability of this data. Yet, 80% of marketers still say they are "very" or "moderately" reliant on third-party tracking. Paying extra to target only to get it wrong with bad data is painful.
What else contributes to poor targeting?
Lange: Inability to determine who’s watching, and lack of knowledge about other members of the household.
Across the U.S., the average number of people using a single streaming account is five. As it happens, I live in a household of five. All of us are in the workforce and buy products and services, and all of us view content using the same streaming account. Yet only one of us is tied directly to the account — and the IP address.
So problem number one is that even if the person tied to the account is accurately represented by third-party data and is the perfect target for a particular campaign or product, [he/she/they] may not be among those watching when your commercial shows up.
Just as frustrating, targeting by IP address means that you aren’t hitting some households in which one or more of the non-account holders are actually members of your ideal target audience. You’re not delivering impressions to potentially large numbers of key prospects.
How can advertisers improve targeting accuracy and return on their streaming campaign investments?
Lange: One important step is asking the right questions to vet prospective partners.
Ask your data provider or DSP how frequently third-party data is refreshed. Then ask how broadly they apply that new data. Does it cover all old data, or just some of it?
How many unique sources of data are used? When evaluating partners, look for those pulling data from multiple sources, for better accuracy.
How robust are the identity graphs? The data for ID graphs should also be refreshed regularly. And advertisers should vet the quality of the additional data points used to create profiles, from email addresses to behavioral data.
Alternately, advertisers can try running campaigns based on content or other targeting options beyond IP addresses. When well-planned and executed, those methods can often yield similar or equal results, minus the premium costs.
And as we know, with privacy regulations expanding, it looks like the ability to use IP addresses for tracking will at some point actually cease to be an option. So being heavily reliant on using that targeting method is not only limited and inefficient; it’s risky. Better to start gaining experience now with alternatives.