This is the second in a series on demystifying the best practices in media today. Read the first on audience suppression.
Frequency management, which first emerged when advertisers tried to calculate reach and media impact of their traditional TV buys, is tricky even for the most sophisticated marketers.
And it’s gotten trickier with the advent of digital. Hit a consumer too few times and it won’t make an impact. Deluge them with ads and marketers waste money and annoy consumers.
Even large marketers who invest time and effort into frequency management run into problems, like showing the same ad to the same consumer too many times, which reduces reach and creates waste.
Programmatic can help agencies and traders manage frequency because it lets them consolidate a good chunk (but not all) of digital ad spend in one place.
But it doesn’t solve all their problems. Optimizing frequency across channels, not just within them, is the latest hurdle marketers are taking on.
Here’s a step-by-step guide for those who want to get up and running.
Crawling: Learning the basics of frequency management
Anyone thinking about frequency capping should start by understanding a couple of basic concepts.
First, when marketers map ad frequency against cost per acquisition (CPA) or click-through rate (CTR), they usually see a curve that peaks, then drops off. Ad exposures after the drop off have less and less of an effect. These graphs can determine optimal ad frequency for future campaigns.
Second, a good rule of thumb for beginners is to set lower frequencies for awareness campaigns and higher ones for retargeting campaigns. Pallavi Yerramilli, senior trading specialist at The Trade Desk, for example, often recommends that clients start with a frequency cap of once every four hours for a retargeting campaign, and once per day for a prospecting campaign.
Walking: Factor in time to conversion and dynamic CPMs
Brands who are crawling should start experimenting to optimize frequency.
“They should set up a testing methodology, cadence and strategy to determine a given frequency,” said Brooks Perry, manager of digital marketing at independent agency RPA.
Brands make the mistake of focusing on capping frequency where conversions peak – but they also need to consider time to conversion.
It might take seven days for someone to book a hotel after seeing an ad, Yerramilli said. If the brand sets a frequency cap of two times a day, showing a total of 14 ads over a week, it may look like the optimal frequency is 14. In reality, the brand might need to show just five ads over a period of seven days to drive the same results.
“The data gets muddled with the time it takes to convert,” Yerramilli warned. Brands need to figure out what frequencies make sense against the average time to convert to avoid setting the wrong cap.
At Essence, the agency looks at how different ad frequencies perform over a 14-day conversion window. To simplify matters, it will analyze by low, medium and high frequencies, said Gila Wilensky, VP of media activation.
MediaMath Chief Product Officer Jacob Ross recommends introducing dynamic CPMs. Instead of paying the same price for a user who is less and less likely to convert, buyers can pay, for instance, a $5 CPM for the first impression and a $2 CPM for the fifth impression, Ross explained. Lowering prices for a later impression is one way to make higher frequencies more affordable.
Running: Managing frequencies across different channels
“Running is where you get into true cross-screen measurement, and you start connecting different platforms and channels,” said RPA’s Perry. For example, “once someone has seen something twice on Facebook on mobile, we may want to stop targeting them and just focus on search, because we know they will primed to go to Google and search for the brand.”
Often, that discipline entails changing frequencies for different channels.
“You should not apply a blanket frequency for display, video, social,” Essence’s Wilensky said. “There are different breakthrough points for different channels because there are different levels of noise.”
But that’s difficult because the digital ecosystem is made up of different platforms and walled gardens that don’t allow tags needed for frequency tracking and capping.
So brands that want to manage frequency must consider which platforms to use.
“You need to consolidate your platforms, so you don’t have five tunnel vision views,” said Mike Venables, executive director of biddable media and ecommerce at Hearts & Science.
Brands can take steps to consolidate by using consistent technology – like the same ad server – across their organization. They can also centralize their technology within a single walled garden: Google.
“If you are a brand that is okay with investing heavily in the Google tech stack, and you use them as your DSP, run paid search and social through YouTube, and use Adometry for multitouch attribution, you are probably going to have the best go at frequency management,” Perry said. “But you are giving up a lot of power and data to Google.”
Brands that don’t want to rely on Google, and that can’t consolidate platforms, must optimize frequency through other means.
Some of RPA’s clients use multitouch attribution tools to optimize frequency across channels, while Essence relies on experiments to manage frequency across channels.
“Walled gardens make it more difficult, but you can use proxies or run experiments to make assumptions,” Wilensky said.
For example, Essence might look at the entire customer journey, which involves five touchpoints, and then see what happens when it increases the average frequency for a single channel from two to three.
What are common mistakes people make?
The biggest mistake is ignoring frequency management altogether.
“Not frequency managing is the most harm you can do to your brand without knowing it,” Wilensky said.
Properly managing frequency involves high-level decisions about technology, so brands need to pre-plan.
“If people manage frequency within a platform, they have the potential to mismanage frequency because they are not looking at customers holistically,” Venables warned.
They also need to make sure their technology is properly stitched together. Cross-device tracking helps avoid duplication across desktop and mobile, and they need to consolidate platforms, use multi-touch attribution or run experiments to understand frequency across display, video, search and social.
How do I measure the ROI?
Some brands look at optimizing frequency to minimize waste.
“The KPI ends up being the amount of dollars you can shift to be more intelligently invested, because you were already hitting a threshold and running into some level of media waste,” Venables said. Those shifted dollars can increase media impact elsewhere.
For others, frequency can help optimize a brand’s immediate KPIs, like CPA or CTR. “You can look at how CPA changes for every incremental impression for users, and see if there is a drop-off,” Yerramilli said. Setting a frequency cap at that impression will then allow brands to remove users who are less and less likely to convert.
What technology do I need?
Managing frequency requires a ton of technology. Agencies had different opinions on which tool they found most important.
“Having one ad server is the first ingredient you need,” Wilensky said. From there, marketers should add in a DMP and DSP.
“Multitouch attribution, or a media-mix model, is one of the more important pieces of technology,” RPA’s Perry said, because the tool enables frequency management across channels.
At Hearts & Science, “the most important thing is plugging into an identity graph,” Venables said. “You need to have a connected ID which ends up being a holistic view of the customer.” Having a people-based, not cookie-based, identity graph allows marketers to avoid overserving ads to customers because they have cleared their cookies.
DSPs offer capabilities around frequency management: Machine learning can help automate frequency adjustments, while a robust reporting suite can give traders and marketers a more hands-on look at optimal frequencies. MediaMath emphasized the importance of the former, and Trade Desk the latter.
Do I need to hire anyone with special training?
“For frequency testing, you really need a mad scientist in your organization who loves to do experiments,” Essence’s Wilensky said. With a caveat: “They have to be people who are comfortable with the success being an incremental gain, like going from zero to two versus zero to 60.”
RPA’s Perry finds that people can be trained for most aspects of managing frequency. But the best results come from building a small team where someone with a detective-like mindset and deep knowledge in statistics is paired with a “big-picture” thinker.
On the front lines, programmatic traders often know people who understand frequency the best. “People who understand how programmatic media buying works are the people who are best positioned to manage frequency universally,” Venables said.
What’s the future of frequency management?
Agencies have a dream for a future – and a reality to deal with now.
The dream involves walled gardens being broken down, being able to stitch together TV data with everything else, and a consolidated ad server. “If you had those three elements, you would have the easiest time managing ad frequency,” Wilensky said.
But the reality means navigating across walled gardens and finding ways to understand optimal frequency through experiments and analysis – not direct measurement.
“Right now it’s not possible to do [frequency management] across any of the walled gardens,” RPA’s Perry said. “Even with a DMP/DSP bundled solution, it’s very difficult because Facebook and Google aren’t willing to share their data or let you tag their media. Obviously, marketers would love to do that.”
In the future, brands will have to weigh the cost and risk of consolidating platforms against a desire for independence.
“If you put all your media in one platform, even though you are limiting yourself on inventory, you can control things much better,” Wilensky said. “For the digital marketer who is [buying] across multiple touchpoints and multiple media environments, it’s mo’ money, mo’ problems.”