As we begin to explore the world of cookieless digital advertising, marketers will likely be focusing much of their attention on walled gardens that have robust amounts of users and, therefore, valuable first-party data. However, even walled gardens come with their issues that will have to be navigated through in order to ensure business goals are achieved and tangible to the financial guardians of brands.
Most marketing landscape analysts predict that walled gardens (in particular Google) will be the safest place to conduct audience targeted buys in 2022. Even while Google’s DSP allows marketers to buy a lot of inventory, it is currently more limited in audio, connected TV and DOOH inventory. These are channels where context is probably more important than the precision of the audience and where there is likely going to be a need to diversify to other advertising platforms to achieve a successful omni-channel strategy.
Facebook does have robust behavioral data from signed-in users; however, iOS 14.5 makes it a lot more challenging to perform audience-based buys and to attribute conversions. Some of our early campaigns showed a 15x increase in CPA within the platform, but nearly no impact on actual sales. This means that conversion data on the Facebook platform was (and is) solely directional for most advertisers. While good for the business, this might be more challenging for marketers who are being asked to prove that their marketing is “working.”
Amazon certainly has robust shopping data on its users, but ad space against those segments are only available on amazon.com and a couple of other sites they own and operate. As a result, the scale is significantly more limited than what Google offers.
The big ad tech players, and thus some agencies, will likely advise brands to ‘trust the algorithm’ even more than they have in the past, as Google, Facebook and Amazon don’t give specialists a lot of control over or insights about many aspects of their buying decisions. Facebook in particular makes it challenging to control frequency, and DV360’s lookalike modeling is very opaque. Against a lack of accurate measurement across each walled garden, brands and their agencies need to develop more holistic, advanced measurement frameworks.
While scale is impacted slightly outside of Google Chrome and Android apps, there are still ample opportunities to bid for inventory in these environments. However, with fewer buying platforms to conduct audience-based buys and fewer impressions to scale against, CPMs will likely increase, in particular on video. This might put pressure on agencies to ‘keep the costs down’, which in turn may increase traffic from bots and fraudulent inventory. Brands need to expect an increase in CPMs while not incentivizing a decrease in inventory quality.
The above can be summarized in a SWOT analysis:
Strengths – Google, Facebook, Amazon and Apple each have huge 1st party data sets. And not just in volume of users, they have robust metadata around each profile as well, from account information, purchase history and behavior. Even if ID-based soutions grow in count, it’s possible we may not be able to append significant amounts of secondary data to each profile to be scalable for marketers.
Weaknesses – There will undoubtedly be adjustments needed in terms of attribution and measurement. Even today, if you were to believe the metrics from each platform, nearly all of your automated marketing channels would have +ROI for the same purchase. Paid search, Facebook conversion ads and programmatic retargeting can’t all have a CPA of $10. They can’t produce 10,000 sales when you only sold 3,000 products. This is because each is taking credit for any time a user touches their ad. Attribution has been among the greatest challenges of digital advertising. Because the walled gardens don’t share a common profile of a user, multi-touch attribution can be at times disjointed and inconsistent. The death of the 3rd party cookie will exacerbate this divide; while we can’t say multi-touch attribution will never be achievable, it is safe to say the methods of achieving this measurement will have to change.
Opportunities – Lean into first-party and second party data in walled garden platforms, and rely less on retargeting. This allows for stronger prospecting and less reliance on audiences that were likely to “convert” anyway and, therefore, inflate marketing metrics.
Threats – Because many marketers and brands will be choosing to lean into walled gardens to circumvent the challenges of third party cookies, it is likely that there will be an increase in costs to advertise on these platforms. Therefore, budgets will need to increase to achieve the same scale previously achieved.
So what does this mean?
At present, our suggestion is to lean into walled gardens for precise audience targeting, but to begin measuring success of your advertising program at a higher level. Some examples of this include matched market tests, media mix modeling, control vs. exposed methodologies. While this will make it more challenging to know which 50% of your marketing spend is effective, it’s the best solution given the reduction of transparency in algorithmic data and therefore less understanding of success from a conversion data standpoint. But, this will also force marketers to start looking at the data as a whole and get away from optimizing towards last-click and last-touch metrics which have provided misleading signals for years.
Regarding adjustments needed to measurement, advertising campaigns need to set-up to achieve business goals rather than just media metric KPIs. To achieve this, individual channels and tactics will need to identify leading indicators, perhaps higher in the funnel, to optimize toward. Engagement rates, reach, completion rate, and measures of media effectiveness like CPM/CPC should become more of a focus rather than CPAs. Fortunately, programmatic platforms are poised to bring marketers success with this measurement approach, and will be able to optimize against important cost and engagement metrics rather than directional conversion-level data.