Amidst rapidly changing technological changes, Digiday sat down with our very own Director Analytics & Technology, Mike Ruff, to inform their industry publication, Media Buying Briefing. Read the article in full here.
By Antoinette Siu • October 23, 2023
No agency can escape measurement challenges these days — but many are looking to tighten up their data standards practices as generative artificial intelligence and privacy changes exert more influence on them.
Not only does standardization become paramount when ensuring accurate measurement, attribution and optimized campaign results, but without good data in place, AI tools would be ineffective.
“The goal of an agency is to try to pull data from disparate sources into one place to make it usable and actionable by a variety of end users,” said Lee Beale, managing partner at Crossmedia. “In all of these kinds of tool sets, it is beyond mandatory that your data is standardized, reliable, properly structured, properly architected.”
Having the right data practices can also improve campaign metrics. Marketing data platform Claravine last week surveyed 140 U.S. advertisers and showed that their data standards resulted in ROI increases of 30% on average. The report found that changing privacy rules and the growth of generative AI are driving agencies to look deeper at data standards and the impact on brand safety, security risks and creative production.
“[What] standards also do is force teams to take a close look at the data they have, what it is and how it can be used,” said Mark Sturino, vp of data and analytics for Good Apple.
Privacy landscape
In the long run, building out standards can cut down on staffing costs as agencies work with clients and adapt to privacy changes. Mike Ruff, director of analytics and technology at media agency Media+, said the agency recently created a new data collection methodology that can reduce costs when onboarding new clients.
“It’s true that each client is unique, but data standardization doesn’t mean data sameness,” Ruff said. “It means that we collect different data in the same way, so there is flexibility to collect very different data, but rigidity in the data collection process … With all of the relatively new privacy regulations, it’s more important than ever to have one compliant practice for data collection.”
Good Apple said it is also building proprietary AI tools to monitor brand safety, QA data feeds and ad operations. The requirements to meet GDPR standards and other subsequent privacy legislation have led to more documentation across the company’s data operations and resulted in better reporting and analysis, noted Sturino.
Building taxonomy and data governance
Standardization starts with data governance to make sure marketing data is “usable,” explained Leanne Smith, svp of business insights and data analytics of CMI Media Group, which specializes in the health field — an area with particularly stringent privacy regulations. For the last two years, CMI and Compas have been reviewing data attributes and nomenclature in order to align all the terminology with client needs. The agency recently rolled out 2024 standards that will classify media and creative in a standardized way to make sure campaign attributes are captured.
“Given the old saying of ‘garbage in, garbage out,’ marketers cannot afford to not adopt standards as more marketing decisions are being made based on the science vs. the art,” Smith said.
Nik Hengel, vp of platform at Novus Media’s data unit Localytics, agreed that agencies need to be tracking at the same level of detail they want to report against. “When we onboard, and even when we pitch, we have dedicated workstreams across planning, analytics and activation teams to make sure we are capturing all the necessary levels of detail, and we understand how we can harmonize within and amongst all channels.”
Implementing across data operations
At Omnicom’s Annalect, chief experience officer Clarissa Season has been developing a taxonomy builder tool for the last five years that is now standard practice for all clients. The tool is a core part of asset creation, especially in managing the AI-produced creative that has taxonomy automatically added. These could include client codes, product codes, audience, placement and name or size of the creative elements – and the fields all need to be standardized from a list of accepted answers, Season explained.
For example, for a buy via The Trade Desk, Season said, “we can use our taxonomy tool to help populate those in the correct taxonomy – and it will create variations of them. So you know if you’re buying 500 different line item buys, you don’t have to create 500 different lines.”
This makes the data come out with the right taxonomy and speeds up the process that someone would otherwise do manually. Season mentioned there is a larger data operations practice across the organization, which employs hundreds of people to run audits, such as reviewing automated QA tools for discrepancies.
Without these systems in place, agencies simply lose accounts – which is often what Season hears from partners about why they are leaving their current agency. To have effective measurement, the data needs to be clean and optimized for media buying, Season contended.
“If I spend a million dollars for a client, I have to be able to show them that they got a million dollars in value. … If I can’t do that, I’ve failed the very basics that my clients expect,” Season added.
AI is useless without good data
With greater use of gen AI across marketing programs, agencies also need to step up their data practices in order to take advantage of the tools. As CMI’s Smith mentioned, creating the standards allows the agency to use an AI bot to automate “rote tasks that are necessary when trafficking media. This automation allows our teams to spend more time on strategic asks from our clients, providing them more value.”
While measurement has always been crucial to agencies, the challenge now is “managing the explosion of data” given the AI integrations, added Pradeep Chelpati, global chief technology officer at Code and Theory. Chelpati said this makes it imperative to revise and enrich data standards that everybody sticks to across data analytics and AI teams.
For Annalect, Season also pointed to the new Omni Assist virtual assistant built on gen AI. The AI tells team members insights on best performing campaigns, audiences and other optimizations – but without the correct taxonomy and clean data and dashboard, “those insights you get from generative AI will not be accurate or clean either,” she explained. “They essentially will become unusable – it makes using generative AI almost pointless, because you can’t really assess where to optimize.”
Looking ahead, it will be vital to also work to standardize processes and taxonomy of an agency’s data vendors and partners, which will present its own set of challenges, Season pointed out. As agencies use their data along with their data vendors’ in a clean room environment, it will become even more necessary to establish wider industry standards and practices.
That, however, is still a work in progress.
