Media agencies and creative agencies may have the same ultimate goal – to drive successful advertising campaigns – but their different priorities within that process sometimes leads to head-butting.
It’s easy to imagine, then, why replacing either team’s duties with an AI tool would seem like an easy way to break the tension, job security be damned.
But Alli, an AI-powered analytics tool offered by independent advertising agency PMG, suggests an alternative: making the information both teams need accessible in an easy-to-parse format, and answering the sorts of questions agency employees used to waste time emailing each other to solve.
“The thing that Alli does really, really well is getting all of our customers’ data in a single place for them to utilize,” Head of Technology Chris Alvares told AdExchanger.
As the second employee PMG hired – third if you count CEO George Popstefanov – Alvares has been instrumental in Alli’s development, from its conception in 2011 to its launch in 2019.
The platform was designed for PMG’s employees and its enterprise brand clients by focusing on use cases that matter most to them.
To that end, Alli boasts a level of customization that, according to Alvares, “would probably be too much for small to medium-size businesses.”
It gathers data from a number of integrable sources, including social ad networks like Meta and Google, DSPs like the Trade Desk and Bidtellect, CRM software like Salesforce and even other analytics platforms like Power BI and Tableau.
Powered by a suite of traditional machine learning models (such as anomaly detection and linear regressions) added in 2017 and LLMs added in 2023, Alli can also learn from sources outside its initial training data using a process known as retrieval augmentation generation (RAG) – although proprietary or sensitive data from each client is blocked off from this process.
Using all this data, Alli can then interpret questions from users (e.g., “What are my top-performing campaigns?”) and generate the results into an easy-to-read format.
“We basically made it a way for anybody who’s using the system to be able to go in and categorize their own data,” said Alvares.
Creative Insights
The data Alli collects covers more than just numbers. It also includes all the creative content housed on existing ad accounts, plus all the media data associated with it.
One of Alli’s latest tools, aptly named Creative Insights, can take this content and analyze it further so users know exactly how well it’s performing across multiple networks.
The practical upshot of this is that creative teams can run their own insights across multiple networks, calculating the top and bottom performers overall based on metrics and funnel placement.
With video ads, it can even track every time there’s a shot change, how frequently the brand logo appears and at exactly what point the user moves on.
Alli can also determine what’s featured in a given piece of content – obvious subjects like “woman” or “dress,” but also other descriptors like colors, patterns and other featured objects – and cross-reference it against performance data.
The Creative Analyzer then aligns these auto-generated tags across a four-quadrant graph to chart how well associated content performs online relative to how frequently they’re used. Frequent and well-performing tags appear on one corner of the chart; uncommon and poorly performing ones get grouped on the opposite side.
This can help brands determine whitespaces and plan future content based on creative-specific factors they might not have noticed otherwise.
One fashion retailer, for example, realized the “hats” tag occurred rarely in their content but saw a high clickthrough rate each time it did, so they increased the number of posts featuring hats to capitalize on the emerging trend.
“We didn’t really think anything of it until I was walking down the street in New York City and every girl had a [base]ball cap on,” Alvares said.
Audience Planning
Traditionally, audience planning is done with information gleaned from static surveys and purchasing data to better understand consumer behavior. But if that data is hard to access or out of date, then all that planning is rendered moot at best.
In contrast, Alli’s Audience Planning tool, which was also released this year, uses first-party data from social media and “about five years of consumer insights” to produce complete, AI-generated user profiles in the style of traditional marketing personas.
This ability to dynamically plan audiences can give creative teams a base from which to generate ideas and media teams a better sense of what their budget allocation should be across different ad platforms, working for both sides of the advertising divide at once.
“The biggest thing I love about this tool is creative and media coming together,” said Alvares. “When you see that synergy between them, the performance goes [up] significantly.”
A marketer working for a pet food brand might type in “cat owners between 18-49 who buy flushable litter,” which would then yield one or more personas designed to include other targetable demographic and psychographic traits (e.g., gender, income or environmental consciousness).
The platform will also suggest geographical locations and preferred social media platforms where consumers fitting the persona can be found, as well as up-to-date data on other search terms or phrases they may identify with.
“When you take customer survey data, or purchasing behavior from other third-party sources, you can’t actually use that data to actually target [audiences] in each of the platforms that are out there,” Alvares said.
But with Alli, “in Facebook, with this persona, there are about 4.5 million people that I can target that have these specific geographies and these specific interests at heart.”
