AI Is Only as Smart as the Data You Feed It: Alliant’s Dave Taylor
CANNES, France – Artificial intelligence may be the hottest guest at every advertising conference, but Dave Taylor isn’t ready to hand it the keys just because it showed up wearing a shiny new algorithm.
Speaking with Beet.TV contributor David Kaplan at the Cannes Lions International Festival of Creativity, the chief product officer at Alliant argued that AI’s biggest challenge isn’t a lack of hype. It’s a lack of high-quality data.
In other words, AI can cook dinner. But if you hand it spoiled ingredients, don’t blame the chef.
Bigger isn’t always better
Taylor said marketers shouldn’t confuse giant datasets with useful ones, especially as AI becomes more deeply embedded in campaign planning.
“Without the best data, the AI is kind of almost useless to a degree,” Taylor said. “Making sure that we hone in on the right data assets… is really paramount.”
Asked how marketers can distinguish quality data from data that simply looks impressive in a PowerPoint deck, Taylor offered a gentle reality check.
“Biggest isn’t always the best,” he said. Instead, marketers need a variety of data assets that provide predictive value rather than sheer volume.
“You don’t want to just have the scale,” Taylor said. “You want to have the predictive nature of the data.”
That’s welcome news for anyone who’s ever been handed a spreadsheet with 47 million rows and exactly zero useful insights.
No single data source wins the game
Taylor said Alliant’s approach combines predictive, probabilistic and contextual data because each serves a different purpose.
When building custom audiences, “the blend and the combination of many different data assets coming together is really what’s going to be powerful,” he said.
Predictive models help forecast future behavior. Historical data explains what already happened. Context adds another layer of relevance. Rather than declaring one methodology the winner, Taylor argued that marketers should match each data type to the objective of the campaign.
“There’s really just a combination of all of those together to help drive the outcome that you’re looking to achieve,” he said.
It’s less Avengers versus Justice League and more everybody reluctantly agreeing to work on the same group project.
Beyond audience segmentation
Custom audiences may now be standard practice across digital advertising, but Taylor believes the next frontier extends well beyond building better audience lists.
“AI… we know AI is here,” he said, while also pointing to the industry’s long-running quest for closed-loop measurement across channels including direct mail, digital and out-of-home.
What fascinates him most, however, isn’t another attribution model. It’s the next generation of consumers.
“I’m interested in… my son’s age,” Taylor said. “How are they going to consume content?”
He even floated a possibility that might surprise marketers who haven’t checked their mailbox lately.
“Is direct mail going to make a comeback?” Taylor asked. “Because it’s the new thing out there now.”
Advertising has officially entered the stage where physical mail is starting to sound innovative again. Somewhere, a catalog printer just smiled.
Privacy starts before the campaign does
Taylor also argued that privacy can’t be treated as a compliance exercise that happens after audiences have already been built.
“You want to make sure that you don’t just do the privacy at the end,” he said. Instead, marketers should build governance, permissions and privacy controls into every stage of the data pipeline.
For Taylor, that means ensuring companies have the proper technology and data controls from the moment consumer information enters the system through campaign activation and measurement.
The goal isn’t simply staying compliant. It’s creating a process that protects consumer data while still producing meaningful marketing outcomes.
As AI continues to reshape advertising, Taylor’s message was refreshingly unglamorous. Fancy algorithms are great. But if the underlying data is a mess, all you’ve built is a very expensive machine that gets the wrong answer faster.
You’re watching Beet.TV coverage from Cannes Lion 2026. For more videos from this series, please visit this page.
CANNES, France – Artificial intelligence may be the hottest guest at every advertising conference, but Dave Taylor isn’t ready to hand it the keys just because it showed up wearing a shiny new algorithm.
Speaking with Beet.TV contributor David Kaplan at the Cannes Lions International Festival of Creativity, the chief product officer at Alliant argued that AI’s biggest challenge isn’t a lack of hype. It’s a lack of high-quality data.
In other words, AI can cook dinner. But if you hand it spoiled ingredients, don’t blame the chef.
Bigger isn’t always better
Taylor said marketers shouldn’t confuse giant datasets with useful ones, especially as AI becomes more deeply embedded in campaign planning.
“Without the best data, the AI is kind of almost useless to a degree,” Taylor said. “Making sure that we hone in on the right data assets… is really paramount.”
Asked how marketers can distinguish quality data from data that simply looks impressive in a PowerPoint deck, Taylor offered a gentle reality check.
“Biggest isn’t always the best,” he said. Instead, marketers need a variety of data assets that provide predictive value rather than sheer volume.
“You don’t want to just have the scale,” Taylor said. “You want to have the predictive nature of the data.”
That’s welcome news for anyone who’s ever been handed a spreadsheet with 47 million rows and exactly zero useful insights.
No single data source wins the game
Taylor said Alliant’s approach combines predictive, probabilistic and contextual data because each serves a different purpose.
When building custom audiences, “the blend and the combination of many different data assets coming together is really what’s going to be powerful,” he said.
Predictive models help forecast future behavior. Historical data explains what already happened. Context adds another layer of relevance. Rather than declaring one methodology the winner, Taylor argued that marketers should match each data type to the objective of the campaign.
“There’s really just a combination of all of those together to help drive the outcome that you’re looking to achieve,” he said.
It’s less Avengers versus Justice League and more everybody reluctantly agreeing to work on the same group project.
Beyond audience segmentation
Custom audiences may now be standard practice across digital advertising, but Taylor believes the next frontier extends well beyond building better audience lists.
“AI… we know AI is here,” he said, while also pointing to the industry’s long-running quest for closed-loop measurement across channels including direct mail, digital and out-of-home.
What fascinates him most, however, isn’t another attribution model. It’s the next generation of consumers.
“I’m interested in… my son’s age,” Taylor said. “How are they going to consume content?”
He even floated a possibility that might surprise marketers who haven’t checked their mailbox lately.
“Is direct mail going to make a comeback?” Taylor asked. “Because it’s the new thing out there now.”
Advertising has officially entered the stage where physical mail is starting to sound innovative again. Somewhere, a catalog printer just smiled.
Privacy starts before the campaign does
Taylor also argued that privacy can’t be treated as a compliance exercise that happens after audiences have already been built.
“You want to make sure that you don’t just do the privacy at the end,” he said. Instead, marketers should build governance, permissions and privacy controls into every stage of the data pipeline.
For Taylor, that means ensuring companies have the proper technology and data controls from the moment consumer information enters the system through campaign activation and measurement.
The goal isn’t simply staying compliant. It’s creating a process that protects consumer data while still producing meaningful marketing outcomes.
As AI continues to reshape advertising, Taylor’s message was refreshingly unglamorous. Fancy algorithms are great. But if the underlying data is a mess, all you’ve built is a very expensive machine that gets the wrong answer faster.
You’re watching Beet.TV coverage from Cannes Lion 2026. For more videos from this series, please visit this page.
Topics Videos
