{"id":75821,"date":"2025-05-05T09:09:56","date_gmt":"2025-05-05T09:09:56","guid":{"rendered":"https:\/\/mailinvest.blog\/index.php\/2025\/05\/05\/progressive-insurance-tests-limits-of-ai-generated-ads-and-learns-when-to-pull-back\/"},"modified":"2025-05-05T09:11:08","modified_gmt":"2025-05-05T09:11:08","slug":"progressive-insurance-tests-limits-of-ai-generated-ads-and-learns-when-to-pull-back","status":"publish","type":"post","link":"https:\/\/mailinvest.blog\/index.php\/2025\/05\/05\/progressive-insurance-tests-limits-of-ai-generated-ads-and-learns-when-to-pull-back\/","title":{"rendered":"Progressive Insurance tests limits of AI-generated ads \u2013 and learns when to pull back"},"content":{"rendered":"

<\/a>\n
<\/p>\n

We meet up with the insurance coverage firm and its advertising firm, Claritas<\/span>, to listen to how generative AI reworked their advert creation, producing 96 audio variants in two weeks, boosting outcomes and exposing each the ability and limits of AI-driven personalization in advertising.<\/strong><\/p>\n

At Possible<\/a>, amid the noise of AI predictions and platform guarantees, one case examine stood out for its readability \u2013 and its candor. We sat down with Robin England of Progressive Insurance coverage and Rex Briggs, chief AI officer at Claritas<\/span>, to listen to how a significant US insurer quietly re-engineered its artistic course of utilizing generative AI. Not simply to make adverts quicker, however to check the boundaries of what automation can actually ship \u2013 and the place the human contact nonetheless issues. <\/p>\n

\n

This wasn\u2019t a press launch masquerading as a case examine. It was a glimpse right into a dwell experiment \u2013 one which produced 96 audio advert variants in simply two weeks, elevated quote begins by 31% and opened up solely new questions on how a lot personalization is an excessive amount of. <\/p>\n

\u201cWe have been the primary to use GenAI to jot down the scripts, create the voices and the music beds \u2013 all of it,\u201d says England, who leads technique, analysis and analytics at Progressive. \u201cWe usually produce a handful of audio adverts over seven weeks. This time, we had 96 made and permitted in 14 days.\u201d <\/p>\n

\n

Need to go deeper? Ask The Drum<\/h4>\n


\n <\/p>\n

<\/miso-explore>\n <\/p>\n<\/div>\n

Claritas<\/span>, Progressive and the Consortium for AI Personalization <\/strong><\/h2>\n

Claritas<\/span>, a advertising intelligence firm with roots going again almost 5 many years, performed a central function within the course of. Recognized for its industry-standard identification graph and segmentation instruments, Claritas<\/span> has more and more moved into AI-driven personalization. It acquired patented AI tech to generate and optimise artistic messaging in actual time, permitting manufacturers to pair the best message with the best particular person \u2013 or persona \u2013 dynamically. <\/p>\n

Briggs, who beforehand helped pioneer digital advert measurement, cellular effectiveness and social media analytics, got here out of retirement to hitch Claritas<\/span>. \u201cI used to be purported to be in Tahoe,\u201d he advised us. \u201cHowever I learn the patent, noticed the potential and realized this was the largest shift in advertising I\u2019d ever seen.\u201d <\/p>\n

Along with MMA International \u2013 a commerce affiliation representing CMOs and advertising innovators \u2013 Claritas<\/span> and Progressive ran the mission via the Consortium for AI Personalization, a analysis initiative the place manufacturers take a look at real-world purposes of AI and share what they be taught. <\/p>\n

Why audio? Why now? <\/strong><\/h2>\n

Progressive selected audio not as a result of it was straightforward, however as a result of it was possible. \u201cOn the time, producing real looking video utilizing AI wasn\u2019t fairly there,\u201d says England. \u201cHowever artificial voices and music have been already ok to place in entrance of shoppers.\u201d <\/p>\n

The primary marketing campaign launched on SiriusXM, then expanded to Spotify, iHeart and different main streaming platforms. The system allowed Progressive to range script, voice and music background in actual time. The outcome: six scripts \u00d7 4 voices \u00d7 4 tracks = 96 mixtures, every tuned to resonate with completely different segments. <\/p>\n

\u201cWe even talked to artificial personas \u2013 constructed from many years of Claritas<\/span> knowledge \u2013 to ask what sort of music they most popular,\u201d says Briggs. \u201cThey advised us nation music was widespread amongst doubtless converters. That wasn\u2019t one of many unique beds. So within the subsequent spherical, Progressive added it.\u201d <\/p>\n

The personalization paradox <\/strong><\/h2>\n

The outcomes have been clearly optimistic. However not good. <\/p>\n

\u201cWhat we discovered is that 96 variations have been possibly too many,\u201d says Briggs. \u201cThere\u2019s an optimum ratio \u2013 what we name the learning-to-earning steadiness. If you happen to solely have two folks changing from 1,000 impressions, you don\u2019t have sufficient knowledge for the AI to be taught from 96 variants.\u201d <\/p>\n

That is the place Briggs provides a dose of realism usually lacking from AI discussions: the so-called 5 nines of accuracy. \u201cEven when your mannequin is 99.999% correct, that\u2019s nonetheless one improper lead to each 100,000 choices,\u201d he explains. \u201cNow think about you\u2019re delivering 40m advert impressions. That\u2019s 400 errors and, in regulated industries like insurance coverage, that\u2019s 400 too many.\u201d <\/p>\n

Therefore, the significance of a human within the loop. Progressive\u2019s artistic director, Chris Monaco, reviewed and filtered 126 GenAI ideas to choose the ultimate 96. Each variant ended with the identical, legally required disclaimer, voiced identically. \u201cWe didn\u2019t reduce corners,\u201d says England. \u201cCompliance was baked in.\u201d <\/p>\n

Environment friendly, sure \u2013 however efficient too <\/strong><\/h2>\n

The numbers converse for themselves. AI-enabled campaigns delivered a 197% elevate over the management group. Quote begins jumped 31%. GenAI-enhanced audio adverts tripled publicity and saved 98% of listeners engaged via to the tip. <\/p>\n

And maybe most usefully, Progressive found a brand new framework for speedy experimentation \u2013 a technique to take a look at hypotheses about tone, message, provide or supply with out ready months for insights. <\/p>\n

\u201cIt compressed the entire cycle,\u201d England says.<\/p>\n

Trying forward: video, electronic mail, brokers <\/strong><\/h2>\n

Claritas<\/span> is already increasing this mannequin to different media, together with generative video and electronic mail. Briggs sees a near-term future through which 20% of manufacturers run totally AI-orchestrated, multichannel campaigns \u2013 personalised in actual time, anchored to first-party knowledge and optimized constantly. <\/p>\n

However even he acknowledges there are limits. \u201cTill the fashions are flawless, people nonetheless have to assessment what goes out.\u201d <\/p>\n

England agrees. \u201cAI doesn\u2019t change advertising,\u201d she says. \u201cIt simply adjustments it. And should you\u2019re a marketer, which means studying the best way to use it \u2013 now.\u201d <\/p>\n

Recommendation for manufacturers nonetheless watching from the sidelines? <\/p>\n

\u201cDon\u2019t be intimidated,\u201d says Briggs. \u201cHowever do become involved. The hole between those that experiment now and people who wait goes to be very troublesome to shut later.\u201d <\/p>\n

England\u2019s recommendation is equally blunt: \u201cDon\u2019t assume AI is another person\u2019s job. It\u2019s everybody\u2019s job. And it\u2019s shifting quick.\u201d <\/p>\n

\n

Recommended newsletters for you<\/h4>\n

\n <\/p>\n

\n
\n <\/p>\n

Daily Briefing<\/h4>\n

Daily<\/h4>\n

Catch up on the most important stories of the day, curated by our editorial team.<\/p>\n<\/p><\/div>\n

\n <\/p>\n

Weekly Marketing<\/h4>\n

Friday<\/h4>\n

Stay up to date with a curated digest of the most important marketing stories and expert insights from our global team.<\/p>\n<\/p><\/div>\n

\n <\/p>\n

The Drum Insider<\/h4>\n

Once a month<\/h4>\n

Learn how to pitch to our editors and get published on The Drum.<\/p>\n<\/p><\/div>\n<\/p><\/div>\n<\/div>\n<\/div>\n