For creative teams to get what they need from AI, they need to collaborate much more closely with their analytics teams.
Effective measurement is often the last frontier for digital advertisers. There is always a new channel, a new creative format, new targeting or new delivery method popping up that complicates measurement.
Ideally, marketers would measure the effectiveness of their creatives with their target audiences at scale across all channels. They would also measure the response from audiences and other targeting elements, and have a reliable attribution model that understands the precise effect on ROI and sales of every campaign element, from ad format to placement to time of day.
Very few, if any, advertisers have all these elements in place.
Taking the measure
Putting together great ad measurement has been historically hard. Testing takes time and is costly, models need a lot of scale to deliver high confidence, and attribution offers feedback that is often too slow to keep up with campaign changes.
Sick of being disappointed, many advertisers get used to what they have and start putting false confidence into their less-than-scientific approach.
Case in point: about two-thirds of advertisers feel confident in their ability to measure ROI in their marketing and advertising, yet many advertisers are using metrics that do nothing to actually provide ROI insights.
While a decent number of advertisers are using some form of attribution, many use faulty methods such as “last touch.”
In just one contradictory example, 77 percent of marketers believe they are using the wrong measurement model for CTV. The reality simply does not jive with the majority of marketers who claim to be confident they are measuring ROI correctly.
Enter AI.
AI does two important things for marketers: save money and prove the value of marketing. Marketers need both badly.
Marketing budgets are only 7.7 percent of company revenue in 2024, down from an average of 11 percent in the preceding four years.
AI is great with crunching big sets of data quickly, which means the cost and time associated with reliable measurement can go down a lot, making it easier for advertisers to prove to their executives that what they are doing is working.
It sounds obvious that AI could help advertisers measure campaigns, but not that many are doing it.
A survey at an AdWeek event I attended last year showed less than half, 46 percent, of advertisers surveyed had implemented measurement using AI. Compare this to the 75 percent of advertisers using AI for content creation.
Jessica Pellen
Test creatives faster for less money
A study from Kantar found that cost was cited by 52 percent of advertisers as the primary barrier to testing ad creative for a campaign, followed by speed to market at 47 percent.
Using AI to aid with testing solves for both issues at once. Unilever notes that using AI testing helped the company achieve greater speed and scale with testing.
Rather than make a creative, test it, isolate specific aspects such as text, imagery or format and test again in a linear, manual way.
AI-based ad testing can dynamically generate and test ads in a matter of days.
Companies such as CreativeX and VidMob provide testing using AI that looks at how specific creative elements contribute to performance, providing much more granular insights than a typical A/B test would provide.
Similarly, there are companies that can evaluate and find target audiences and content such as Dstillery and Cognitiv that use AI and deep learning to find opportunities that pull from a huge variety of variables.
Compare that to a keyword list or demographic segment that groups audiences by income or how many kids they have.
In addition to speed and cost savings, AI campaign testing also opens up new possibilities.
An algorithm can also test a huge variety of variables all at once, such as optimal timing, ad format, placement, context and sentiment.
Advertisers can identify new audiences and new media placements with deep learning that they never could have identified manually or even through machine learning.
Gain confidence and get bolder
Most advertisers get excited about new opportunities, but are held back knowing that measuring something new can be difficult.
With multi-touch attribution getting more difficult with the erosion of third-party cookies, advertisers are turning to MMM and incrementality measurement, but they can be time consuming and complicated unless they are automated.
Adding something new to a media plan and testing it means creating a control group and suppressing spend against a particular audience or on a particular platform to measure incremental lift.
Determining how to do that can require political negotiation internally and, without the right amount of budget, end up not even working right.
Getting inconclusive results from a test often makes the whole process seem like a waste of time. Stakeholders do not have the proof points they need to expand their efforts and no one knows how to improve or optimize.
Companies such as Prescient AI and UpWave provide AI-based measurement that can make it easier to test campaigns with confidence.
At the end of a campaign, advertisers can actually get insights about the campaign to understand everything from brand metrics to how effectively they engaged with their target audience, to incremental sales lift. Attribution models become more reliable and accurate, too.
THE POWER THAT AI-based measurement brings to advertisers should not be underestimated.
AI opens up a new world of insights across elements that were previously unmeasurable. It allows advertisers to find new opportunities to advertise profitably, and it delivers insights about a campaign that advertisers can actually believe.
Feeling confident about test or campaign results means advertisers can make smart choices going forward: getting rid of campaign elements that do not provide value, leaning into strategies that do, and testing innovations that create new growth.
Jessica Pellen is research and insights director at Kargo, New York. Reach her at jpellen@kargo.com.