Why brands are shifting toward ‘less precise, more accurate’ gauges for paid social

As they struggle to estimate the impact of their paid social and creator marketing activity, more marketers are turning away from deterministic measures of media success, and towards hybrid approaches and mixed media modelling (MMM) AI software.

It’s a sea-change for marketers used to depending on last-touch attribution metrics that come directly from social platforms.

Practitioners in the paid social and influencer marketing space have long been skeptical regarding the use of last-touch attribution models because they risk unduly crediting lower-funnel marketing channels with sales. 

According to a November, 2026 ANA study, 40% of marketers report that proving the return-on-investment (ROI) of their creator marketing spend is their biggest challenge, while 36% said their top problem was attributing sales back to influencers.

Related Insights

But such measures give clear, deterministic answers to marketers worried about proving the worth of their work to skeptical finance teams. Alternatives, such as mixed media models, didn’t provide answers fast enough for clients weighing up mid-flight changes to campaigns.

“Historically, marketers have leaned towards those deterministic KPIs,” said Rita Steinberg, vp, media at Toronto-based full-service agency Fuse Create. “[But] the consumer journey has shifted so much that it’s just not really realistic to rely on that.”

That’s begun changing over the last year. Executives at four media agencies told Digiday their clients and buyers had begun embracing a more nuanced collection of performance metrics to guide investment decisions, such as UTMs (urchin tracking module, a means of tracking website visitors), brand lift and post-purchase surveys, platform pixels or affiliate tracking, and often using generative AI tools to stitch them together.

“[There] is a shift in thinking across the paid social space,” said Chris Robinson, interim head of paid media at creator and paid social agency Goat.

It’s leading to clients adopting an approach that is “less precise, more accurate,” for clients trying to understand which investments pushed the needle, according to Matthew Chappell, global client success officer at WPP-owned Gain Theory, which provides effectiveness monitoring for brands like BetMGM.

In short, efforts to measure the effect of paid social investments (which account for 32.6% of digital ad spending in the U.S., according to eMarketer) are moving on from attempts to exactly credit a given channel with a conversion or sale, and towards modeling that aims to be good enough to allocate a budget.

At Goat, Robinson said the agency used brand lift survey provider ThisThat and marketing data platform Circana to provide a surround-sound view to clients. Steinberg said that Fuse Create often used software called Tracksuit to achieve a similar result.

The availability of generative AI solutions has catalyzed this shift, however. Steinberg, for example, said that Fuse teams had recently begun using custom GPTs within Perplexity and ChatGPT equipped with API access to platform data sources, to quickly compare and contrast campaign metrics. The agency’s clients include protein brand Grenade and Jose Cuervo.

Six months ago, indie media agency Go Fish began using “high velocity” MMMs for clients built upon Google’s open source Meridian model, which has been available since January 2025. 

David Dweck, Go Fish’s president, said the models had helped provide evidence for paid social advertising’s positive impact on e-commerce sales and enabled clients to map the effect of combining Meta ads with spending on Amazon.

“The idea was to give us a better sense of how we can optimize day to day vs. six months,” said Dweck. On the back of a six-week test conducted between July and August last summer, one unnamed DTC client increased its media investments from $100,000 to $500,000 a month.

The generative AI explosion, and the industry’s drawn-out downgrading of the third party cookie has led agency and client practitioners to re-evaluate MMMs for modern advertising. Last October, agency Ars X Machina launched an agile mix modeling proprietary platform that promised full campaign measurement across the entire spread of marketing channels. “It’s easier to access, it’s faster than it was before,” said Gain Theory’s Chappell.

This doesn’t always entail building bespoke models. Media by Mother has been using MMMs from software-as-a-service company Mutinex, according to Olamma Nzeribe-Williams, social activation manager, to provide a more holistic understanding of paid social spending.

Such models allow markets previously wedded to measurement models like last-touch attribution to understand the impact of their marketing activity across a broader swathe of channels.

In turn, it’s gradually making it easier for buyers to persuade clients to invest more in upper-funnel ad formats – which in a paid social context, would be units such as Instagram Reels or TikTok Pulse ads.

“That is where the non deterministic models [are] going to make the biggest impact,” said Nzeribe-Williams.

Goat’s Robinson agreed. “The shift we’re seeing is a move away from a singular, siloed view of determining the success of a campaign to looking at lots of factors and then pulling in information from multiple sources… because buying decisions aren’t linear, and so we shouldn’t measure them as linear.”