Viral Isn't the Same as Measurable: Rethinking Influencer Marketing ROI

In 2022, a Clinique lipstick shade launched in 1971, Black Honey, started resurfacing across TikTok: no paid push, no coordinated campaign, just creators rediscovering an old product and talking about it. Estée Lauder Companies' measurement framework tracked what happened next: earned media value tied to the moment grew from $6.4 million to $8.7 million over two years, with effects traceable across awareness, web traffic, and conversion.

Most brands would have logged this as a nice EMV spike and moved on. Estée Lauder Companies could instead show how a single organic moment moved measurable business outcomes across the funnel. That gap, between "we saw a spike" and "we know what it produced", is exactly where most influencer measurement programs still sit today.

The channel that outgrew its own metrics

Influencer marketing didn't creep up on marketing budgets. It overtook paid search outright to become the world's largest digital advertising channel, with global investment reaching $32.5 billion in 2025, up 35.6% year over year. Unilever has publicly committed to raising its influencer media investment from 30% to 50% of total advertising spend, a structural reallocation, not a test budget. As Unilever CEO Fernando Fernandez put it, the brands that let influencers, celebrities, and creators speak for them at scale are building the systems that matter now.

The metrics most programs still report on (Earned Media Value, impressions, and engagement rate) were built for a much earlier moment: proving the channel deserved a budget line at all. They were never built to optimize a nine-figure program or defend it in a cross-channel allocation conversation.

Two problems: what gets measured and when

Most influencer measurement gaps trace back to two distinct issues that compound each other:

• Brand-building and conversion are getting blended into a single number. Organic reach and paid amplification behave nothing alike; they have different saturation curves, different audiences, and different credibility with the viewer. A model that aggregates the two produces a figure that's wrong in both directions at once: it understates the compounding brand value of organic reach while overstating what boosted content alone is actually converting. Estée Lauder Companies' own framework avoids this by modeling desirability, consideration, and conversion as distinct, fund-mapped signals rather than one blended metric.
• Most programs measure creators by tier when they should measure them by moment. Engagement rate declines as follower count rises, which means a creator's best return doesn't line up with their follower peak; it lines up with their engagement peak, a separate and often earlier point in their trajectory. A program built around static tier labels (nano, micro, and macro) will keep misjudging which creators deserve the next dollar, because tier tells you almost nothing about where a specific creator sits in their own lifecycle right now.

Solving both requires the same prerequisite: unified data. Influencer platform data, paid media data, CRM, and commerce data need to sit in one place before any model can meaningfully separate these effects. Without that, sophistication on top doesn't help.

What emotionally-driven categories reveal about the upside

Beauty and fashion consistently show the clearest evidence of what's possible once measurement catches up to the channel. Maybelline's TikTok-first livestream strategy in Indonesia generated 1.1 billion views and TikTok Shop sales outpacing the category by 2x, with ROAS exceeding benchmarks by 2.7 to 3.5x. The result didn't come from reach alone; it came from treating livestream commerce as a measurable pathway in its own right, distinct from the brand-building effect of the same content circulating organically. That distinction is exactly what most measurement programs still collapse into a single number, which is why so few brands can point to a comparable result with the same confidence.

The reverse is also true, and worth naming: markets like China show ROI declining despite continued high investment, a saturation signal that only shows up if a brand is tracking incrementality rather than reach.

Where Ekimetrics starts with clients

Every brand that comes to us with an influencer program already has data. What they usually don't have is a single view of it. Our starting principles:

• Break down the silo between influencer, media, CRM, and commerce data before attempting any modeling refinement on top.
• Disaggregate by platform, tier, format, and lifecycle stage rather than aggregating by tier alone, since tier-level aggregation is what causes over- and under-valuation in the first place.
• Model halo (brand) and call-to-action (conversion) pathways separately inside a full-funnel Marketing Mix Model, validated with uplift tests or synthetic control groups rather than assumed from platform-reported engagement.
• Benchmark ROI by tier, channel, and market maturity, since a program that looks flat in aggregate is often strong in some segments and saturated in others.

The output isn't a cleaner engagement dashboard. It's a reallocation signal: which creators are at their engagement peak right now, which content is building brand equity that converts later, and which spend is simply generating reach that isn't moving anything.

"Creating marketing activity systems where others can speak for your brand at scale is incredibly important", Unilever CEO Fernando Fernandez said

of the shift toward creator-led marketing. The scale is no longer in question: influencer marketing has already claimed the top spot in digital advertising. What's still catching up, for most brands, is the ability to know what that scale is actually producing.

‍

Want to go deeper?

Read our full white paper for the complete measurement framework, case studies, and capability model.

Discuss your influencer measurement framework with our data science experts.