Words Are Only 7% of the Sale

01 /The Concept

Influencer marketing has become a major industry. In 2023, spending exceeded $20 billion.

But most brands choose influencers on purely statistical grounds – follower count, view numbers, or simply personal taste. None of those metrics reliably predict whether a given influencer will actually move product when they record or speak a brand's message. The result: brands waste roughly 40% of their influencer budgets on partnerships that don't deliver.

Trudy helps brands evaluate influencers not by statistics but by what it calls "human metrics" – signals that can more accurately forecast how effectively an influencer will carry a brand's message. The primary format Trudy focuses on is video.

Trudy's AI reviews an influencer's existing video content and analyzes more than 160 variables: what they say, how they say it, which words they choose, what emotions they express. The output is two profiles – behavioral profile (observable traits) and personal values (what they actually believe).

The same engine can analyze a brand and compare that profile against an influencer's, predicting how well the two fit – in other words, how convincingly the influencer is likely to carry the brand's message.

At the current stage, the brand's profile appears to be built primarily from videos of influencers who have already proven effective for that brand. A brand submits a high-performing video, Trudy builds that influencer's profile, and then matches it against other influencers under consideration – predicting their effectiveness based on how closely their profiles align.

There's a useful side effect. Once Trudy understands an influencer's profile, it can advise brands on how to brief that influencer to get their best work. For a highly competitive influencer: "Take a look at this great piece another creator made for us – think you can top it?" For a highly creative one: "Here's a direction we think could work – but we'd love to hear how you'd approach it."

Beyond the human metrics, Trudy also reports standard reach, audience demographics, and engagement data – so brands can combine behavioral signals with quantitative reach to find their optimal fit.

The startup launched only in April of this year and has already closed its first £500K funding round.

02 /Why It Matters

At first glance this could look like pseudoscience. Does it really matter how an influencer's voice sounds, what expression they wear, what emotions come through – when the words themselves may have been written by the brand's marketing team?

But consider the famous line of theater director Stanislavski: "I don't believe it!" – delivered even when an actor was speaking the lines exactly as written. Words alone aren't the message.

Research on communication puts words at just 7% of what a speaker actually conveys. Vocal delivery – tone, pace, inflection – accounts for 38%. Body language, facial expression, eye movement, gesture: 55%.

In other words, 93% of a message's persuasive power comes from everything other than the words. That "everything else" is precisely what Trudy tries to measure and predict.

The same gap in persuasiveness shows up when comparing an influencer who was simply paid to endorse a product versus one who actually tried it, liked it, and is talking about it from genuine experience.

Several startups are approaching influencer effectiveness from that angle specifically.

Zipr – rebranded as Lava and [covered here](/review/iskrennjaja-reklama-luchshe-prodajot) last summer – turns a brand's actual customers into brand ambassadors. Real buyers naturally talk about products more convincingly. Their AI tool plugs into an online store and offers buyers discounts in exchange for social posts about their purchases. Zipr claimed those Instagram posts delivered nearly 3x the ROI of paid Instagram ads, counting the discount and the platform fee as the ad spend. The startup, now operating as Lava, raised $3.2M in new funding.

Stack Influence ([covered here](/review/sarafannoe-radio-vmesto-reklamy)) and Hummingbirds (also covered) take a product-seeding approach: brands ship samples to micro-influencers in exchange for honest posts about products they actually use. Stack Influence raised $1.27M; Hummingbirds raised $4.4M.

03 /Opportunities

Influencer advertising is large and still growing. The defining trend on this market is that advertisers are shifting attention from raw reach to actual effectiveness.

One strand of that shift is toward micro-influencers over celebrities with massive followings. Stack Influence puts influencer engagement at 6.7x higher for micro-influencers than for large ones.

Another strand is toward authenticity. Lava routes brand messages through real customers. Stack Influence and Hummingbirds seed products with micro-influencers for genuine reviews. The Desire Company ([covered here](/review/chestnaja-reklama-bolshaja-glupost-ili-bolshaja-cel)) publishes expert video reviews from specialists who receive products to evaluate but don't know whether the review is sponsored – enforcing honest opinions by design.

Trudy fits into the same trend, approaching it from a different angle: helping brands understand in advance how convincingly a specific influencer will deliver their message. Less "authentic" in the editorial sense – but effective is effective.

The overall direction: platforms and tools that help brands extract more value from influencer spending. The approaches are varied, some of them genuinely surprising. But the market is large enough that multiple models will find their customers. The question is which path makes sense for you.