The audio signal has never been measured
A senior media buyer watches a new ad cut. The hook is strong. The creative concept is solid. The product looks great. They approve it. It goes live. It underperforms.
Why? The platform shows CTR and ROAS. Outcome data. It shows nothing about the causal factors in the creative. The buyer watches the ad again. Listens more carefully. "There's something off about the CTA," they say. "The energy drops." But what does that mean? At exactly which second? By how much? And what would it look like if it didn't?
"The things that separate a high-performing ad from a low-performing one are often completely invisible to the people reviewing it. Not because they lack skill. Because the instruments to measure them didn't exist."
This is the problem AdZhi was built to solve. Not "what performed well". Platforms tell you that. But "why did this specific delivery work or fail, and exactly what should change."
The insight is acoustic. The signals your audience's subconscious processes in the first 150 milliseconds: pitch, energy, credibility, delivery pattern. These determine whether they swipe or stay. They've never been measured at scale for ad creative.
Four forces converging
Four convictions we build from
The voice carries more persuasion weight than motion or colour
The visual gets all the attention: motion, colour, faces. But the words, and specifically how they're delivered, carry the persuasive load. Pitch authority. Vocal warmth. Energy at the moment of the ask. Your audience hears all of this before they consciously process any of it.
Pre-production is where the leverage is
Post-production reporting tells you what happened. Pre-production analysis gives you a chance to change it. The best creative intervention costs a creator 20 minutes to re-record, not £20,000 to re-shoot.
Specific beats generic
"Improve your CTA" is advice. "Re-record from 00:24. Your CTA was delivered at 68% of your average energy. Stand up. Mean it. Target 85+ CTA energy." is a brief. Every AdZhi output is a brief, not a score.
The data should be honest about what it knows
Current predictions are calibrated against observed patterns across thousands of ad analyses. As more brands connect their real performance data, the signals get validated against actual ROAS and CTR outcomes. We say which is which. We don't call a pattern a proof.
Ad + 知 — to know, from the source
The name has a meaning. Ad — what we analyse. Zhī (知) — the Mandarin character for "to know", "to be aware of". 知道 (zhīdào): to know, to understand.
It comes from time spent studying in China, and years working as a data analyst across social publishing and influencer marketing. The insight that kept surfacing was the same: you could measure reach, engagement, and conversion — but never the creative quality that caused it. Not systematically. Not at scale.
"The signals your audience's subconscious processes in the first 150 milliseconds — pitch, energy, credibility — have always been the differentiator. Now they can be measured."
AdZhi is that measurement. Built in London by people who've spent years on the buying side of creative, watching ads underperform with no instrument to say why.
Background in the work, not just the tool
AdZhi was founded with a background in data analysis across social publishing, influencer marketing, and DTC performance. Not as an observer — as a practitioner building reporting infrastructure, watching creative decisions get made without the data to support them.
The specific frustration: platforms give you outcome data. They give you CTR, ROAS, CPM. They tell you what happened after the money was spent. They give you nothing about the causal factors in the creative. That gap — between what performed and why it performed — is what AdZhi is built to close.
Built AdZhi after seeing creative briefing collapse repeatedly as the cause of underperforming campaigns — not targeting, not budget, not platform. The specific problem: platforms give you outcome data (CTR, ROAS, CPM). They give you nothing about why a piece of creative performed. That gap is what AdZhi closes — from the raw audio file, before media spend.
The frustration came from spending years on the data side of social publishing and influencer marketing — building reporting infrastructure, watching ad after ad go live, and having no instrument to explain why some delivered and others didn't. The platform would say CTR was low. The creative team would watch it back and say "something feels off about the CTA." Nobody could say exactly what, or exactly where, or exactly how to fix it.
I tried listening frameworks, post-production review templates, creative scorecards. None of them were specific enough to brief a creator. So I built the instrument I kept wishing existed: one that reads the raw waveform, tells you at exactly which second the energy drops, and gives you a brief you can hand to someone today.
Trained in statistical methods; designed AdZhi's scoring to require p<0.05 significance before any correlation claim is surfaced to users. We do not call a pattern a proof.
The first cohort of brands using AdZhi are shaping what it becomes: which signals matter most in their vertical, which output format actually gets used in a brief. If you're running video ads at scale, we want to hear from you — and early users get direct access to the roadmap.
Get in touch →Live and working. More coming.
We're early. Come help us build it.
Whether you're running ads and want a faster feedback loop, or you're just curious what your creative actually sounds like. The product is live and the first analysis is free.