The product
AdZhi reads the creative behind your ad’s numbers.
AdZhi reads the hook, voice delivery, CTA energy and persuasion structure straight from the raw file. It identifies the likely creative weaknesses and the change worth testing first: a timestamped brief in under a minute.
Keep the confession hook and the relatable problem beat. Rebuild the final 8 seconds: add one proof point, then a confident, high-energy CTA.
- The confession hook
- Specific problem detail (£80 serums)
- Warm, credible vocal tone
- Apologetic CTA — make it direct
- Flat pitch on the benefit claim
- No social proof before the ask
CTA direction Re-record standing, higher energy: “Tap the link and start tonight.”
See the signals behind your results, straight from the creative.
AdZhi starts with the raw audio waveform and measures pitch, energy and delivery quality at the signal level: the signals to test against your CTR and ROAS. You know what to fix, and where, before you run it.
- Pitch: expressiveness, monotony, emphasis
- Energy: where it peaks and drops, CTA momentum
- Voice credibility: trust, clarity, warmth
- Disfluency: fillers flagged by exact timestamp
- Audio environment: noise, music bed, clarity
- Rhythm: pacing, pauses, cadence
- Voice IQ81 strong delivery
- Hook energy118% of its average; drops 32% at the CTA
- Disfluency3 2 fillers near the CTA
Every persuasion act mapped. Every gap named.
AdZhi maps the entire script to named persuasion acts (Hook, Problem, Solution, Social Proof, CTA), rates how strongly each one lands, and names what’s missing and what that costs you.
- Each act placed on the timeline and rated strong, adequate or weak
- Structure rated against AIDA, PAS and Hook-Story-CTA, scored in the Structure section
- Missing elements flagged explicitly
- Acoustic signals feed the verdict
- Timestamped to the exact moment
Each act placed on the ad’s own timeline and rated strong, adequate or weak. The ask lands weak, with no proof before it: that’s where this ad leaks.
A brief your creator can act on today.
Three rewrite variants with specific delivery direction, then a ready-to-send creative brief. Not “improve your CTA”: “re-record from 0:24 with 20% more energy and a direct imperative verb.”
- Rewrites informed by every acoustic signal
- Delivery notes target the specific problems found
- Brief: one-liner, keep/fix, hook + CTA direction
- Word count calibrated to the original pacing
- PDF export: hand to a creator or client
- 1Re-record the CTA from 0:24, 20% more energy, a direct imperative verb.
- 2Cut 1.5s of dead air at 0:11, just before the proof beat.
Measured from the file itself — not inferred from your results.
Eight sections, each one question about your ad, read straight from the upload. Every ad platform can only tell you what happened after the spend; this is the read you get before it. Upload a file, get a timestamped brief in under a minute.
- 21%
Hook
Does the opening stop the scroll? 2 measurements, e.g. Hook delivery time, Primary tone.
- 17%
Call to action
Is the ask clear and delivered with intent? 3 measurements, e.g. CTA timings, Power words, Benefit vs feature.
- 8%
Offer
Does the ad give people a reason to act now? 5 measurements, e.g. Urgency cues, Social proof, Cialdini principles.
- 13%
Attention
How long does the ad hold a viewer? 6 measurements, e.g. Pause & silence ratio, Dramatic pauses, Pacing variance.
- 13%
Structure
Does the ad build to its point? 4 measurements, e.g. Energy dynamic range, Peak-energy timing, Sentiment (VADER).
- 13%
Voice IQ
Does the delivery sound credible and engaged? 9 measurements, e.g. Pitch (F0), Pitch variation, Voice clarity.
- 9%
Alignment
Does the voice match what the script says? 1 measurement, e.g. Speech rate (wpm).
- 6%
Visuals & sound
Do the picture and sound carry the message, including with the sound off? 22 measurements, e.g. Loudness (dBFS), Tempo & beat, Musical key & mode.
52 measurements, 8 sections, one AdZhi Score. Alongside them, eight named diagnostics (Trust Index, CTA Momentum and the rest) sit beside the section they explain, and Persuasion DNA profiles the ad to find similar past ones. Neither is added into the score.
Connects to where your ads run
Connect an account and the loop closes: AdZhi pulls your real outcomes back in, scores each forecast against them, and tunes itself to your account.
Meta Ads Available
Pull your full creative library with spend, ROAS and CTR. Queue underperformers in one click.
TikTok for Business Available
Connect and pull your library with performance data. Spot which hooks lose the scroll.
YouTube URLs Available
Paste any URL: your ads, competitor ads, research. No download. Analysis starts immediately.
Performance correlation Available
Import a CSV of ROAS/CTR/spend. AdZhi matches each row to an analysis.
Webhooks Available
A signed POST fires when a job completes. Wire it into Slack, Zapier or your CMS.
PDF + CSV export Available
Full analysis as PDF for a creator or client. History as CSV.
Google Ads, Pinterest, LinkedIn and Snapchat connections are in beta: built, but not yet verified against live accounts.
From the read to the next decision
Connect an account and AdZhi doesn’t stop at a score. It turns every read and every real result into the next move, and you choose how much of that call to hand over.
The agent decides and tells you why. It never changes your ads or your spend: live execution is off by default, and only you can switch it on.
- 01
One ranked queue
Every analysed ad becomes a decision (scale, fix or retire), ranked by impact, with its trigger, the evidence and the expected lift.
- 02
You decide, or the agent does
Keep the call, or delegate any ad, ad set or campaign. On a delegated scope the agent makes the decision and emails you what it chose and why.
- 03
A verdict on every live ad
Scale, fix or retire: the creative signals read beside the live numbers, so you know which running ads to back, which to re-cut and which to switch off.
- 04
A costed budget plan
A draft allocation across your top creatives, ranked by realised ROAS — a plan you sign off before anything runs.
- 05
It earns trust on your numbers
See how the forecasts score against your real outcomes: out-of-sample, per metric, beside the naive baseline. The record is the proof.