Every performance marketer knows the first three seconds matter. The hook. The pattern interrupt. The opening line. If you lose them there, you've lost them for good.
All true. But it's produced a collective blind spot: an obsession with the opening that leaves the close almost entirely unmeasured. And in acoustic terms, the close is where ads quietly fail.
When you analyse the audio waveform of a DTC ad (mapping pitch, energy, and words-per-minute across every second), one shape turns up again and again. The hook delivers. The middle builds. And then, right at the call to action, something happens that the media buyer never sees in the platform data: the voice drops.
Both bars are indexed to the ad's mean vocal energy (= 100), on the same scale as the segment charts below. The ask lands at 68 — a third below the mean, and the quietest moment in the ad.
The percentages, scores and thresholds in this article are worked examples of how AdZhi reads a CTA, chosen to make the mechanism concrete. They aren't statistics measured across a validated dataset, and consistent with our methodology, we don't present a relationship as proven until we've measured it against real outcomes.
What we measure at the ask
CTA Momentum is one of AdZhi's eight proprietary metrics. It measures the trajectory of vocal energy leading into and through the call to action: specifically, whether the ad builds toward the ask or decays before it.
To compute it, we extract root mean square (RMS) energy from the raw audio signal at millisecond resolution across the ad's full runtime. We identify the CTA window (typically the final 15 to 25% of the ad) using a combination of transcript pattern matching (urgency phrases, action verbs, time constraints) and pitch analysis.
We then compare the energy level at the CTA moment to the ad's running average. A CTA Momentum score above 70 means the ad builds into the ask: energy rises, WPM increases slightly, pitch drops intentionally on key words. Below 50 means the opposite: the creator has run out of steam by the time they get to the line that matters most.
Creators tend to spend the most cognitive effort writing the CTA copy — the exact words of the ask — and the least on how they deliver it. Our working hypothesis is that delivery carries as much weight as wording. Acoustic analysis exists to put that hypothesis to the test.
The anatomy of a low-momentum CTA
Here's what the acoustic profile of a typical underperforming CTA looks like. You'll recognise it immediately once you know what to listen for.
The creator opened with conviction. They built a case in the middle. And then, somewhere in the final third, the cognitive work of getting to the end of the script caught up with them. The delivery became rote. The urgency drained from the voice before the words of urgency were even spoken.
None of this is a craft failure. It's what happens when every hour of optimisation goes into the script and none into the delivery: the effort lands on writing "limited time only" rather than on meaning it.
Why viewers feel it even when they don't hear it
A viewer forms an impression of how much a speaker believes their own words in a fraction of a second, and most of that impression rides on the voice — its pitch contour, its loudness, the small catches before an important word — rather than on the sentence itself. Decades of vocal-cue research point in this direction; our methodology leans specifically on Zuckerman & Driver's 1989 review for the expressiveness signal. The upshot for a CTA is blunt: the delivery is carrying information the copy can't.
When you say "click the link now" at 68% of your usual vocal energy, the viewer registers the incongruence: the words say urgent, the voice says tired. They won't consciously think "this person sounds unconvincing." What they get is a vague sense of friction — a micro-hesitation that, in an environment of infinite scroll, resolves as a swipe.
Take a DTC skincare ad with a CTA Momentum score of 28. AdZhi's acoustic alert: "CTA delivered at 68% of your average energy, slowest WPM in the entire script. Re-record the last 6 seconds." That's a fix in hand before a penny of media: re-record the close with the energy the words are asking for, then put it back in the test.
The WPM factor
Energy alone doesn't tell the full story. Words-per-minute at the CTA moment carries its own signal. A CTA delivered significantly slower than the ad's average WPM, even if energy is maintained, reads as uncertain. The pause-and-drop pattern ("Click the link [pause] below [long pause] to get yours") acoustically communicates doubt.
A strong close usually holds 95 to 110% of the ad's average WPM: quick enough to feel deliberate, without tipping into a gabble. The pace says: I know what I'm asking, I expect you to do it, here's the action. The delivery matches the intent of the words.
Fixing a quiet ask
The diagnosis is acoustic. The fix is almost always in the recording, not the script.
- Re-record the last 6 to 10 seconds only. Most editors can drop in a single clip without re-shooting the whole ad. The hook is fine. The middle is fine. The close needs new energy.
- Get on your feet for the close. Voice coaches have people stand for a reason: it gives the breath more room to work, and the extra support usually comes through in the take.
- Run the CTA line out loud a couple of times before rolling. The first delivery is often a rehearsal masquerading as a performance. Roll after the line has settled into your mouth.
- End the script with the emotional benefit, then the action. "Get yours now" closes on mechanics. "You'll know exactly what your ads are doing — get yours now" closes on the promise. The emotion in the voice follows the content of the words.
- Don't read. Speak. If your eyes are on a script while you deliver the CTA, the viewer hears it. The micro-pauses as you track to the next line are acoustically identical to hesitation. Know the last line from memory before you record it.
What good CTA energy looks like
Here's the acoustic profile of a high-performing CTA from the same category (DTC skincare, 30-second format). Same script structure. Different delivery energy.
The hook is slightly softer: less "look at me," more "come closer." The energy builds continuously, and the CTA arrives above everything that came before it. The viewer's nervous system reads that as confidence. The swipe threshold rises. The click becomes easier.
The briefing implication
If delivery energy at the ask carries this much weight, your creative brief needs to address it explicitly — which is a document problem, and a solvable one. We've published a complete, copyable acoustic brief in How to Brief a Creator for Acoustic Performance.
The three-sentence version costs nothing to add today: "The CTA should be your highest-energy moment. Stand for the final segment. Record the ask early in the session, while your voice is still warm." Small instructions, but they change what the microphone picks up.
The cheapest fix in performance creative
Hook copy is important. It's also fixable in post: you can swap intros, test different opening lines, A/B the first frame. The hook is a creative variable.
CTA energy is different. It's captured at the moment of recording, in the creator's body, in the room where it was filmed. It can't be edited in. It has to be delivered right the first time, or re-recorded.
The good news: re-recording the last 6 seconds of an ad is the cheapest creative intervention that exists — no new shoot, no new script, no new edit. One creator, one phone, and a different energy level at the moment of the ask.
The argument is simple: how the ask is delivered plausibly shapes the response as much as the words do, and unlike the words, the delivery is measurable. Your CTA energy deserves at least the attention your hook copy already gets, and it's far easier to fix.