🤖 Here’s what QuillBot's $3.2M creative test can teach you

Here are the three hooks, and what QuillBot ultimately spent on each:

Ad 1

"This AI humanizer tool helps me write like me." 

$1.6M total spend, $433k in the last 30 days.

Ad 2

"Nobody teaches you how to rewrite AI text so that it actually sounds like you."

$892k total spend, $158k in the last 30 days.

Ad 3

"Okay, your robotic writing is bad for business." 

$710k total spend, $114k in the last 30 days.

The gap isn't subtle. Ad 1 has generated more spend than ads 2 and 3 combined.

And the 30-day numbers suggest the gap is widening - ad 1 is being scaled up while the other two are quietly being wound down.

The counterintuitive result

Here's what makes this interesting for anyone running DR ads: the most aggressive hook lost.

Conventional DR wisdom says lead with pain. Poke the bruise. Create urgency.
Ad 3 does exactly that - "your robotic writing is bad for business" is a direct threat, designed to trigger loss aversion. Ad 2 uses a curiosity gap, another reliable hook format. Both are legitimate, proven approaches.

Ad 1 does neither. It's calm, product-confident, and almost matter-of-fact: "This AI humanizer tool helps me write like me." 
No threat, no gap, no manufactured urgency. Just a straightforward product claim delivered with quiet confidence.

And it outspent the fear-based hook by more than 2:1.

There are a few ways to interpret this.

The most likely explanation is audience temperature. 

QuillBot is targeting people who are already using AI to write - they don't need to be scared into awareness of the problem. They're living it.
For a warm, problem-aware audience, a calm product demonstration is more persuasive than a threat, because the threat feels redundant. They already know their writing sounds robotic. What they want is the solution, not more anxiety about the problem.

The fear hook works best when you need to create urgency in someone who doesn't yet feel it. When your audience already feels it, showing them the exit is more powerful than describing the fire.

Breaking down the body: a script that earns every second

Once the hook lands, all three ads run the same 23-second body. Here's what's happening beat by beat:

0:03 - "So let me show you how. Here's a sentence I got from AI." 

Within 3 seconds of the hook, they're already showing the product. No story padding, no brand history, no feature list. The demo starts immediately.

0:05 - "And here's what it looks after using QuillBot's humanizer. Same message, way more me."

The before/after is the entire argument. The viewer doesn't need to be convinced intellectually — they can see it. This is the highest-conversion format in software ads because it converts skepticism without requiring trust.

0:10 - "Literally a humanizer for your writing."

This line functions as a category definition the viewer can remember and repeat. QuillBot isn't just selling a feature — they're naming a new type of tool and planting the flag.

0:12 - "And QuillBot isn't just a writing tool. It's how I protect my tone and credibility at work."

This is the sharpest reframe in the script. It moves the purchase from a productivity nice-to-have to a professional reputation safeguard. That's a much stronger buying motivation — and a much higher perceived value.

0:16 - "If you're using AI but still sound awkward, QuillBot fixes that."

This validates the viewer's existing frustration before offering the fix. It makes the CTA feel earned rather than pushed.

0:19 - "Try QuillBot Premium to get access to the humanizer, plus all of the must-have features for less than $9."

"Less than $9" is deliberately vague-yet-specific. It anchors low without committing to an exact number, and it lands at the very end - a classic direct response move to reduce perceived risk at the moment of decision.

Why a real person matters here

The creator is not a tech reviewer. She's positioned as a regular professional who uses AI and cares about how she comes across.

This is a deliberate casting choice - QuillBot's buyer isn't a developer, it's someone in marketing, HR, content, or any knowledge-work role who's using ChatGPT but still second-guessing the output.

The bedroom-and-desk setting reinforces this. It reads as authentic, not produced. And that's not an accident - it's the visual equivalent of the script's central promise. When your product's core claim is "write like yourself," your ad cannot look like it was made by a committee in a film studio.

Every production choice is a signal. QuillBot got that right.

What the spend curve tells us

All three ads launched on the same date and started with similar spend levels.

The divergence didn't happen immediately - it built gradually over the first few weeks as performance data came in and budgets were reallocated. This is normal for a properly run hook test: you start even, let the data accumulate, then shift weight toward the winner.

What's notable is that QuillBot didn't kill the worst-performing ads immediately.

Ads 2 and 3 are still running at meaningful spend ($158k and $114k in the last 30 days, respectively).

This is smart for a few reasons.

First, different hooks reach different people at different scroll moments - some viewers respond better to threat framing depending on their mood or context.

Second, running multiple creatives protects you from algorithm fatigue on any single ad.

Third, the data from the worst-performing ads is still valuable - it tells you which audiences to stop targeting with that angle, which informs future creative decisions.

But the direction of travel is clear. The budget is consolidating around ad 1, and if the trend continues, the other two might be cut within the next few months.

Takeaways 

1. Match your hook to your audience's awareness level. 
   
   Fear hooks work on cold audiences who don't yet feel the pain. Product hooks work on warm audiences who already do. QuillBot's audience is already using AI and is already frustrated with the output - they needed a solution, not more diagnosis.
   

2. Isolate your hook variable. 
   
   QuillBot changed nothing except the first 3 seconds. Same script, same creator, same publish date. If you're testing hooks, this is how you do it - everything else held constant so the data is clean.
   

3. Name a category, not just a product.
   
   "A humanizer for your writing" is a concept, not a feature. If you can make viewers feel like they've discovered a new category of tool, they'll remember you when they're ready to buy.
   

4. Move from pain to identity.
   
   The sharpest line in this script is "it's how I protect my tone and credibility at work." That's not a product benefit - it's a self-concept. People pay a premium to protect how they see themselves professionally.
   

5. Let the demo do the heavy lifting.
   
   A before/after is the highest-conversion format in software ads because it bypasses skepticism. If your product has a visible transformation, show it in the first 5 seconds and let the viewer's brain close the sale.
   

6. Don't kill your losing variants too fast.
   
   Ads 2 and 3 are still generating hundreds of thousands in spend. The data they produce - which audiences convert, at what frequency, on which platforms - feeds future creative decisions. A losing hook isn't worthless. It's research.

‘Til next time!

💌 Like this newsletter? Let's continue the conversation

Get in touch with us by responding to this email, tagging us on LinkedIn or Instagram, and sharing your thoughts. Your feedback helps us keep our newsletter relevant and interesting.