The $7 AI Ad That Spent $70K: How One Podcast-Style Video Changed Everything We Knew About Creative Production

By Kamal Razzak, Founder of MHI Media | February 2026

In November 2024, we uploaded an ad that cost $7 to make.

By February 2025, that ad had spent $70,000 in media budget, generated 1,400 website purchases at a $50 CPA, and was still running profitably.

No actors. No studio. No production crew. One person, a laptop, and an AI toolstack that costs $424/month.

This is the story of how that ad happened — and why it convinced me that the economics of creative production have permanently changed.

The Problem We Were Trying to Solve

> Traditional video ad production costs $5K-$15K per asset, making systematic creative testing financially impossible for most DTC brands.

Here's the dirty secret of performance marketing: the creative bottleneck kills more scaling efforts than bad targeting, weak offers, or platform changes combined.

At MHI Media, we manage ad accounts for 200+ DTC brands. We'd built systems, hired media buyers, developed frameworks — but we kept hitting the same wall. Our clients needed 10–20 new creatives per week at scale. Traditional UGC production gave us 5–10 per month.

The math didn't work.

A single UGC video costs $200–$500 and takes 2–4 weeks from brief to delivery. A brand spending $100K/month on Meta needs to test constantly because creative fatigue hits every 10–14 days. By the time the next batch of UGC arrived, the current winners were already declining.

We were always behind. And so was every other agency we talked to.

The Podcast Format Insight

> AI-generated podcast-style ads feel like organic content, bypassing ad fatigue because listeners perceive them as entertainment rather than advertising.

Before the AI breakthrough, we'd already noticed something interesting in our data: podcast-style ads were crushing it.

Not actual podcast clips — ads designed to look like podcast conversations. Two people talking about a product, a problem, an industry. No hard sell. Just a conversation that happened to be about something the viewer cared about.

Why do podcast-style ads work?

    • They match consumption behavior. People spend hours listening to podcasts. The format signals "this is content worth my time" rather than "this is an ad interrupting my feed."
    • They feel native. A podcast clip in your feed looks like organic content. It triggers curiosity rather than ad avoidance.
    • They allow depth. A 60-second conversation can cover objections, benefits, social proof, and CTA — naturally, without feeling scripted.
    • They build trust. The podcast format implies authority. If someone is being interviewed about a topic, they must be an expert. That's the subconscious assumption.
The problem? Producing podcast-style ads traditionally meant coordinating two people, scheduling recording time, renting studio space, editing, post-production. $2,000–$5,000 per video, 2–3 weeks turnaround.

At that cost and timeline, we couldn't test enough variations to find winners consistently.

The Experiment

> MHI Media produced an AI podcast ad for $7 using synthetic voices and minimal editing, then tested it against $15K professional video creative.

In late 2024, AI voice and video generation crossed a threshold. The output wasn't perfect, but it was good enough — good enough to stop the scroll, hold attention, and convert.

I decided to test a hypothesis: could we produce podcast-style ads with AI at a fraction of the cost, and would they actually perform?

The Setup

Client: A health and wellness DTC brand (supplements category) Challenge: Needed fresh creative to scale past $50K/month. UGC pipeline was maxed out. Goal: Test AI-generated creative as a viable alternative.

The Production Process

Here's what creating that first AI podcast ad actually looked like:

Step 1: Script (30 minutes) I wrote a conversational script between two people discussing the client's product category — not a hard pitch, but a genuine conversation about the problem the product solves. The script used voice-of-customer language pulled from reviews and support tickets.

Key script principles:

Step 2: AI Voice Generation (15 minutes) Using AI voice tools, I generated two distinct voices — one male, one female — reading their parts of the conversation. The voices sounded natural, conversational, with appropriate pauses and emphasis.

Total voice generation cost: approximately $2.

Step 3: AI Video Generation (20 minutes) Generated podcast-style video — two people sitting at a table/studio setup, talking. The visuals match what you'd expect from a real podcast clip. Lip sync wasn't perfect, but for a social media ad viewed on a phone screen, it was convincing enough.

Video generation cost: approximately $4.

Step 4: Post-Production (15 minutes) Basic editing — added captions (critical for sound-off viewers), adjusted pacing, added the brand's product imagery as B-roll during the product mention, included a subtle end card with a CTA. Step 5: Upload and Launch Uploaded to Meta as a standard video ad. No special settings. Broad targeting via Advantage+. Let the algorithm do its thing. Total production time: ~80 minutes Total production cost: ~$7

The Results

> The $7 AI ad generated $70K in revenue, 1,400 purchases, and outperformed the professional creative on every metric including ROAS and CPA.

The first 48 hours were unremarkable. The ad spent $200 with mediocre results. We almost killed it.

Then something shifted. The algorithm found its audience.

Week 1

Promising but not exceptional. We let it run.

Week 2–3

Performance stabilized. CPA held steady even as spend increased. This was the signal — the ad had legs.

Month 2 (December 2024)

December CPMs spike for everyone. The fact that CPA decreased while CPMs increased meant the ad's conversion rate was improving as the algorithm optimized delivery.

Month 3 (January 2025)

Still running strong. January CPMs dropped (seasonal), and the ad continued to convert efficiently.

Month 4 (February 2025)

Performance finally began declining. Creative fatigue was setting in after nearly 4 months — an exceptionally long lifespan for any ad creative.

Cumulative Results

Why This Changes Everything

> At $7 per ad, brands can test 100 creative concepts for the cost of one traditional video, fundamentally changing the economics of creative testing.

That $7 ad didn't just perform well. It reframed how we think about creative production entirely.

The Old Model

Brief → Source creators → Negotiate → Ship product → Wait for delivery →
Review content → Request revisions → Wait again → Receive final →
Upload → Test → Hope it works

Timeline: 2–4 weeks
Cost: $200–$500 per video
Output: 5–10 videos/month

The New Model

Write script → Generate voices → Generate video → Edit → Upload → Test

Timeline: Same day
Cost: ~$7 per video
Output: 60+ videos/month (1 person)

The implications ripple through the entire performance marketing stack:

1. Testing velocity explodes. When creative costs $7 instead of $350, you can test 50x more variations. More tests = more winners found = faster scaling. We went from finding 1–2 winning creatives per month per client to finding 4–6. 2. Creative fatigue becomes manageable. When you can produce 60+ ads per month, the 10–14 day fatigue cycle stops being a crisis. You always have fresh creative in the pipeline. The treadmill becomes walkable. 3. The barrier to entry drops. Small DTC brands spending $10K/month can now access the same creative volume as brands spending $200K/month. Creative production is no longer a function of budget — it's a function of skill and systems. 4. Iteration speed transforms strategy. In the old model, testing a new angle meant a 2–4 week production cycle. Now you can test a new angle in hours. "I wonder if a podcast ad about ingredient sourcing would work" goes from hypothesis to live test the same day.

The Craft Layer: Why This Isn't "Easy"

> AI tools are available to everyone, but knowing what hooks convert, what formats work, and how to optimize requires deep performance marketing expertise.

I want to be honest about something: producing AI ads that actually perform is harder than it looks.

The technology is accessible. Anyone can generate an AI voice and video. But the gap between "AI-generated content" and "AI-generated content that converts" is enormous.

Here's what separates our AI ads from the ones that flop:

1. Script Quality Still Matters Most

AI handles production, not strategy. The script — the words, the structure, the hooks, the emotional beats — is still 80% of what determines performance. A brilliantly produced AI video with a bad script will fail. A roughly produced AI video with a great script will win.

Our scripts work because they're informed by $50M+ in ad spend data. We know which hooks stop the scroll, which objections need addressing, which emotional triggers drive conversion. AI doesn't know any of that.

2. Authenticity Engineering

The biggest risk with AI creative is the "uncanny valley" — content that looks almost real but feels off. We've developed specific techniques to avoid this:

3. Performance Marketing Knowledge

Knowing how to make an AI video is a production skill. Knowing how to make an AI video that converts is a performance marketing skill. You need to understand: The craft layer — knowing what looks real vs. what doesn't, knowing which scripts convert vs. which just sound good, knowing how to engineer authenticity into AI content — is where the actual value sits.

What We've Built Since

> MHI Media has since produced 500+ AI ads across 50+ brands, achieving an average 3.8x ROAS with production costs under $50 per asset.

That first $7 ad was proof of concept. Since then, we've scaled AI creative production across our entire client portfolio:

We've also expanded beyond podcast format into AI-generated: Each format follows the same principle: AI handles production, humans handle strategy. The technology is the tool; the taste and experience is the differentiator.

SaruGeneral: When AI Creative Meets Aggressive Scaling

> SaruGeneral scaled from $5K to $50K daily ad spend in 60 days using AI-produced creative, proving the format works at scale.

One of our most dramatic case studies combines AI creative production with our broader scaling methodology.

SaruGeneral, a DTC brand we work with, scaled to $100K days using a combination of founder-led creative, AI-generated ads, and systematic scaling frameworks.

The AI creative component was critical to maintaining the creative volume needed at that spend level. At $100K/day, you're burning through creative fast. Without the ability to produce 20+ new assets per week, scaling to that level would require a creative team 5–10x larger.

The Future of Creative Production

> Within 12 months, AI-produced creative will become the default for DTC brands, with human expertise focused on strategy rather than production.

I believe we're in the first inning of AI-generated creative for performance marketing. Here's what's coming:

Near-term (2026): Medium-term (2027–2028): What won't change:

What This Means for You

> Any DTC brand can start testing AI-produced ads today with minimal investment, but results depend on strategic expertise, not just the tools.

If you're a DTC founder or marketer reading this, here's the practical takeaway:

The cost of creative production is no longer a valid excuse for not testing enough.

$7 per ad. Same-day turnaround. 60+ variations per month with a single person.

The brands that will win in 2026 and beyond aren't the ones with the biggest production budgets. They're the ones with the best systems for producing, testing, and iterating creative at speed.

That $7 ad didn't just generate 1,400 purchases. It proved that the future of performance creative isn't about spending more — it's about moving faster.

FAQ

How do AI podcast ads compare to real podcast ads?

In our testing, AI podcast ads perform within 5–10% of ads using real podcast footage, at 1/30th the cost. The gap is closing as AI quality improves, and for most social media contexts (mobile viewing, sound often off with captions), the difference is negligible.

What AI tools do you use to create podcast-style ads?

Our full stack costs $424/month and includes AI voice generation, video generation, and editing tools. We don't publicly share the exact stack because it changes frequently as better tools emerge, but the principle is consistent: voice + video + editing, all AI-assisted.

Can any brand use AI-generated ads?

Yes, with caveats. AI creative works best for brands where the ad format is native-feeling (social media, podcast clips, testimonial style). It works less well for brands requiring specific product demonstrations or highly regulated industries where accuracy of visual claims matters.

Won't consumers get tired of AI-generated content?

Consumers don't care whether content is AI-generated or human-created — they care whether it's interesting, relevant, and trustworthy. The $7 ad worked not because viewers couldn't tell it was AI, but because the conversation was genuinely engaging and the product solved a real problem.

How do I get started with AI ad production?

Start with scripts. Before investing in any AI tools, write 10 podcast-style conversation scripts for your product. Test the best ones as static text or audio-only first. Once you have a winning script, then produce the AI video version. The script is the hard part — the production is the easy part.

What's MHI Media's role in AI creative production?

We handle the full pipeline for our clients: strategy, scripting, AI production, testing, and optimization. The craft layer — knowing what converts, what looks authentic, what hooks work — is our core value. The AI tools are available to everyone; the expertise to use them effectively is not.

Frequently Asked Questions

How much does it cost to produce an AI podcast ad?

AI podcast-style ads can be produced for as little as $7-$50 per asset using synthetic voice tools and minimal editing. The primary investment is in strategic expertise — knowing what messaging, hooks, and formats convert — rather than production costs. MHI Media produces AI ads at scale for DTC brands starting at $500 per month.

Do AI-generated ads perform as well as professional video?

Yes. MHI Media's data shows AI-produced ads match or outperform professional video on key metrics including ROAS, CPA, and conversion rate. The $7 AI ad in this case study outperformed a $15K professional video. The key advantage is testing velocity — you can produce and test 100x more creative variations.

What platforms work best for AI podcast-style ads?

Meta (Facebook and Instagram) delivers the strongest results for AI podcast ads, particularly in feed and Reels placements. TikTok is growing rapidly for this format. The podcast style works because it mimics organic content, reducing ad fatigue and increasing engagement rates across all platforms.

Can any brand use AI-produced creative?

Any DTC brand can benefit from AI-produced creative, but results depend heavily on strategic execution. The tools are accessible to everyone, but understanding what converts — hooks, messaging angles, audience psychology — requires deep performance marketing expertise. Start with proven frameworks before experimenting.

About MHI Media

MHI Media is a London-based DTC performance marketing agency specializing in founder-led creative. Founded in 2020, MHI Media has helped 200+ ecommerce brands scale through data-driven paid media, creative strategy, and performance content.

Kamal Razzak is the founder of MHI Media, a performance marketing agency that has managed $50M+ in ad spend across 200+ DTC brands. MHI Media specializes in founder-led creative and AI-powered ad production for direct-to-consumer brands.