- The Method
- Posts
- ⚙️ Turn Testimonials Into Ads
⚙️ Turn Testimonials Into Ads
+ Read this or keep being misunderstood
Generate AI Influencer Ads From Text
If you are in the market for user-generated content (UGC) ads, it might be worth using AI to create your own.

A screenshot of the videos we’ve generated using Google’s Flow that we’re now running as ads
We’ve been using Google’s latest video tool, Flow, to create hooks, testimonials, and body content for our clients' UGC ads.
And the results are good.

Early campaign data from our AI ads
Here’s how to create AI ads and not feel like you are deceiving your customer:
(Bonus tip: If you want to see how to use AI to write your entire ad script - read this)
1. Head over to Google AI plans
2. Select the Google Ultra AI Plan ($173.75 if you are already a Google Workspace user)

3. Once you’ve signed up, head to Google Flow > select “Create with Flow”
4. Select “Create New Project”
5. You’ll then be given the prompt box - change the settings to “Text to Video” and change the settings to “Veo3 - Fast (Text to Video)” and select 1 “Outputs per prompt”

Screenshot of the settings
Google gives you 12,500 monthly credits, and every video output you produce costs 20 credits (roughly 625 videos - you’ll spend a lot of credits trying to get your prompting right for the content you are after)
6. In terms of prompts, here is the best format:

Example prompt given to Flow
Outline the video output type first: “4×5 vertical video”
Choose the setting: “podcast interview style”
Tell Flow the details of the person: “Feature a professional-looking European female, late 30s”
Tell it the shot: “Shot from the shoulders up”
Tell it the dialogue: “They made getting the cash we needed easy.”
7. Here is the output:

Voice and video are indistinguishable from normal creators, albeit the eyes are sometimes creepy
8. To turn these into ads we feel happy to advertise, we’re placing a watermark saying, “These are real customer testimonials using AI actors.”

Don’t let the AI write the subtitles, add them in manually in your editing
From our testing with the 12500 credits, you can get 20 or so ads, each 25 - 40 seconds long.
This works out to be about $8.70 an ad in terms of AI.
Add a short-form video editor to create a high-quality ad that converts for roughly $40.
Pricing
We're addressing our pricing problem at Compound.
We're at capacity and have a ton of members paying old low rates. More than 1/6th of them pay less than what it costs us to serve them and we have another 1/3rd who are floating around our break-even mark.
That's what happens in capacity constrained subscription businesses dealing with inflation, as your break-even amount increases you lose money on an increasing amount of your customers.
We'll do the standard stuff like lift our minimum rate for all members to at least our break-even rate.
But given we're at capacity I want to introduce a few pricing mechanisms to better adapt our pricing to meet our supply problem.
Dynamic pricing seems the obvious solution. As demand increases so does your price. For us that means making sure our members who train at peak times, are paying the most and visa-versa.
But another idea I've been toying with is making sure that our peak-time members are also our most committed.
Members on our annual contracts are worth 3x what our members on cancel-anytime agreements are worth.
Traditionally, gyms handle this disparity by making the non-committal memberships more expensive, but even a 30% price premium doesn't close that gap much.
So for a start we're testing only offering our peak training times to members willing to commit to an annual contract. Another option we have is increasing our minimum term on our memberships.
Losing long-term members because we're too busy with short-term members is poor business.
With these adjustments, we hope to make our pricing slightly more intelligent and better reflect the value of training at certain times in our gym.