How We Rebuilt CHOQ's Ads After Seven Years of Measuring the Wrong Thing

By Kevin Veitia

Client

CHOQ, a US men's health supplement brand sold by subscription

Channels

Meta Ads, Google Ads, Microsoft Ads

Timeframe

Audits in July and August 2026, management since August 2026

Engagement

Meta and Google audits, then full management

Goal

Cost per purchase, measured against store orders

The challenge

CHOQ sells men's health supplements by subscription, with national radio, podcast and TV behind the brand. It isn't a small company finding its feet. It had spent hundreds of thousands of dollars on ads and could prove very little of what they'd done.

By the time the founder asked us for an audit in July 2026, CHOQ had been through a revolving door of operators: about a dozen agencies over seven years, plus freelancers and in-house buyers. Each one inherited the last one's setup, added to it and moved on. What we found wasn't one bad decision. It was years of layers.

We audited Meta and Google separately. They told the same story from two angles.

Meta: five accounts for one brand

CHOQ had five Meta ad accounts. Three were spending at the same time under different operators, and two had been permanently restricted. About $900 a day was split across three pixels, three audience libraries and three learning histories, so none of them learned much. One live account was borrowing lookalike audiences built from a different account's pixel.

Over the 90 days we reviewed, the live accounts spent $13,998.81 for 60 purchases: $233.31 each, for products priced between $49 and $182. A campaign called "Scaling winners" had spent $1,765 for a single sale and was still switched on. The flagship prospecting campaign was 81% retargeting by spend, so it mostly re-bought people who already knew the brand.

And the numbers were only half true. A site-speed plugin had been set to hold back every script until a visitor scrolled or tapped, the Meta pixel and analytics included. On a test page, nothing fired for 65 seconds. About 60% of paid visitors never showed up in Meta, Google Analytics or the email platform at all. The account was being judged, and was optimizing, on the minority of visitors it could see.

Google: seven years on the wrong scoreboard

Google had a different problem with the same root. CHOQ had spent $855,436 on Google Ads since 2021, and 94% of every conversion the account had ever recorded was an analytics event called "Engaged User", worth $0.00 and set as a primary goal. Google's bidding did exactly what it was told. It found 436,541 people who would scroll a page, not people who would buy.

That made non-brand campaigns look spectacular. Searches like "ashwagandha benefits" reported conversion rates of 61 to 69%, which no supplement brand gets from cold traffic. An agency would scale them, sales wouldn't move, the agency would be let go, and the account would retreat to bidding on CHOQ's own name, the one thing that reliably sold because radio and TV had already created the demand. Then the next operator found the same spectacular numbers, and the loop started again.

Our conclusion for the founder: non-brand advertising on Google hadn't failed. It had never been measured. It's the same trap we wrote about in why we don't run link-click campaigns. Ask a platform for cheap activity and it will find you plenty.

What we did

Fixed measurement first

Nothing else matters if the account can't see its own customers. We specified the tracking fixes and CHOQ's developer built them: the script delay came off the landing pages, a duplicate pixel that was double-counting purchases came out, and paid traffic moved to fast, standalone landing pages. On the new pages, the share of ad clicks that registered as a landing-page view went from about 40% to 76%.

On Google, the plan demoted "Engaged User" so that purchases became the only goal bidding chases, and every result is now checked against real store orders rather than the platform's own count.

One account, one pixel, one structure

On Meta, everything moved into a single account with one pixel, one audience library and three campaigns: prospecting, retargeting sized to the warm audience that actually exists, and a creative lab where new ads prove themselves before they get budget. We cut retargeting that wasn't adding anything and switched off placements and campaign types that had never sold, like Google's Demand Gen.

On Google, branded search went into one campaign, which ended four campaigns bidding against each other on the word "choq". Non-brand got its first honest test, on high-intent searches only. Microsoft Ads came next, built off the Google structure.

Creative with a system, inside the rules

The old Meta setup had about 53 live ads leaning on one duplicated video hook, with no naming convention and no rule for when to stop. Now every ad is tagged by concept and hook, and any ad that crosses a set cost per purchase gets switched off.

Supplements are one of the most closely policed categories on Meta and Google. Health claims get ads rejected, and enough rejections get accounts restricted. So every creator script and every ad goes through a compliance pass before launch: hedged language, no promised outcomes, nothing the product page can't back up. When Google did reject ads, the cause was usually the page they pointed to rather than the ad itself, so each product got a landing page built for paid traffic.

The creative that won wasn't the flashiest. It was product-led creator video: a real person, the actual product, plain language. In the first weeks, one product-led ad brought in purchases at about $21 each, the lowest cost the account had recorded.

Results

The founder hired us to run the accounts straight after the audit and handed them over within days.

• Blended cost per purchase across Meta, Google and Microsoft of about $80 in the first month and about $120 in the second, against a $300 ceiling. Before we took over, Meta alone was running at $233.
• One product-led ad at about $21 per purchase, the lowest the account has recorded
• No account restrictions since we took over, and only a handful of rejected ads, each fixed by changing the copy or the landing page
• Meta, Google and Microsoft Ads all running within target
• Two months in, the brief has changed from "make it work" to "scale it"

Cost per purchase rose in month two because the budget went up. That's what scaling does: each extra dollar reaches people a little less ready to buy. Well under the ceiling, there's room to keep going.

What you can take from this

• If most of your visitors are invisible to your tracking, every number in the account is fiction. Fix measurement before you judge anything.
• A 61% conversion rate on cold traffic isn't a win. It's a sign you're counting the wrong thing.
• Five ad accounts don't give you five chances. They split the learning five ways.
• In supplements, compliance belongs in the creative brief, not in a review at the end. Write inside the rules and the account stays alive long enough to scale.
• When cost per purchase rises as you scale, compare it with last month and your ceiling, not with your best-ever week.

Want a second set of eyes on your account? Book a free 30-minute consultation and we'll walk through what's working, what isn't, and where the next win is. Curious what it costs first? Our pricing is public.