Before AdMagic was a product, it was a spreadsheet open at midnight, trying to explain why a client’s ad account looked great and their bank balance did not.

We did not start with code. We started with a client losing money on ads that every dashboard called a win.

The account that looked perfect

Years before AdMagic existed, we ran paid social for ourpetslife.co.uk as part of our agency work. The ROAS numbers were healthy. The click-through rates were fine. By every metric the ad platform surfaced, the campaigns were working. But when we sat down with the actual finances, margin after product cost, fulfillment, returns, and platform fees told a different story. Spend more here, and profit went down, not up.

That was not a one-off. It is the default state of paid social for most DTC and ecommerce brands. Marketing optimizes for ROAS because that is what the ad platform hands you. Finance cares about net profit, and almost nobody has built a real bridge between the two. We lived that gap firsthand, long before we ever thought about building software.

The expensive way to find out you were wrong

The other thing we learned running that agency: testing your way out of the problem is slow and costly. The industry default is something like 50 creatives, three to four weeks, and $10,000 to $20,000 in spend, and even then the result is often inconclusive. You do not find out you picked the wrong creative on day one. You find out three weeks and a real budget later, after the damage is already on the books.

We managed over $4 million in ad spend and generated more than $2.1 million in client revenue in that pre-AdMagic work. That track record is exactly why we could not unsee the pattern: good-looking accounts quietly losing money, and testing cycles too slow to catch it before the budget was gone.

Why we started building AdMagic

So we started building the tool we wished we had back then. AdMagic is built around two ideas that came directly out of that experience. The Context Engine ties every dollar of ad spend and every click back to true net profit and contribution margin, not just ROAS, so you can see the version of the account your finance team sees, not just the version your ad platform shows you. And instead of finding out a creative was wrong three weeks in, UNisal, our proprietary neuro-creative scoring model built on AdMagic’s own benchmarking and baselines, scores a creative before you ever spend a dollar on it.

None of this runs on autopilot. The AI layer proposes a recommendation with a reason and an audit trail, and a human approves before anything executes. We built it that way on purpose, because the point was never to remove judgment from the process, it was to remove the three-week, $15,000 delay before you get the information you need to use that judgment.

AdMagic is still in active MVP development. We are not claiming live customers or a finished product, we are building toward our first design partners, targeting around January 2027. What we do have is the exact problem, lived firsthand, and a clear idea of what a real fix looks like.

If the ROAS-versus-bank-account gap sounds familiar, that is the whole reason this exists. Join the AdMagic waitlist and get early access as we build toward launch.