Iron Gate Knives

Where they started

Iron Gate Knives is a small Australian knife maker. Izac Yeaman runs the workshop, drops new knives every Thursday night, and builds the rest to order. His content did the hard part. It made Meta traffic cheap to buy. Converting that attention was the problem.

A lot of that traffic landed on products nobody could buy. Most of what the best-seller page showed could not be bought. On the drop page, the main call to action was to join the knifedrop email list. Clicking it did nothing. The page just refreshed.

The ad account had the same shape. The highest-spend Meta campaign in the 30 days before we started was a boosted Instagram post. It produced around 28,000 link clicks and not one purchase. Dozens more campaigns were set to chase visits rather than sales, on mixed attribution settings.

Google was a closed door. The platform had rejected his ads under its weapons policy, and he had stopped trying. He was running all of it himself, late at night, after a day in the workshop. And he did not trust the numbers he was reading.

"I don't really know what's true and what's not." - Izac Yeaman, February 2026

The constraint

The audit ranked one constraint above the rest. Paid traffic was being sent to products that were already gone, and nothing in the account was built to prevent it. It named this in February, and the account has run on that diagnosis since March. The damage ran in a loop. A good ad earns the click. The click lands on a sold-out page. No purchase follows, and the platform concludes the ad failed. So the ads doing their job looked exactly like the ads that failed.

It also refused the easy diagnosis, in the audit's own words: "You don't have a 'marketing' problem. You have a demand capture + system design problem". We could not stop a knife from selling out. We could stop the account from paying for clicks instead of customers. So the work started with the ad account and with the way the business measures itself.

What changed

In March 2026 Izac stopped buying his own media and we took the account over. The first thing to go was its best-funded campaign, the boosted post with 28,000 clicks and no purchases. The click and landing-page-view objectives went with it. The sales lanes were rebuilt to buy purchases, so the budget behind them stopped paying for clicks that could not convert.

Google restarted the same month. Our team wrote a keyword set around brand searches and everyday kitchen and fishing terms. No tactical framing. That kept the ads inside the policy that had rejected him. By August the channel he had given up on was carrying about half the account's ad spend.

The third change was how the account gets judged. Shopify net sales are the truth. New-customer cost and payback are the decision metric. Platform return is directional only. That rule is why the return figure below is Shopify's, not a platform's.

The Meta rebuild and the Google build went live in the first month; the reporting standard followed. The audit also called for store-side work, and that is where the record stops. We hold no artifact confirming it went live, and no one has re-measured the store's conversion rate since February, so we will not claim either.

What happened

In the four weeks to 2 August the account brought in 36 new customers a week, at about A$139 each and 6.96X blended. In the four weeks to 31 August it brought in 47 a week, at about A$87 and 9.39X. Cheaper customers, more of them, on a smaller weekly budget. Both windows sit inside the engagement, so this is month against month.

The budget shrank because Izac shrank it. He had cut pre-orders back to five or six SKUs - they were eating the workshop's capacity - and set spend to match what he could make. So the number that matters here is efficiency, not size.

The earlier window overlaps a sale the sheet flags, which flatters the baseline, so the gap is more likely understated than overstated.

August as a whole month: about A$161,000 in revenue on about A$18,000 of ad spend. It brought 200 new customers at about A$90 each. Nearly two in three orders came from someone who had never bought before. Those are Shopify's figures.

The best week of the engagement came in late June: about A$83,000 in revenue on about A$6,500 of spend.

The thing capping this business at the start was its storefront. It could not sell to the people already standing in front of it. We cannot tell you that problem is fixed. We can tell you what caps it now.

"our drops been selling out in 20 minutes" - Izac Yeaman, August 2026

That is the same mechanism we named in February, running for a different reason. In February the store could not capture the demand. Now the demand is outrunning one man's bench - a queue outgrowing a workshop, not proof the storefront is fixed. On his own forecast, not our result, he is planning drops of up to 1,000 knives in October and November, up from 20 to 50. Six months ago the account was paying to send buyers to pages with nothing left on them. The question now is how many he can make.

"When we've got stock, we can... accelerate the ad spend because we've got a really healthy ROAS. When we don't have stock, the ROAS falls out of the sky... I don't like sending people to a website where the first brand experience is I can't buy and I have to wait months and everything's out of stock."

Verify this case study

• The audit itself. Twelve pages, dated 19 February 2026, with the diagnosis, the numbers and the plan in it. We will walk you through the diagnosis and the financial pages on your call.
• Every performance figure above comes from one weekly reporting sheet that holds the weeks before we started and every week since. The before-state details come from the audit and from recorded calls. It is Izac's account data, so it stays with him. Ask on the call and we will tell you where each number came from. Figures from his account are in Australian dollars. Our own figures are US dollars: the $5,000 audit fee, the $75K threshold and the $30K a month spend bar.
• Revenue and new-customer figures are Shopify's. Meta and Google each count their own slice. Added together, they report more revenue than the store actually took. That is why we do not run on them.
• What is missing, and one disclosure. There is no recorded interview with Izac yet, and no public review. He has agreed to record one twice, and it has not happened. And he came to us in February 2026, before we set today's bar of $30K a month in ad spend. His account still runs below it.