Day 210: How a Duplicated Ad Set Silently Destroyed My Sales.

Some days in this journey are about celebrating wins — Day 210 is not one of those days.

This is the story of a near-total sales collapse, a frantic 24-hour investigation, and a discovery inside Facebook Ads Manager that genuinely made my blood boil.

- Key Takeaway 1: Duplicating a Facebook ad set can silently shift your audience targeting to completely unintended countries.

- Key Takeaway 2: Always audit spend by country in Ads Manager — CPMs can vary wildly (£2 in the US vs £151 in South Africa in this case).

- Key Takeaway 3: A drop in sales doesn't always mean your creative or funnel is broken — sometimes it's a targeting problem hiding in plain sight.

- Key Takeaway 4: Facebook's algorithm will not self-correct when it drifts into low-converting markets — you have to manually intervene with exclusions.

The Sales Collapse That Triggered the Alarm

On the 29th of March, I logged just $143 in sales for the day. That number sent me straight into panic mode. Looking at a broader trajectory, we'd already been seeing a gradual downward trend with peaks and troughs, but this felt like a cliff edge. Something had changed — and I needed to find it fast.

Over the 28th and 29th combined, we did around $600 in total sales. The problem? Ad spend for those same two days was $1,200. That's a $600 net loss in 48 hours, and that's not a trend you can afford to ignore.

What I Changed — And What I Suspected

Just before the collapse, I had duplicated an ad set with the intention of introducing some fresh creatives. The plan was to run the new set, see it gain traction, and then shut off the original. Standard practice — or so I thought.

I'd seen this cause issues before.

Sometimes when you duplicate an ad set, Facebook's algorithm resets in ways that aren't obvious on the surface. My first suspicion was that the system had drifted into low-quality traffic markets — places where you get volume but zero purchase intent.

The Ads Manager Metrics That Confused Me

Here's what made this so frustrating initially: on the surface, the new ad set looked like it was working. The CPM was actually better than before. Click-through rates were broadly similar. There were plenty of outbound clicks and landing page views, and the cost per landing page view seemed reasonable.

But there were zero sales. A cost per purchase of £225 staring back at me. The top-level metrics were masking a serious underlying problem, which is exactly why you have to go deeper than the dashboard summary.

Breaking Down Spend by Country: The Moment Everything Became Clear

The breakthrough came when I filtered the data by country inside Ads Manager.

Looking at the date range between the 28th and 30th of March, the picture that emerged was staggering. Of approximately £450 in total spend, around £225 — half the entire budget — had been allocated to South Africa.

South Africa has never been a significant market for Guitar Coach Magazine. There's no historical purchase data there to speak of. And yet Facebook's algorithm had decided, entirely on its own, to send half my money there.

The CPM Disparity That Says It All

If the misallocation itself wasn't bad enough, the CPM data drove the point home with force. In South Africa, the CPM was running at £151. In the US — my core, proven market — the CPM was £2. That is not a typo.

I'm running an ad account with substantial purchase history. There is significant conversion data baked in. And Facebook's much-talked-about AI decided the best use of my budget was to spend it at 75 times the cost in a market that doesn't convert. It's difficult to describe just how maddening that is when you're watching it happen in real time.

The Malaysia Problem Too

South Africa wasn't the only culprit. Malaysia also appeared in the spend breakdown as another market absorbing budget without delivering results.

These are markets characterised by high traffic volume and extremely low purchase rates for a product like ours — exactly the kind of drift that can quietly hollow out your ROAS before you even notice.

Both South Africa and Malaysia are now going on the exclusion list immediately.

Performance Snapshot: Day 210

Sales (29th March): $143

Combined sales (28th–29th): ~$600

Combined ad spend (28th–29th): ~$1,200

Net loss over two days: ~$600

South Africa spend (28th–30th): ~£225 of ~£450 total

CPM South Africa: £151 | CPM United States: £2

What I'm Doing to Fix It Right Now

The immediate action is straightforward: exclude South Africa and Malaysia from the ad set targeting. It should have been in place already, and this is a reminder that whenever you duplicate an ad set, you need to re-verify every targeting parameter — including geographic exclusions — from scratch.

Beyond the exclusions, I'll be monitoring daily spend by country far more closely going forward. This isn't a one-time audit; it needs to be a regular part of the campaign review process.

The Bigger Lesson for Course Creators Running Facebook Ads

If your sales suddenly drop and your top-line metrics still look reasonable, don't assume the problem is your creative, your landing page, or your offer. Before you rebuild anything, go into Ads Manager and break your spend down by country.

You may find, as I did, that the algorithm has simply started spending your money in the wrong place.

Facebook will not tell you this is happening. It will not flag it as unusual. It is entirely on us to catch it — and the sooner you catch it, the less it costs you.

It's been a bruising couple of days, and losing $600 net in 48 hours is genuinely painful.

But the silver lining is that we now know exactly what went wrong, we have a clear fix in place, and we've added a new layer to our ongoing audit process.

Tomorrow's update should tell us whether the exclusions have done their job. Onwards.

Resources & Next Steps

Free Top 10 Split-Tests: https://www.jonathanhowkins.com/split-testing

Join the Facebook community: https://www.facebook.com/groups/coursecreatorads

Subscribe on YouTube: https://www.youtube.com/@jonathanhowkins-coursecreator?sub_confirmation=1