Day 141: Losses, Audience Quality, and Another Split-Test

Today is Monday, January 19th, and this update starts with a day I really didn’t enjoy looking at.

Yesterday slipped into loss-making territory, which is something I actively try to avoid.

Ad spend came in at around $661 against revenue of roughly $540–$550, leaving a loss of about $110 on the day. That’s not something we’ve seen often, and it’s definitely not something I want to see become a pattern.

So far today, we’re at $378 at around 4:20pm.

There’s still time left in the day, but it’s already clear that this is going to be another tough one unless things turn around fairly sharply.

What’s Actually Causing the Problem

Looking at the numbers, this isn’t a funnel or offer issue. It’s very clearly an audience quality problem driven by Facebook’s delivery.

Over the last 30 days, the average order value has been close to $44. Over the last couple of days, that’s dropped to around $35. At the same time, the conversion rate has fallen hard.

Across the 30-day view, conversions were sitting around 3%. Over the last two days, they’re closer to 1%, with virtually nothing happening on the upsells.

That tells me the ads are currently being shown to a much weaker segment of the audience. This is one of the frustrating realities of algorithmic delivery.

Sometimes you’re in front of buyers who are ready and motivated, and sometimes the system drifts into less responsive pockets of traffic.

There isn’t a lot I can do about that in the short term other than let it run and see if it corrects itself. I could pull the budget down further, and that’s still an option, but I want to see how today finishes before making another adjustment.

Sustaining losses for more than a very short period is not acceptable, but I also don’t want to panic and overcorrect.

Continuing the Split Test Rollout

Alongside managing the ads, I’m continuing to push forward with setting up split tests so they can run in the background while I’m away later this month.

Following the success of the stripped-back “anti-design” approach on the main landing page, I’ve now applied the same thinking to the abandoned cart page.

This is the page people are sent to if they start the checkout process but don’t complete their purchase. Importantly, this test is purely aesthetic. There are no changes to the copy at all.

The current abandoned cart page already converts well, so this isn’t about fixing something broken.

It’s about seeing whether the same minimalist, black-and-white approach that improved landing page conversions can squeeze out additional gains here too.

That split test is now live inside the cart abandonment retargeting flow.

What’s Left to Test

There’s now only one major page-level split test left to set up, which is the upsell page. I’ll be getting that live shortly so it can run alongside the others.

Once that’s done, most of the page-level optimisation work will be in motion. At that point, the focus has to shift away from pages and toward what happens after the sale.

Post-Purchase and Email Focus

The biggest remaining opportunity is the 30 to 45 day post-purchase window.

That’s where lifetime value is either built or left on the table.

I’ll be looking closely at how people are being warmed up after buying, how and when upsells are introduced, and whether there’s an opportunity to run a short warm-up or reactivation campaign to the existing list to give profits a bit of a lift.

The month started strongly, but there’s clearly been a lull recently. I’ll probably step back tomorrow and look at the month as a whole to get a clearer sense of whether January is just wobbling or whether it’s genuinely struggling.

For now, it’s about keeping losses contained, letting the tests run, and making decisions based on trends rather than emotions.