Day 76: The Hidden Cost of Split-Testing

It’s Saturday, November 15th — Day 76 — and today I’m reflecting on one of the realities of digital marketing: split testing always comes with a cost.

Sometimes that cost is monetary, sometimes it’s time, and sometimes it’s clarity — but it’s always the price of progress.

Daily Numbers

Yesterday was rough:

• Sales: $360

• Ad Spend: £363 (≈ $477)

• Profit/Loss: -$117

That’s officially my biggest single-day loss since starting this $240K journey. It doesn’t feel great — but that’s the nature of testing. You can’t find the breakthroughs without paying for the experiments that don’t work.

Today, as of 4 p.m., sales are sitting at $289, so I’m hoping for a late recovery to push back into profit before day’s end.

Google Ads: Still Waiting for Movement

The Google Demand Generation campaign is live, but nothing has started yet — no impressions, no clicks, no data.

The campaign is marked “Learning” with five days remaining in that phase, so it’s probably still being processed by the algorithm.

Google’s ad delivery system spans across YouTube, Display, and Discovery — which makes it harder to know exactly where ads will appear or how long it will take to gain traction.

At this stage, patience is the only play. I’ll give it the full week to see whether it starts moving before I make any changes.

Split Test Reflection: When “Three” Performs Worse Than “One”

I’ve been running a split test on my Classic Riffs funnel, comparing:

• Version A (Control): One “Muscle Memory” bundle of 42 riffs.

• Version B (Test): Three separate course volumes of 14 riffs each.

Here’s what happened after just three days:

• Control (One Bundle): ~3% conversion rate.

• Test (Three Volumes): ~1% conversion rate.

That’s a huge drop — and a clear example of how split tests can reduce short-term profit while you search for long-term gains.

But here’s the real lesson: I think the test itself was flawed.

It wasn’t just one box versus three boxes — it was one strong proposition versus one weak one.

The original page was positioned around “Muscle Memory Training: The Secret to Effortless Playing.”

It was specific, emotional, and benefit-driven.

The three-volume version, however, simply said “Classic Riffs” — which lacked any real hook or clear promise.

So, I wasn’t really testing structure or perception of value — I was testing a great headline versus a weak one.

The Plan: A Fairer Test

I’m going to pause the current split test and redesign it properly:

• Keep the “Muscle Memory” theme and emotional proposition.

• Retain that strong headline and benefits.

• Simply change the visual presentation — showing three volumes instead of one.

That will give me a true test of perceived value without altering the core message.

I’ll aim to rebuild those creatives tomorrow or Monday, depending on time.

Once live, I’ll run both versions for a full week to gather cleaner data.

The Takeaway

Split testing is a double-edged sword. Every new test introduces instability — ad sets go back into learning, CTRs fluctuate, and conversions can dip.

But this is the price of getting better.

You can’t A/B test your way to breakthroughs without enduring a few losing days.

The important thing is to learn from every test, refine fast, and make sure your next iteration isolates the right variable.