SpeedCubeShop

Where they started

By October 2025, Cameron Brown was still signing off the ad budget every month. He no longer believed what it was telling him.

SpeedCubeShop has sold speedcubes and puzzles since 2009. The average order is about $50. The audit put it plainly. We were not going to play the lifetime-value game with a product like this. Cameron's own standard is that a new customer has to be profitable on the first order. Every wasted dollar in the ad account shows up in the bank account the same month.

Two agencies had already come and gone by the time we first spoke. He had not gone looking for a third. The introduction came through a mutual contact in October 2025. He describes the years before as "this just sinking feeling of, yeah, we're spending all this money on ads, but it's not making anything." On camera he goes further: "As far as the paid ads go, I had a lot of confidence problems in that realm specifically... from that point on, my confidence was just completely shaken."

The numbers were going the same way. Store revenue fell about 14% from 2024 to 2025, on almost exactly the same ad spend. The audit was written on month-to-date figures. In them, new-customer revenue was down 16%. Revenue from existing customers was down 9%. New-customer acquisition cost had peaked at $16.55 in May 2025, one month taken from before the engagement began. Blended return on ad spend that month was 5.99x.

Google Ads spend had just been pulled from about $43,000 in August 2025 to about $19,000 in September. Nobody could say what that cut had saved or cost. Nobody could say which part of the account was working either. The platforms, the dashboards and the agency reporting each told a different story.

The constraint

The ceiling was not product demand. It was that nobody in the business could tell who was actually a new customer. Until that was fixed, nothing else in the account could be judged honestly. That included whether the offer was the real problem.

The account could not separate a new customer from a returning one. It could not separate either of them from someone who had just typed the brand name into Google. Every mechanism for that confusion was sitting in the accounts, in writing.

About 20% of the main Performance Max campaign's spend was going to branded search terms. Separate branded campaigns were already running, and that one campaign outspent them on brand. The Google account was set to credit clicks up to 90 days before a purchase, on an order of about $50. Its customer match list had not been updated in over 100 days. On Meta, about 25% of spend was reaching existing customers, on top of a separate remarketing campaign.

There was also no break-even ROAS defined anywhere in the business. There was no line to be above or below, so nothing could ever tell him to stop.

Every one of those was chosen by somebody, or never defined at all. Together they made one problem. Nobody could see who was actually new. So nobody could tell whether Cameron's rule was being kept.

What changed

The first job was to build the measurement the account had never had. A weekly reporting stack, a profitability sheet and a monthly scorecard. Delivered every Monday before 3pm, holding 2024, 2025 and 2026 side by side. Two lines on it matter most here. One is what a new customer costs. The other is the share of orders coming from people who had never bought before, which nothing in the account had counted.

Then a break-even chart. It shows, for each level of monthly spend, what return that spend has to produce before the business makes money. One rule runs it. Stay a full point of return above break-even and there is room to keep scaling. That chart is now the tool used live on calls with Cameron.

Cameron's own account of the change is in the interview: "You can see the ad spend is following revenue, which has been a really simple concept that I've been wanting for years."

On Google we split brand, non-brand and retargeting into separate campaigns. Each could then be judged on its own, instead of one line covering all three. We built dedicated new-customer campaigns and excluded existing customers from them. On Meta, spend moved into small incremental campaigns. In May 2026, at about $700 a week, those campaigns were buying new customers between $4.40 and $11.91.

Then the part that only works once the reporting exists. The budget came down. The rule we set on the audit call in October 2025 was simple. If revenue is falling, the ad budget falls with it. Across weeks 1 to 35 of 2026, spend ran 27% below the same weeks of 2025.

What happened

The work started in January 2026. Take August 2025 against August 2026, one month a year apart.

New customers 28% cheaper, $15.30 to $10.97.

Total return on ad spend was 9.41x, up from 6.87x. That is measured across the whole business, not one platform's claim.

57.3% of orders came from people who had never bought before, up from 51.3%. That is a mix shift toward first-time buyers, not the existing list being sold to again.

Every dollar spent to win a new customer now brings back $2.90 in gross profit over 90 days. A year earlier it was $2.10. That runs on a modelled product cost of 41%. Recent confirmed months have come in higher than that. The latest actuals are still outstanding, so read the ratio as directional.

August was not a lucky month. Every month of 2026 beat its 2025 counterpart on new-customer cost. The January to August average went from $13.59 to $9.30.

Across weeks 1 to 35, ad spend went from $429,057 to $311,306. Now look at what revenue did. It fell 4% over those same weeks. A year earlier, the same 35 weeks had fallen 8%. So the decline was already there, and it slowed while the budget came down.

Then look at the back half. Across weeks 23 to 35, revenue turned positive, up 4.1% on 12.6% less spend. That is the decline reversing, and it happened on less money, not more.

The rule that produced all of it only becomes possible once there is a line. The same chart that told him to cut tells him when spend can go back up. Cameron's version of the payoff is one line: "Just knowing that every ad dollar that we're spending is strategic and profitable."

One of those four August numbers has not cleared the bar yet. At $10.97, a new customer costs 28% less than a year ago. It still costs more than the model allows, once a month of fixed costs is carried inside it. That is not a footnote. The same gap is what moved the constraint in July, and the June break-even simulation had already shown it.

Six months in, with the numbers finally honest, the ceiling had become the offer. Our live profitability model puts the break-even cost of a new customer at $9.38. It runs on a blended net order value of $50.58 and monthly fixed costs of $152,000. Its fixed-cost input was last updated on 9 July 2026. At $10.97 we are buying them about 17% above that. Some of the small incremental campaigns come in under it. None of them holds it at the volume the business needs. And a new customer's first order is worth $46.49 after discounts, which is the number the next fix has to move.

So in July we delivered a five-offer package. It is built to raise a new customer's first order from $46.49 towards $70. The test is written into the document. If that number does not move, the thesis is wrong and the constraint is elsewhere. It has not moved yet. There is no launch result on record, and we are not going to write one before there is.

In August we handed over a full audit of an email program we do not run and do not bill for.

What all of it bought him was the ability to decide.

Last October he was signing off a budget he did not believe. Today he runs the same business against a number he can see. When that number says stop, the budget stops. The next thing to fix is the offer, and the test for it is already written down.

Cameron on camera, in his own words. He talks about the two agencies that did not work. He talks about the years of spending he had stopped believing in, and what the reporting changed.

Every year-over-year figure on this page compares like with like. This business swings hard with the season. Picking the best month of one year against the worst of another would flatter it. New-customer cost was also already falling in late 2025, before we began. Two pre-engagement months ran below where 2026 sits today. Months are not comparable to each other in a business like this. That is why every figure here is set against the same period a year earlier. The same month, the same run of weeks, or the same eight months. August 2025 was also the last month of a Google budget that was cut in half the following month. That is why this page never rests on that pair alone. The eight-month average sits beside it, $13.59 to $9.30. So does the movement inside 2026: $6.94 in March, $12.05 in July, $10.97 in August. That is the best month, the worst month, and the month the headline rests on. Every one of them beat the same month of 2025. Ask us for those pre-engagement months on the call and we will show you. The before-state figures are not before-and-after claims about our work.

"It sounds cliche to say but it helps me sleep at night."

Verify this case study

• Watch the full recorded interview above. It is unscripted and on camera.
• Look the business up yourself. speedcubeshop.com has traded since 2009, and Cameron Brown is its named public face.
• The diagnosis came first, and it is dated. The audit was written in October 2025 off September 2025 data. Two audit calls were recorded. The work started in January 2026.
• Every month-by-month and week-by-week figure here comes from one live reporting sheet, updated weekly. The first-order economics come from the dated July 2026 offer analysis. The account findings come from the October 2025 audit. Ask us on your Strategy Call how any figure here is sourced and dated. We will walk you through the method and put you in touch with references.