A new customer cost $332.A year later, $163.Media spend fell 33.96% on the way there.
A US pet tech brand selling a premium GPS collar: track the dog in real time, set safe zones, get an alert when it leaves one. The collar costs $700 to $990 and an owner buys it once. Those two facts set a hard ceiling on what a customer is allowed to cost, and the account was running at double it. Below is what changed in a year, and which numbers in the source do not add up.
- $332 → $163
- what one new customer cost, year on year
- -39.60%
- company marketing costs behind that figure
- +22.66%
- more units sold, on less spend
- 1.95 → 3.84
- ROAS as the study reports it, on a basis it never defines
The product set the ceiling before a single campaign ran.
A $160 ceiling on something bought once
The collar sells for $700 to $990 and the purchase is one-off, so there is no second order to recover an expensive first sale. The economics allowed a maximum of $160 to acquire a customer. The account was paying $332. At that point this stops being a performance conversation and becomes an arithmetic one.
A buying cycle nobody could attribute
Four to six weeks passed between first contact and purchase, and the measurement in place could not connect a sale back to the campaigns that started it. Awareness work on YouTube and Google Display therefore looked like cost with no return, which is how awareness work usually gets cut, whether or not it was doing anything.
One flat budget against a seasonal product
Demand peaked from the end of February through June and fell away in winter, but spend was spread evenly across the year. That burns money when nobody is buying and buys too little visibility when everybody is. Google search trends agreed with the sales data, so this was never a hidden pattern, only an unused one.
Measure it, then cut what the measurement exposed.
One Performance Max campaign became two
Awareness, new customer acquisition, remarketing and brand had all been running inside a single Performance Max campaign, competing for one budget. Brand searches took the largest share, because they convert most cheaply and the system optimises for exactly that. After the split, one campaign handled brand queries and the other went after new users and category share.
Attribution wired for views, then tested for lift
Northbeam went in to track impressions alongside clicks, so conversions could be attributed to YouTube and Display instead of vanishing. Haus measured incremental impact on top of that. The first tool tells you where credit belongs, the second tells you whether the sale would have happened anyway. Budget moved to the channels that survived both questions.
Keywords split three ways, with the argument following the intent
A single generic message had been running against every search. It was replaced with three keyword groups and three arguments: product phrases such as "GPS wireless dog fence" got the technology and the absence of a physical fence, category phrases such as "electric dog shock collars" got the safety case against shock collars, and competitor terms got the alternative. Someone searching for a shock collar needs a reason to consider something else, not a spec sheet.
Budget followed the season instead of the calendar
Account data, the CRM and previous years of sales put the peak between the end of February and June. Spend was raised into that window and pulled back through the quiet months. Flat budgeting on a seasonal product is a decision to overpay in winter for the privilege of being invisible in spring.
One number carries this case. It landed three dollars above its target.
Year on year, the cost of acquiring one customer, set against the ceiling the product economics fixed before any of the work started. Three figures, all in dollars, all from the published study.
A halved acquisition cost is the kind of claim that usually arrives without a mechanism. This one has one. The study reports company marketing costs down 39.60% and units sold up 22.66% over the same year, and cost divided by customers turns that pair into a fall of a little over half, which is exactly where $332 to $163 lands. Note that it is the marketing cost line that produces it, not the 33.96% fall in digital media spend on its own. Revenue rose 18.97% while all of that happened, and the study puts the revenue coming from Google ads specifically at 117% higher, with category share up 60%. Two things in the source stay off this chart, and you should know why before you click through to it. First, the reported ROAS of 1.95 to 3.84 does not follow from 33.96% less spend and 18.97% more revenue, and follows even less from 117% more revenue out of Google ads. The study never says which revenue and which spend sit in that ratio, and separately mentions a bROAS that carries operating costs. It keeps its own label up top, apart from the figures it fails to reconcile with. Second, the summary panel at the top of the source page reads 49.98% revenue growth, a 110.77% ROAS increase and a 58.41% CAC decrease. None of those three match the body of the same page, and the third contradicts its own headline of a CAC cut by over 50%. They are not used here.
The ceiling comes first. Everything else is downstream of it.
Start with the number your product forces on you. An item priced at $700 to $990 and bought once gives you a maximum you can pay for a customer, and here it was $160. Until that figure is written down, every campaign decision is a matter of taste, and an account acquiring customers at $332 can be defended with any metric you like. Once it is written down, there is nothing left to argue about.
The second move that travels is the dull one. A single Performance Max campaign covering awareness, acquisition, remarketing and brand will put most of its budget in front of people who already know you, because those are the cheapest conversions available to it. Splitting brand out is not a clever optimisation. It is the only way to see what a genuinely new customer costs you, which is the number your margin actually cares about.
The third sits outside the ad account. The product carried an average rating of around 3 stars, largely from reviews left by people who had never bought it, and that rating met every visitor the campaigns delivered. Fixing it was not media work and it was recommended anyway, before the season opened. If the page your ads point at argues against them, you are paying to send people somewhere that talks them out of it.
This work was delivered by MTA Group, which Zero Fluff Digital is part of, by the same specialists who would work on your account. Every figure here comes from the case study MTA published, republished here with their consent. The reviews behind it are public and verified on Clutch, where clients rate the work rather than the agency describing itself.
Your margin sets a ceiling on what one customer may cost.
30 minutes, no deck. You leave with the maximum your margin can pay for a customer, and whether your account is anywhere near it today.
Book a callNext case studyAll Ears Vet: 37% more new clients from Google and Meta Ads →