From $0 to $5k/Day: A Week-by-Week Anatomy of Launching a Brand-New Ad Account

The most expensive part of launching a brand-new ad account isn't the ad spend. It's the nerve it takes to keep spending while the numbers look terrible.

Let me put a figure on that. Picture a brand-new account with a target of a A$30 cost per acquisition, and a hard deck of A$75 a day that you've agreed to hold no matter what. Through the first few weeks that account runs at a A$100 CPA. So every single day, you're knowingly losing roughly A$45. Over a month of bad early performance, that's well over a thousand dollars set on fire on purpose, before you've found a single thing that works.

Most people can't sit through that. They panic at day four, start yanking budgets, swapping creatives daily, second-guessing the product, and they strangle the account before it ever gets going. The hard deck is the discipline that gets you to the other side.

I want to walk you through a brand-new account, week by week, the way it actually unfolds. Invented numbers, anonymised brand, but the shape is exactly what these launches look like. A wellness drink, never advertised before, A$1,000 in lifetime revenue from friends and family, an empty ad account. Target CPA A$30, AOV around A$42, so we're chasing roughly a 1.4x return that works because it's a subscribe-and-save product where the real money is in the repeat.

Here's how it played out.

The four rules I set before spending a dollar

Before launch I locked in a few decisions, because the worst time to decide your rules is in the middle of a bad week when you're emotional about it.

One. Purchase objective from day one. No traffic campaigns, no warming up, no "let's get some engagement first". I optimise for the purchase event from the first dollar, even on a cold account with zero data. Broad targeting, no interests, no lookalikes. The creative does the targeting.

Two. NCPA is the north star. New customer cost per acquisition. Not ROAS-on-the-dashboard, not cost per click, not CPM. The job of paid is to bring in new customers, so that's the only number I let drive decisions. Everything else is noise I glance at, not steer by.

Three. One campaign, a hard deck, and a baseline to beat. A single CBO campaign. A hard deck I will not scale below regardless of how ugly it gets early. And whatever ad is currently performing least-badly becomes my baseline. The whole game becomes "beat the baseline", not "panic at the number".

Four. Seven days before I judge a creative. On a tiny budget you barely get any data per day, so I give each test a proper week before calling it, unless it tanks performance hard on day one. The exception matters: if a creative takes over spend and destroys the account in 24 hours, it's gone. I'm not married to the seven days, I'm married to not making twitchy decisions.

With that set, we launched.

Weeks one to three: bleeding slowly, on purpose

Week one. Budget capped at A$50 a day. Worth saying plainly: a brand-new account is often limited to a small daily spend whether you like it or not, and the card sometimes gets declined in the first fortnight while the bank gets used to a daily charge. We spent about A$340 for the week and got one sale. A A$340 CPA. Every dropshipper on earth would have quit the product right here.

We didn't. We had a hard deck and a target, and one week of data on a cold account tells you almost nothing.

Week two. Still A$50 a day, about A$310 spent. Our simple opening video settled in as the best of a bad bunch at roughly a A$58 CPA. Nowhere near the A$30 target, but it's now the baseline. I did not turn it off. People ask me all the time whether they should kill an ad that's over target, and my answer is: not if it's your best one and nothing's beaten it. Killing your only consistent seller to chase ads that aren't spending is how you end up with no sales and a lot of decisions to make.

Week three. A new creative took over spend, ran for the week, and got zero sales off about A$150. Easy kill. It actively hurt the baseline, so it's gone. This is the pattern that repeats the whole way through: an ad that eats budget but doesn't sell is worse than no ad at all. Notice the kill was cheap because the budget was small. The same mistake at A$1,000 a day would have cost thousands.

Three weeks in, still bleeding, still holding.

Weeks four to seven: holding the deck through the ugly middle

Week four. A billing knock-back took the account offline for most of the week. We spent about A$50 total. New accounts do this. Annoying, not fatal. We did get two new creatives launched.

Weeks five and six. Around A$180 then A$420 for the weeks. One of the new videos, a slightly odd one we nearly cut, started taking spend at a A$59 CPA and became the de facto winner for a stretch. Roughly the same as the old baseline, but with more legs. I started nudging spend up a touch, partly to test where the next budget ceiling sat.

Week seven. Here's a thing worth knowing. I pushed daily spend from A$50 up towards A$100 to test the common theory that more budget gives the algorithm more data and improves performance. It did not. CPA went from about A$60 to A$88. That told me something clean and useful: the problem wasn't budget, it was creative. We simply didn't have the winning ad yet. More money on weak creative just buys you more expensive losses.

So we went quiet on launches for a week or two while a batch of properly briefed creator content was in production behind the scenes. This is the part nobody shows you: long stretches where you're not launching anything because the good material isn't back yet. A creator we'd hired in early August took the better part of five to six weeks to ship product, film, and get the content back. That lag is real, and you plan around it or you stall.

Weeks eight to nine: the winner

Week eight. Daily spend around A$90. We launched a batch, including a dead-simple video, no voiceover, just music, that looked like something a real person posted rather than an ad. It opened at a A$27 CPA on its first chunk of spend. First time we'd seen anything under target.

Week nine. It took over. Spend climbed to roughly A$150 a day, the new video carried the account, and we landed around a A$32 CPA for the week. After two months of bleeding, we were finally a couple of dollars off the A$30 target.

I want to be honest about why it won, because it's the whole lesson. It wasn't slick. It was the opposite. It looked native to the feed - the kind of clip you'd scroll past without clocking it as advertising - but it still had the fundamentals underneath it: a clear hook, a real reason to care, the right person talking. Camouflage with substance. That combination beat every polished, obviously-an-ad creative we'd run.

And finding the winner moved the goalposts on us. The bar was a A$58 ad, then a A$32 ad. Every new winner raises the standard the next creative has to clear, which is exactly why people get stuck. You're not failing to beat your winner because you're unlucky. You're failing because you haven't made anything better than it yet.

Weeks ten to thirteen: scaling on a rule, not a feeling

From here the job changed from "find a winner" to "scale without breaking it". The rule is boring on purpose: 20% up on a good day, 20% down on a bad run.

When yesterday's NCPA sits at or under target, I increase the budget by 20%. When it runs hot, I do not touch it for three to four days before cutting, because a single bad day usually auto-corrects, and lunging at it just turns a wobble into a crash. From a client's seat it can look like I'm doing nothing. Doing nothing, deliberately, is often the highest-skill move in the account.

We ran that. Daily spend stepped from a couple hundred, to A$800-ish, towards A$1,000 a day across these weeks, with the main winner holding a A$28 to A$31 CPA the whole way up. Newer iterations got a lower CPA on tiny spend, but I left the big winner alone - a A$60 spend with a slightly better CPA isn't a reason to kill a A$7,000 spend that's still under target. Anything beating my top spender's CPA isn't hurting me, so it can stay, but it's not big enough to crown.

The trick to those iterations, by the way, was not reinvention. It was the same winning clip in new locations and with different people. Same script, same structure, filmed in a car, then a park, then a kitchen. That's where a lot of the next leg of scale came from.

The rules the story keeps proving

Strip the weeks away and the same handful of rules did all the work:

• Hold the hard deck through the ugly early CPA. The account that quits at day four never finds week nine.
• Set a baseline and beat it. Don't kill your best seller to chase ads that aren't selling.
• An ad that takes spend and doesn't sell is a kill, fast. Cheap to do early, brutal to do late.
• More budget doesn't fix weak creative. If pushing spend lifts CPA, the answer is a better ad, not more money.
• Scale on the number, not the nerves. 20% up on target, sit on your hands for a few days before cutting.

None of that is clever. It's just held, consistently, while the numbers are frightening enough to make most people break the rules.

The honest part I'll leave you with: this account took the better part of two months to get from a A$340 CPA to a A$32 one, and most of that time it looked like a failure. So if you're staring at week three of your own launch wondering whether the product is broken or you are - which one is it really, and have you given it long enough to know?