The Rule of 10,000: Exactly How Many New Ads Your Brand Should Ship Each Week

Here's what creative starvation is quietly costing you, in plain money.

When your account runs out of fresh ads to test, the winners you already have keep getting shown to the same people, more often. Frequency climbs, the audience gets tired, and your cost per acquisition drifts up week after week. Nothing looks broken. There's no error, no disapproval, no obvious culprit. The account just slowly gets more expensive, and you blame the algorithm or the season.

Put a number on it. A brand spending A$100k a month that lets CPA creep from A$30 to A$38 because the creative pipeline dried up has just made every new customer 27% dearer. On that spend, that's thousands a week handed back for no reason other than running out of things to test.

The real ceiling on your spend isn't your budget. It's creative diversity. You can pour more money in, but if you've only got three live concepts, more budget just buys the same three ads more frequency and a worse CPA. So the question that actually matters is mechanical: how many new ads should you be shipping every week to keep the machine fed?

Here are the two ways I'd answer it. One's a back-of-the-napkin rule. The other is the precise version. Use whichever you'll actually stick to.

The quick version: the Rule of 10,000

If you want one number and you want it in ten seconds, here it is.

For every A$10,000 a month in ad spend, ship one new ad per week. Minimum two, no matter how small you are.

That's the whole rule. Take your monthly spend, divide by 10,000, and that's your weekly new-ad target.

• Spending A$100k a month? That's 10 new ads a week.
• Spending A$300k a month? That's 30 new ads a week.
• Spending A$3k a month? The maths says less than one, so you default to the floor: 2 a week, or if the pipeline's genuinely hard to fill, 4 every fortnight. Just keep it consistent.

One thing to be clear on, because it's the most common way this goes wrong: an "ad" here means a genuinely new concept or a real variation, not a headline swap. A new line of copy on the same video is not a new ad. We'll come back to why that distinction matters, but for now, when I say ten new ads, I mean ten real swings.

The Rule of 10,000 is deliberately blunt. It doesn't know your margins or your test budget. What it's good at is giving a team a clear, non-negotiable weekly target so the pipeline never silently runs dry. For a lot of brands that's enough, and it beats the usual approach of "make some ads when we remember to".

The precise version: spend, test budget, and your real CPA

When you want the number that's actually right for your account, you build it from what a test costs you rather than a flat rule. The logic is simple and it's worth doing once properly.

The formula is:

(Monthly spend x testing %) / in-platform CPA = conversions you can buy for testing, and from there you back into how many concepts that funds.

Let me unpack each piece, because each one is a real decision.

Monthly spend. What you're actually putting through the account in a month. Start here.

Testing percentage. The slice of budget you ringfence for testing new creative rather than scaling proven winners. Most brands I see sit somewhere around 20%. Push it higher when you're hunting for new winners or entering a new category, lower when you've found something working and you just want to milk it.

In-platform CPA. Not your blended number, not your MER. The cost per purchase Meta actually records in the account, because that's what governs how fast a test gathers enough data to call. This is the number people get wrong most often, and it throws the whole calculation out.

Divide the test budget by that CPA and you get the number of conversions your testing money can buy in a month. Then you decide how many conversions you need on a concept before you'll trust the read - call it somewhere around 8 to 10 purchases to make a confident decision - and that tells you how many concepts you can properly test. Spread that across the month and you've got your weekly number, built from your economics instead of a rule of thumb.

Worked example: a 7-figure brand

Let me run it end to end so it's not abstract. Invented numbers, but the shape is exactly what these look like.

A homewares brand spending A$80k a month. Sitting at a A$35 in-platform CPA. They ringfence 20% for testing.

• Test budget: A$80k x 20% = A$16k a month.
• Conversions that buys: A$16k / A$35 = roughly 457 testing purchases a month.
• At ~9 purchases to call a concept with confidence: 457 / 9 = about 50 concepts a month the budget can genuinely support.
• Across roughly four weeks: about 12 to 13 new concepts a week.

Now sanity-check it against the quick rule. A$80k / 10,000 = 8 a week from the Rule of 10,000. The precise method says 12 to 13. They're in the same postcode, and the gap is informative: this brand's relatively low CPA means each test is cheap, so they can actually afford to push past the blunt rule. The Rule of 10,000 would have left testing budget on the table.

That's the value of doing the real maths. The rule keeps you safe. The formula tells you where you've got room to be braver.

Worked example: an 8-figure brand

Now scale it up, because the dynamics shift.

A supplement brand spending A$600k a month. Higher CPA at A$55 because the category's competitive. Same 20% testing.

• Test budget: A$600k x 20% = A$120k a month.
• Conversions that buys: A$120k / A$55 = roughly 2,180 testing purchases a month.
• At ~9 purchases to call a concept: 2,180 / 9 = about 240 concepts a month.
• Across four weeks: about 60 new concepts a week.

The Rule of 10,000 says A$600k / 10,000 = 60 a week. This time the two methods land almost exactly on top of each other.

Here's the honest part though. Sixty genuinely new concepts a week is a serious production operation. No founder is shooting that off the side of their desk. At this scale the number stops being a creative question and becomes an operational one: can your pipeline physically produce 60 real swings a week, consistently, without the quality collapsing into slop? Most can't, and that gap between the number you should hit and the number you can hit is exactly where the next section comes in.

How to actually hit the number without an in-house studio

Most brands can't staff a creative team big enough to feed these numbers, and they don't need to. The volume's only worth chasing if the inputs stay decent, so here's where I'd get it cheaply.

Iterate your winners instead of inventing from zero. This is the biggest one. Once a concept works, you don't reinvent the wheel for every new ad. You go deep on the winner and ship a dozen variations of it: same idea, different setting, different person, different opening. A huge slice of your weekly number can come from extending what's already working, which is far faster and cheaper than net-new concepts and tends to perform better too.

Brief creators properly and batch it. A handful of UGC creators on a clear brief will out-produce an expensive studio for feed-native volume. Worth knowing the lag is real: a creator can take four to six weeks from brief to footage in your hands, so you commission continuously, not when you've already run dry.

Use AI as an accelerant, not the strategy. The brands getting value from AI aren't generating fifty ads in an afternoon and dumping them in the account. That's how you manufacture slop at scale. They're using it to research angles, draft scripts, and spin variations on a concept that's already proven, on top of a real creative strategy. Accelerant, not autopilot.

Mine your competitors and your own back catalogue. Formats that have run a long time in your niche are running because they work. They're a legitimate starting point for new concepts, and the brief half-writes itself.

Stack those four and the scary weekly number gets a lot less scary. A meaningful share of it is variations of proven winners and well-briefed creator content, not 60 original ideas dreamed up from scratch.

Where I'd start

So here's what I'd do this week. Work out your real in-platform CPA, pick your testing percentage, and run the formula once. Compare it to your Rule of 10,000 number. If you're shipping well under either, that gap is almost certainly where your creeping CPA is coming from.

This is genuinely the exercise we run for clients before we touch anything else, because there's no point optimising budgets and bidding on an account that's quietly starving for creative. Set the weekly number, build a pipeline that can hit it consistently, and most "the algorithm hates me" problems sort themselves out.

So what's your number, and how far is it from what you actually shipped last week?