$783,518 of Meta Ads: What 23 Accounts Actually Bought

Twenty-three Meta Ads accounts. Australian businesses across five industries, the oldest campaign built in 2013, the newest in the week we pulled the data. Between them they spent $783,518 AUD, bought 999,000 clicks from 77.7 million impressions, and built 1,134 campaigns, of which 67 were running on 9 September 2026.

Our $32M Google Ads series went deep on one account. This series does the opposite. It asks one question of twenty-three accounts and reports what recurs. A finding that shows up in one account is that advertiser's mistake. A finding that shows up in nineteen of twenty-three is the platform.

Every number was pulled from the Meta Marketing API v23.0 on 8 and 9 September 2026. Nothing is illustrative.

How the accounts are described

No client is named, and no single account's spend is published. Accounts appear as letters (A to W) and as one of five industry bands, each holding at least three accounts:

Two rules follow from that. Money is only ever shown pooled across all twenty-three, or pooled by placement, segment or year. Rates, ratios and counts are shown per band or per account letter. Two of the five bands are dominated by a single advertiser, so a band-level spend figure would identify them, and we do not publish one.

Where a ratio rests on fewer than 30 conversions or under $500 of spend, the post says so in the sentence that uses it. Several of these accounts are small. A 4x figure on three purchases is not a finding, and we have tried not to present one as if it were.

Start here

If you read one post, read the first. A column Meta names with the word "leads" reported 260,849 for an account that had 2,003. Nobody had done anything wrong. That is the point of the series: most of what follows is not advertiser error but the platform's own definitions, defaults and reporting, and what they quietly do to a budget.

Reading order

Number order is the order the argument builds: what the reported numbers are made of (1 to 7), what the platform did underneath everyone (8 to 12), what the advertisers chose or declined to choose (13 to 18), where the money actually went (19 to 23), creative (24 and 25), and the two findings we could only publish with a caveat.

Parts 1, 3, 13 and 19 stand alone. Parts 5, 6 and 7 are one argument in three pieces. Parts 9 to 12 share one placement table and are best read together.

What this series is not

It is not a benchmark. Twenty-three accounts in five industries with one advertiser dominating two of the bands is a cross-section, not a sample of the Australian market. A rate in this series tells you what happened in these accounts, and the post tells you why. Whether the same mechanism is at work in yours is what the "what to check in your account" section of each post is for.

It is also not an argument that Meta Ads does not work. Several of these accounts buy leads at $14 to $39. The argument is that the number on the dashboard is a sum of things Meta configured, at some point, for some reason, and it is only as honest as the least honest thing in it.

If you would rather have someone pull your account apart the same way, that is what our Facebook Ads management work starts with.