Why 24 beats 240.

Rule count is the easiest marketing number in this category. It scales with effort: ship another rule, the badge ticks up from 100 to 101. It looks like more value for the same price. It reassures a buyer who's never thought hard about what a "rule" actually is.

The problem is that an audit tool isn't a checklist app. Every rule is a statistical test, and statistical tests have a false positive rate. When you stack them, the false positive rate stacks too. That's not a marketing footnote — it shows up in your audit as findings that aren't real.

When you run a statistical check at 95% confidence, you accept a 5% chance of being wrong by pure noise. Run one check, expect roughly zero false positives. Run a hundred, expect about five. The numbers compound. Here's the expected false-positive count at the standard 95% threshold:

Expected false positives = n × α. The math assumes the worst case — every test is testing pure noise. A real audit has fewer because some rules are testing real signal. The point is the direction: more rules, more wrongness.

This is the multiple comparisons problem. Every textbook on statistical inference covers it. The standard corrections — Bonferroni, Holm, Benjamini-Hochberg — exist specifically because running many tests at the naive threshold produces findings that aren't real.

Every false positive is a finding that fires, recommends a fix, and turns out to be nothing when the customer implements it. The customer loses trust. They churn. The math doesn't care that the marketing said "150+ checks."

Honest Growth applies Benjamini-Hochberg correction at q = 0.10 to the rules that produce many sub-findings — including audience overlap pairs and placement comparisons. Most audit tools don't apply this correction. The math gets buried in the marketing.

The "24 rules" framing is the surfacing layer — what shows up on your report. The rigor is in how each rule fires against every eligible entity in your account.

Take a real account: 14 ad sets, 38 ads, 12 audiences, 7 placement types. Our 24 rules produce something like this:

• 200+ individual statistical checks per audit — 23 patterns multiplied across eligible entities (15 at launch, +2 in v1.1, +6 deterministic strict-evidence additions in v1.2)
• 47 statistical signals computed across the dataset (rolling windows, sibling deltas, audience-pair overlaps)
• 20,000+ data points examined on a typical mid-size account
• Wilson 95% confidence intervals on every rate comparison — proportions, not raw averages (Wilson 1927; the interval doesn't collapse at the boundary the way the textbook normal-approximation one does)
• Sample-size guards — at least 30 conversions per entity before any comparison fires
• Benjamini-Hochberg correction (q = 0.10) on rules that produce families of sub-findings (Benjamini & Hochberg 1995)

The trade is deliberate. More checks, fewer false positives, better findings — because every check is built to survive its own statistical scrutiny before it ever surfaces on your report. That's the right trade.

We're not naming names — most of these tools are competent at what they do, and the structural problem is the category, not the team. But three forces push every audit vendor toward inflating rule count:

1. 1. Volume justifies pricing.

   When your pricing is "$24/month for 100+ rules," cutting to 15 well-built rules makes the price look high. The headline number is doing the conversion work. Volume is marketing, not methodology.

2. 2. Black-box methodology hides the math problem.

   When the rules are proprietary, no one outside the company can audit the false positive rate. There's no incentive to publish your statistical corrections — or even to apply them — if no buyer is going to check.

3. 3. More findings = more dashboard signal.

   A customer who sees 24 findings feels they got value. A customer who sees 4 high-confidence findings feels they paid for "only 4 things." The truth runs the other direction: 4 things you can actually act on beats 24 things you have to sort through. But the dashboard optic favors the 24.

Don't trust our argument. Run our audit. Then run Markifact's, Madgicx's, anyone's. For each tool, count which findings produced a real recoverable dollar amount when you implemented the fix.

That's the number that matters. Rule count isn't a quality signal — implementation rate is. A tool that surfaces 4 findings and gets 3 of them implemented is doing more work than a tool that surfaces 40 and gets 2 implemented.

We'd rather lose that comparison than win on the marketing badge.

24 rules. 200+ statistical checks per audit. Wilson intervals. Benjamini-Hochberg correction. Every finding traceable to a specific row of your data and a specific calculation.

Read the methodology. Run the sample audit. Decide for yourself.

If we're wrong about one of the 24, tell us. The rules update. Read the changelog before any other audit tool publishes one.