How to spot a real CRO audit from a useless one in 5 questions

If the audit doesn't give you concrete numbers on screen, you're in a PowerPoint, not in an audit.

You pay 1,000 to 4,000 EUR for a CRO audit and you get back 47 slides full of "heuristics", "best practices" and "strategic recommendations". Zero numerical predictions. Zero testable hypotheses. Zero priorities.

It's the same story three consultants told you before. And your conversion is still 0.4%.

Why 80% of CRO audits are theatre

The CRO industry has a problem. Anyone who read Nielsen Norman and finished a 12-hour course can write an audit. The audit costs 1,000 EUR, takes 5 days, ships a long PDF.

The PDF isn't the problem. The problem is that nobody predicts anything. Nobody says "if you change X, conversion goes up by Y points". Just lists of "should".

An audit without quantified hypotheses is an essay, not an audit. An essay won't lift your conversion.

The 5 questions that separate the real audit from the theatre

Before you pay one cent, ask the consultant 5 questions. His answers tell you everything.

1. "How many hypotheses do you ship and what lift do you predict for each?"

A real audit ships 8-15 prioritized hypotheses. Each one with estimated lift (e.g. "+0.4 percentage points conversion") and confidence interval ("between +0.2 and +0.7 with 80% confidence").

Bad answer: "depends", "we can't predict", "CRO doesn't work like that".

Good answer: "I ship 12 hypotheses, each with estimated lift and statistical significance threshold, prioritized by ICE".

2. "What data do you actually analyze and over what period?"

A real audit asks for GA4 access, Hotjar/Microsoft Clarity, Shopify Analytics, order database export. Analyzes minimum 90 days. Identifies the segment with worst conversion and why.

Bad answer: "I'll just look at your site and tell you".

Good answer: "I need GA4 + Clarity + 90 days of order export, plus 50 session recordings from abandoned carts".

3. "How do you prioritize what I test first?"

The only acceptable answer: explicit framework. ICE (Impact, Confidence, Ease), PIE or PXL. Each hypothesis gets a 1-10 score on three dimensions. Sort descending. Start with top 3.

Bad answer: "we'll do them all".

Good answer: "ICE on all 12, I attack the first 3 with estimated impact above 7".

4. "How long does a test need to run to be credible?"

The math doesn't lie. At 100,000 monthly visitors and 1% baseline conversion, to detect a 20% lift with 95% confidence you need 3-4 weeks. Below 10,000 visitors per month, traditional A/B testing is statistical suicide. You go Bayesian or you make changes big enough to skip testing.

Bad answer: "we run 7 days and see".

Good answer: "sample size calculator gives 3 weeks minimum for 20% lift, I bring power calculation on every test before launch".

5. "What happens if your hypothesis lowers conversion?"

Honest answer: "happens in 30% of tests, we kill the variant and learn". Dishonest answer: "doesn't happen ever".

Real CRO has 30-40% test win rate. The rest invalidates the hypothesis, which is also useful information. Anyone promising 100% wins is lying or not testing.

BEFORE vs AFTER: what a real CRO audit looks like

Typical case study: Shopify fashion store with 60k monthly visitors

A Shopify fashion store paid 1,500 EUR for an audit that recommended "add urgency on the product page". They added it. Conversion grew by roughly 6%, below statistical significance on the 60,000 monthly traffic.

Second time, they asked for an audit with the 5-question protocol above. The audit shipped 11 hypotheses, 4 of them with estimated lift above 0.3 percentage points. They tested the top 3 in 9 weeks.

Typical cumulative lift in this kind of scenario: 2 to 3 times the initial conversion, measured on at least 4 weeks post-implementation. AOV usually climbs 15-30% through bundles and cross-sell. Monthly revenue doubles or triples on the same traffic.

The difference between the two auditors wasn't price. It was methodology.

An audit with no quantified hypotheses, no sample size, no ICE prioritization, no implementation follow-up is a paid essay. Run the 5 questions before any bank transfer.

Action plan before you pay for your next audit

1. Bring the 5 questions to the first call. Write down answers word by word. Match them against the patterns above.
2. Ask for a 30-minute free audit. Any serious consultant looks at your site for 30 minutes and gives 2-3 concrete observations with numbers. Anyone refusing is selling smoke.
3. Verify the portfolio with clickable links. No logo-grid. Real sites, accessible. Call an existing client before signing.
4. Write sample size and estimated lift in the contract. If the consultant refuses to commit numbers to paper, you don't pay.
5. Pay 50% upfront, 50% after implementation with measured results. Not 100% in advance.

Final lesson

A CRO audit is not a PDF. It's a series of prioritized hypotheses, quantified, statistically predicted, walked through implementation. The rest is theatre.

You pay for testable hypotheses, not for colored slides.

Now: open the last audit you paid for. Count the hypotheses with estimated lift. If the answer is zero, you have your answer.