Centsbot · audit · samples · guides

Landing page A/B testing with low traffic: what to do instead

Everyone tells you to A/B test your landing page. If you get a few hundred visitors a month, a test can run for months and still mislead you. This page shows how much traffic a test needs (with sources), the mistake that ruins most small tests, a plan for when you can't test, and a $19 review that gives you 3 evidence-based changes to ship within 24h.

$19 · 3 fixes · within 24h · the BTC invoice opens instantly, no account
Free observation first — not ready to pay? Send your URL and get 1 concrete observation on your page, free, no account. Request your free observation →

How much traffic does a test need?

Evan Miller's sample size calculator answers "how many subjects are needed for an A/B test?" With its example settings — a 10.2% baseline conversion rate and a minimum detectable effect that takes it to 13.2% — it shows 2,545 visitors per variation. Smaller expected lifts need more. Try your own baseline in the calculator: if the answer is more visitors than you get in a few months, a classic test isn't practical yet.

The mistake that ruins small tests: peeking

Evan Miller's How Not To Run an A/B Test explains that if you check an experiment repeatedly and stop when it first looks significant, instead of fixing the sample size in advance, all the reported significance levels become meaningless. With low traffic the temptation to call it early is strongest, which is exactly when it misleads you most.

What to do instead when you can't test

Nielsen Norman Group's advice is to complement A/B split tests with user research to identify true causes and develop well-informed design variations. A practical plan for a small site:

  1. Find the likely cause first. Watch 3–5 people from your audience use the page, or run a 5-second test.
  2. Get an outside read on headline, CTA and proof, such as the value proposition review.
  3. Change the 3 things you can justify, all at once, and write down the metric you expect to move.
  4. Measure before and after over the same number of days, and note anything else that changed (campaigns, seasonality).
  5. Test later when traffic supports it, with the sample size fixed in advance.

A before and after comparison is weaker evidence than a properly sized test. It tells you what changed, not for certain why.

What the $19 review gives you

We read your live page and pick the 3 gaps most likely to cost you signups. You get 3 before → after rewrites (hero, CTA, offer or proof), each with a one-line reason and a metric hypothesis you can check later, ready to paste, within 24h of 1 BTC confirmation. No call and no account. See an example deliverable (excerpt from a public sample) or browse all public samples.

Honest limits: we don't run tests, set up tools or promise a lift. If you already have the traffic for tests and want a testing program, see audit vs CRO agency.

$19 · 3 fixes · delivered within 24h · paid in BTC

You get a BTC invoice page with address, amount, a QR code and an open-in-wallet link. Work starts once payment is seen. Not sure yet? Ask for 1 free observation on /audit.

FAQ

Can I A/B test my landing page with low traffic?

You can, but it may take a long time to reach a trustworthy result. Evan Miller's sample size calculator, with its example settings (a 10.2% baseline and detecting a move to 13.2%), shows 2,545 visitors needed per variation. A smaller expected lift needs more. Check your own numbers in the calculator before you start.

Why shouldn't I stop a test as soon as it looks significant?

Evan Miller explains that if you check an experiment repeatedly and stop when it first looks significant, rather than fixing the sample size in advance, the reported significance levels become meaningless. Decide the sample size before you start.

What can I do instead of testing when traffic is low?

Nielsen Norman Group recommends complementing A/B split tests with user research to identify true causes and develop well-informed design variations. With low traffic, that means getting an outside read and talking to a few real users, then shipping the changes you can justify and measuring before and after.

Is a $19 audit a replacement for A/B testing?

No. It gives you 3 evidence-based copy changes to ship, each with a metric hypothesis you can check later. It can't prove a change works; only a properly sized test or a clear before and after can.

What do I receive for $19?

3 before and after rewrites for the highest-impact spots on your page (hero, CTA, offer or proof), each with a one-line reason and a metric hypothesis, ready to paste, within 24h of 1 BTC confirmation. No call and no account.

Related

Questions: [email protected]