Both methods answer the same question: which version brings more conversions? They differ in effort and in the traffic required.
- A/B test: one change, two versions. Quick to set up, clear to evaluate and feasible with medium traffic. The right choice for most websites.
- Multivariate test: several elements in all combinations. Three headlines and two images already make six variants. Each needs enough visitors, so the method suits pages with a great deal of traffic.
- No test: with little traffic, no test delivers a reliable result in a reasonable time. In that case we implement well-founded improvements directly and compare before and after.
Typical starting points are product and landing pages, forms, the checkout process and the wording of headlines and buttons. Often the cause lies upstream of the page: an ad promises something different from what the landing page delivers. That is why we work closely with our SEA and SEO teams. A high bounce rate is often the first indication.