KNOWLEDGE 4 min READ

PART OF DIGITAL MARKETING

Traffic arrives, traffic leaves, and the diagnosis lands on button colours, page speed, or the absence of an exit popup. Sometimes that is fair. Often the visitor left for a simpler reason: they could not work out what was being sold, to whom, or why it was different. Optimising the mechanics of a page nobody understands is polishing the wrong thing.

Why it matters

Every improvement in conversion improves the return on every rupee spent sending people to the page, for as long as the improvement holds. It is the multiplier under performance marketing, SEO and content alike. It is also where a company's upstream problems become measurable: the funnel reports the leak, and the cause is usually one step earlier than the stage that shows it.

The order of operations

1. 01 The offer

   What exactly does the customer get, at what price, with what risk removed? If this is unclear, no page can fix it. See a better funnel cannot rescue a confused offer.

2. 02 The message

   Can a stranger repeat the offer after one read of the first screen? Clarity converts before the CTA does.

3. 03 The journey

   Where in the actual path do people stop, from the customer journey map and the analytics? Fix the biggest drop first, not the easiest.

4. 04 The mechanics

   Form length, page speed, trust signals, layout, the button. Real, and worth doing, and last.

Reading funnel data upstream

• High add-to-cart, high checkout abandonment: often a pricing surprise, not a checkout-UX problem.
• Strong traffic, weak opt-ins: usually a who-is-this-for problem.
• Great signups, brutal churn: the offer promised something the product delivers differently.
• Good conversion, bad revenue: the wrong people are converting. Look at targeting and positioning.

How to test without fooling yourself

• One hypothesis per test, written down before the test, with the number that will decide it.
• Enough traffic to reach a conclusion. Small sites should test big changes, because small changes cannot be measured on small numbers.
• Measure the business outcome, not the click. A test that raises clicks and lowers qualified enquiries has failed.
• Keep a log. Losing tests are knowledge; a team that only remembers the wins repeats the losses.
• Where traffic is too low to test, use qualitative evidence: the five-second test (show the first screen to a stranger for five seconds, then ask what is sold and to whom), session recordings that mask personal details, and customer conversations. Evidence does not have to be statistical to be useful.

Common failure modes

• Optimising the mechanics of an offer nobody understands.
• Testing trivial changes on traffic too small to detect them, then acting on noise.
• Optimising for the platform's conversion definition rather than the business's.
• Stopping at purchase. Retention is conversion that keeps paying.

Before rebuilding the funnel, run the cheaper experiment: make the first screen understandable to a stranger. It is remarkable how often the funnel turns out to have been fine.

What is a good conversion rate?

One better than yours last quarter, for the action that matters to the business. Benchmarks vary so much by offer, price and traffic source that the industry average is a distraction. Compare against yourself, over time.

Do we need an A/B testing tool?

Only once you have the traffic to use one and the hypotheses to test. Many businesses get further, faster, with the five-second test, a handful of customer conversations and a clear rewrite of the first screen.

How is CRO different from UX design?

UX designs the experience; CRO measures whether it produces the action and changes it until it does. Good CRO uses UX methods and holds them to a business number.

GO DEEPER

Reading is free. The diagnosis is a conversation.

Say Hi

NEXT TOPICS TO LEARN