Conversion Rate Optimization (CRO)

Conversion rate optimization, or CRO, is the practice of systematically increasing the share of visitors who complete a goal, such as buying or signing up. It works as a cycle: research where and why people drop out, form a hypothesis, test a change, measure the result and keep what improves the outcome.

A process, not a list of tricks

CRO is often reduced to tips such as changing button colours. Those tips rarely transfer from one store to another, because each audience and product range is different. The lasting value of CRO comes from the method: evidence first, a clear prediction, a controlled change and an honest reading of the result.

The cycle repeats. Each test, win or loss, adds knowledge about the customers, which feeds the next round of research.

Good CRO also looks beyond the conversion rate. A change that raises orders but lowers average order value, or that attracts more returns, can reduce profit. Teams that define success as revenue or gross profit per visitor avoid celebrating results that do not help the business.

Step one: research

Good ideas come from understanding the problem. Research combines numbers that show where people leave with observations that suggest why.

  • Analytics: funnel reports, step drop-off, device and source segments
  • Behaviour: session recordings, click and scroll maps, site search terms
  • Voice of the customer: short on-site surveys, support tickets, reviews and returns reasons
  • Usability sessions: watching a handful of people try to complete a purchase
  • Market context: how prices, delivery terms and product ranges compare with other stores

Market context is often skipped. If shoppers leave a product page because the same item is cheaper elsewhere, no amount of layout testing will fix it. Tracking competitor prices gives the research step that missing view.

Step two: writing a hypothesis

A hypothesis links an observation to a change and a measurable outcome. A useful format is: because we saw this, we believe changing that will improve this metric for this audience.

For example, a shop selling bicycle parts sees in surveys that mobile shoppers are unsure about delivery costs, and the funnel shows heavy loss between cart and checkout on phones. The hypothesis: showing the delivery cost on the product page will raise mobile checkout starts. It names the evidence, the change, the metric and the segment, so the result can be judged cleanly.

A weak hypothesis, such as make the checkout better, cannot fail and therefore cannot teach anything. If the test shows no change, a well-written hypothesis still tells the team that delivery cost clarity was not the main barrier, which narrows the next round of research.

Step three: prioritising ideas

Research produces more ideas than any team can test. Simple scoring models help rank them. Two common ones are ICE, which scores impact, confidence and ease, and PIE, which scores potential, importance and ease. Each idea gets a score for each factor, and the total sets the order.

Scores are judgement, not science, but they force a discussion about evidence and effort. Give extra weight to ideas that affect high-traffic pages or high-margin products, since a small gain there is worth more than a large gain on a page few people see.

Step four: testing and reading results

The standard method is an A/B test, where visitors are split at random between the current version and the change. Decide the main metric, the sample size and the test length before starting. Run tests across full weekly cycles so weekday and weekend behaviour are both included, and avoid stopping early the moment a result looks good.

Low-traffic stores may not reach a reliable sample in a reasonable time. In that case, focus on larger changes, test on the busiest pages, or use before-and-after comparisons with care and accept that the evidence is weaker.

After a test ends, check segments before rolling out. A change can win on desktop and lose on mobile, or help new visitors while slowing down returning buyers. Rolling out to everyone on the strength of the average can undo part of the gain.

Price as a CRO lever

Price affects conversion directly, so it belongs in CRO, but it needs extra care. Showing different prices to different visitors at the same time can upset customers who notice, and rules on price display and fairness vary by country. Many businesses test price presentation instead of the price itself: how a discount is shown, whether shipping is included, or how bundle savings are explained. If you plan to test actual prices, check the rules where you sell and take legal advice.

Whatever is tested, judge the result on profit per visitor, not conversion alone. A change that adds orders but relies on a deeper discount may leave the business worse off.

Competitor moves can also distort any test. If a major competitor cuts prices halfway through a test, both versions may lose conversion, and the comparison between them becomes harder to read. Note market events in the test log so results can be interpreted in context.

Pitfalls that waste tests

  • Testing without research, so ideas are guesses
  • Changing several things at once and not knowing which one worked
  • Calling a winner too early on a small sample
  • Ignoring segments, so a gain on desktop hides a loss on mobile
  • Not recording failed tests, which leads to repeating them later

Keep a simple log of every test: hypothesis, dates, traffic, result and decision. Over time this log becomes one of the most valuable records the team has.

Key takeaways

  • CRO is a repeatable cycle of research, hypothesis, test and learning.
  • Research should include market context, such as how your prices compare.
  • A good hypothesis names the evidence, the change, the metric and the audience.
  • Prioritise with a simple score and weight high-traffic, high-margin pages.
  • Judge price-related tests on profit per visitor, not conversion alone.

Frequently asked questions

How is CRO different from conversion rate?

Conversion rate is a metric. CRO is the ongoing process of researching, testing and improving to raise that metric, ideally alongside revenue and margin.

How much traffic do I need for CRO?

Research and fixes for obvious problems need no minimum. Reliable A/B tests need enough conversions in each version to detect a difference, which can take a long time on small sites. Low-traffic stores should test bigger changes on their busiest pages.

Is it legal to A/B test prices?

Rules differ by country and sector, and price tests can also harm trust if customers notice. Many stores test how prices are presented rather than the amount. Check local rules and seek legal advice before testing different prices.

Related terms

In BlackAha, Pricing Reports helps you analyze pricing and competition by category, brand and domain.

Back to the Pricing Dictionary