What is conversion rate optimization?

The CRO work follows a cycle: tracking, hypothesis, prioritization, test or implementation, learning, new hypothesis. The cycle is not new. It follows the same logic as hypothesis-driven product development and scientific experimentation in general. What is often missing is not the methodology but the discipline to follow it when time pressure pushes the work toward quick decisions. Three things separate a serious CRO engagement from ad-hoc test work:

1. 01Tracking foundation. The right events, the right funnel, the right data. Before anything is tested you have to know where visitors drop off, which steps have the most friction, and which channels produce conversion versus pull-through. Google Analytics↗ and Tag Manager are the default foundation, but what is measured is customer-specific: e-commerce measures add-to-cart flow, B2B SaaS measures trial activation, lead gen measures form progression. Validating that events actually fire correctly and that funnel steps reflect the real customer journey is the first task.
2. 02Hypothesis formulation. A hypothesis has structure: if we do X, then Y will happen, measured via Z, because Q. Vaguely formulated hypotheses produce vaguely formulated test results and vaguely formulated decision support.
3. 03Prioritization with ICE or PIE. Impact, Confidence, Ease (or Potential, Importance, Ease). Tests cost time and traffic. We test what carries the largest business impact if the hypothesis holds, not everything that is easy to test.
4. 04Test or direct implementation. Sometimes you test (A/B), sometimes you implement directly (when the hypothesis is high-confidence or traffic is too low for significance). Both are legitimate. Test religion does not solve the problem.
5. 05Measurement and learning. A test gives three possible answers: the hypothesis holds, does not hold, or is incomplete. All three are worth recording. Learning between tests is what makes the hypothesis pipeline improve over time.

All work is documented in a monthly report. Decision support, not status: the hypotheses, the priorities, the result, and what is to be tested next. One hour of an experienced hand carries further than four inexperienced ones, which means even short engagement forms have time to give a concrete picture.

The most common question in CRO is the wrong question. "What is a good conversion rate?" implies there is a global answer. There is not. The right question is: what is a good conversion rate for your industry, your product price, your customer segment, and your traffic source? The variance is too wide for global averages to be meaningful.

• E-commerce: a typical range is 1 to 3 percent. Low-price products typically convert higher; high-ticket business lower. Industry and season affect strongly.
• B2B SaaS: 2 to 7 percent is typical for trial signup, 10 to 25 percent for trial-to-paid. Long decision cycles dilute the figure in the short term.
• B2B lead gen: 2 to 10 percent form submission is common. The quality of leads is often more important than the volume. A high conversion rate on wrong leads is not valuable.
• Traffic source matters: paid search traffic typically converts higher than organic because intent is sharper. Social channels are typically lower, because the visitor is more in discovery mode.

Benchmark against your specific combination, not against a global average. A 1.5 percent conversion rate can be poor in an industry where 4 percent is the norm, or exceptional in an industry where 0.5 percent is the norm. Without reference points there is no baseline to improve against, and without a baseline, improvement work is subjective. The first thing we do in a CRO audit is establish what your normal looks like.

CRO work rarely gets stuck on lack of ideas. It gets stuck on four recurring pitfalls:

1. 01Tests without tracking. A/B tests are run before the data setup is clean. The result is credible numbers on the wrong measurements. Funnel steps are mis-defined, so tests show winning variants but the wins never reflect in revenue.
2. 02Hypotheses without business connection. "The button should be green" is not a hypothesis, it is a design preference. A real hypothesis has a measurable consequence on a specific business KPI. Without that link, CRO work becomes a design-agency deliverable in disguise.
3. 03Best-practice application. "Top 10 CRO tips" works for someone, somewhere, in some situation. They are not hypotheses for your funnel. Hypotheses have to come from your specific traffic and your specific visitors, not from a generic list.
4. 04PDF without implementation. A common model is an audit report: the customer receives 47 recommendations in a PDF. Three months later two of them have been implemented, because nobody had time or mandate to go into the code. Recommendations that are not implemented are unanswered questions.

Nielsen Norman Group's research on conversion design↗ shows that the most impactful changes are rarely the most dramatic. They are most often the most considered. CXL's analysis of experiment design↗ goes deeper into prioritization and statistics pitfalls. The common thread in the pitfalls is that they are not about lack of ideas or lack of tools. They are about lack of discipline: skipping hypothesis formulation, not tying to business KPIs, not verifying tracking, not implementing. Discipline is what separates a CRO engagement that moves the needle from one that moves dashboards.

Not all conversion problems are CRO problems. One of the most common reasons CRO work fails is that the underlying problem lies elsewhere:

• It is CRO work if: traffic converts poorly compared to the industry, funnel steps show clear drop-offs, copy or UX flow has friction, or hypotheses about user behavior are untested.
• It is web work if: the site is technically weak (load time, mobile rendering, fundamental UX errors that copy cannot fix), or the structure blocks the conversion flow in ways CRO cannot optimize away.
• It is SEO work if: the traffic that lands has the wrong intent. Search engine optimization filters out the wrong visitors. CRO cannot convert the wrong audience into the right buyer.
• It is copy work if: the message does not meet the audience's situation. CRO tests variants of copy, but if copy is generally vague or misdirected, CRO iterations are not enough.

The dividing line is not exact science, but it should be one of the first questions in a CRO audit: is this CRO work, or should the resources go to web, SEO, or copy first? Google's guide to page experience↗ is a good entry point to the web-vs-CRO disambiguation. If this points at your situation, see how Memorise works with conversion.