The CRO process: five steps to turn more visitors into customers

The CRO process is a repeating cycle: research what’s broken, form a guess about how to fix it, test that guess, and learn from the result. Then do it again. It’s how you stop guessing about your website and start knowing what works.

Most companies skip this entirely. An Econsultancy study of 450+ businesses found that 63% have no structured approach to conversion rate optimization. They just change things and hope for the best. The companies that do follow a process? They’re 84% more likely to see real improvement.

The five-step cycle below works whether you’re a solo marketer or a full team. Kick off the audit phase with the free CRO checklist to find your biggest conversion gaps.

What is the CRO process?

The CRO process is a loop, not a checklist. You research, hypothesize, prioritize, test, and analyze. Then you start over with what you learned.

Think of it like going to the doctor. You don’t just take random medicine. You describe your symptoms (research). The doctor makes a diagnosis (hypothesis). They decide what to treat first (prioritize). You try the treatment (test). Then you check if it worked (analyze). Next visit, you adjust.

The difference between this and what most companies do? Most companies skip the doctor entirely. They Google their symptoms and buy whatever’s on sale. According to CXL founder Peep Laja, “CRO is roughly 80% research, 20% experimentation.” Most teams flip that ratio. Or skip research altogether.

If you want a full CRO guide that covers the bigger picture, we have one. For a look at how the process fits into a broader CRO strategy, start there. This post is about the process itself, the step-by-step workflow you repeat every cycle.

Step 1: research (find what’s actually broken)

This is the step everyone skips. And it’s the one that matters most.

Research is detective work. You’re trying to figure out why visitors show up, look around, and leave without buying (or signing up, or filling out the form, or whatever you need them to do).

There are two types of research, and you need both:

Number-based research (what’s happening): Open your analytics. Look at where visitors drop off. Which pages get the most traffic but the fewest conversions? Where in the checkout do people abandon their cart?

Baymard Institute found the average cart abandonment rate is 70.19%. If yours is higher, this step will tell you where the leak is. Check the metrics that matter for which numbers to focus on.

People-based research (why it’s happening): Numbers tell you where visitors leave. They don’t tell you why. For that, you need surveys, customer support logs, and session recordings (video replays of real visitors using your site). Ask recent customers: “What almost stopped you from buying?” You’ll be surprised how honest people are.

Our take: If you only have time for one research method, read your last 50 customer support tickets. Every complaint is a conversion opportunity hiding in plain sight.

If you want a structured checklist for this step, our CRO audit framework walks through it in detail.

Time estimate: 1-2 weeks for a solo marketer reviewing analytics and reading support emails. 2-4 weeks for a team running surveys and session recordings.

Step 2: hypothesize (turn observations into testable ideas)

A hypothesis (your educated guess about what will work) has three parts: what you observed, what you want to change, and what you expect to happen.

Say your analytics show 60% of visitors leave your pricing page without clicking anything. Your hypothesis might be: “If we add a comparison table showing what each plan includes, more visitors will click ‘Start free trial’ because they’ll understand the value faster.”

That’s a good hypothesis. It’s grounded in data. It has a clear change. And it predicts a specific outcome.

A bad hypothesis? “Let’s try a green button.” No research. No reason. No way to learn anything even if it works.

CXL’s process has a useful sorting trick. Put every idea into one of three buckets:

• Test it when you need to shift behavior (like the comparison table example)
• Just fix it when something is clearly broken (like a 404 error on your signup page)
• Get more data when you’re not sure what’s happening yet

The “just fix it” bucket is underrated. Peep Laja’s agency found that simply detecting and fixing bugs generated $100 million for clients over seven years. Not from testing. Just from fixing things that were broken. Sometimes the landing page just has a dead link.

Step 3: prioritize (decide what to test first)

After research, you’ll have a list of ideas. Maybe ten. Maybe fifty. You can’t test them all at once. So you rank them.

Keep it simple. Rate each idea on two things:

	Low effort	High effort
High impact	Do these first	Do these second
Low impact	Do these third	Skip these

“High impact” means it affects a lot of visitors or a high-value page. “Low effort” means you can make the change quickly without a developer.

This might surprise you. Running more tests doesn’t automatically mean better results. An Optimizely analysis of 127,000 experiments found that impact per test peaks when teams run 1-10 tests per year. Beyond 30 tests per engineer per year, expected impact drops by 87%. Quality beats quantity.

One well-researched test beats ten random ones. Every time.

For more specific ideas on what to prioritize for CRO, we have a full breakdown. And if you want a practical walkthrough of how to optimize your conversion rate with tactical actions you can start today, that’s a separate guide.

Step 4: test (run it and be patient)

A/B testing is the core of this step. You show two versions of a page to different visitors and measure which one converts better. Half your traffic sees Version A (your current page), half sees Version B (your change).

Before you start, you need three things:

1. Enough traffic. Roughly 1,000 visitors per version per week for results you can trust. Less than that and you’re flipping a coin. Check your CRO testing methodology to understand why sample size matters.
2. A clear goal. What counts as a conversion? A purchase? A signup? A form submission? Pick one per test.
3. One change at a time. If you change the headline, the button color, and the hero image all at once, you won’t know which change made the difference.

How long should you run a test? At least 2-4 weeks. Never stop a test early because it “looks like it’s winning.” Early results are unreliable, like judging a movie by the first five minutes.

With Kirro, the gap between “I have an idea” and “the test is live” is about three minutes. Paste a script, pick a page, use the visual editor to make your change, and hit start. No developer needed. You can set up your first test right now and see for yourself.

For a deeper look at which CRO tools do what, we have a comparison.

Step 5: analyze and act (learn from the result)

Your test will end with one of three outcomes:

Winner. Version B clearly outperformed Version A. Implement it. Then go back to Step 1 and research what to improve next.

Loser. Version B performed worse. That’s not a failure. You just saved yourself from permanently making your page worse. Write down why you think it didn’t work.

Inconclusive. Neither version was clearly better. This usually means your change was too small to matter, or you didn’t have enough traffic. Try a bigger change or run the test longer.

The numbers are humbling. At Google and Bing, only 10-20% of experiments produce positive results. At Microsoft broadly, about one-third of ideas help, one-third are neutral, and one-third actively hurt. Even the best teams in the world fail most of the time.

And yet: one Bing headline change, initially dismissed as low priority, generated over $100 million in annual revenue when tested. You don’t need every test to win. You need the process to keep running. For more examples like this, see our CRO case studies with real results.

Our take: The companies that get the most from CRO aren’t the ones with the highest win rate. They’re the ones that document every result and never test the same thing twice. A simple spreadsheet works. Date, what you tested, what happened, what you learned. That’s it.

Measuring A/B test conversion rates accurately is its own skill. Kirro handles the math using an approach that works with smaller traffic (experts call this Bayesian statistics). It tells you “Version B gets 23% more signups” instead of making you decode a stats formula. Try it free.

The compounding effect is real. A 5% lift this month, a 3% lift next month, a 7% lift the month after. These stack. After six months of consistent testing, you’re looking at a noticeably better conversion rate.

What the CRO process looks like at different scales

Every CRO article on the internet describes the same process as if you have a full team, a six-figure budget, and enterprise tools. Most of you don’t. So let’s get specific about what the cycle actually looks like at three different scales.

Solo marketer (just you): You’re doing everything between meetings. AI in the CRO process can help here, and tools that auto-generate test ideas or analyze session recordings save hours per cycle. Research means spending 30 minutes in GA4 and reading customer support emails. Your hypothesis backlog is a note on your phone. You run one test at a time, 2-4 week cycles. Time commitment: 3-5 hours per week. Expect 1-2 tests per month. That’s perfectly fine.

Small team (2-3 people): Someone owns the research phase. You keep a shared list of test ideas (a spreadsheet, a Notion board, whatever works). You can run multiple tests per month as long as they’re on different pages. Time commitment: 10-15 hours per week total. Expect 3-5 tests per month.

Agency or in-house team (5+ people): Full research process: structured page audits, analytics deep dives, customer surveys, usability testing. Formal prioritization framework. QA checks before every test goes live. A developer dedicated to building test variations. Expect 8-15 tests per month. If you’re at this scale, you’re building a CRO program, not just following a process.

The biggest mistake? Copying the enterprise playbook when you’re a team of one. Start where you are. One test per month using a simple tool and a simple process beats zero tests per month using a “perfect” process you’ll never actually follow.

For CRO training resources on building your skills at any level, we have a list.

Why most CRO processes fail (and how to fix yours)

The Speero + Kameleoon 2024 benchmark surveyed 206 companies. 47% didn’t have clear program goals. 91% felt underfunded. 95% had no training program. No wonder most CRO processes break down. These are the five ways they fail, and what to do instead.

Mistake 1: Skipping research. You jump straight to “let’s test a new headline” without knowing if the headline is actually the problem. That Optimizely analysis of 127,000 experiments? It found that 88% of ideas teams implement aren’t winners. When you skip research, you’re basically gambling.

Mistake 2: Stopping after one test. CRO is a cycle, not a one-time project. One test tells you almost nothing. The value comes from compounding small wins over months.

Mistake 3: Testing too many things at once. Each test needs enough traffic for reliable results. If you split your traffic five ways, you need five times the visitors. For most sites, that means waiting months for a single result.

Mistake 4: Not documenting results. Six months from now, someone on your team will suggest the exact same test you already ran. Without documentation, you’ll run it again. And waste the time.

Then there’s mistake 5: tracking the wrong metric. The same Optimizely study found that three of the five most commonly used metrics have the lowest impact on business outcomes. Tracking page views when you should be tracking purchases? That’s how you “improve” numbers that don’t matter.

The fix? Follow the five steps. Start small. Document everything. Be patient. And check out proven CRO best practices for the techniques that actually move the needle. If you’re thinking bigger picture, our post on CRO strategy covers the longer-term plan.

Conversion rates dropped 6.1% year-over-year according to Contentsquare’s 2025 benchmark of 6,000 websites. At the same time, the cost per website visit went up 9%. You’re paying more for traffic that converts less. A structured CRO process is how you fight back.

For a visual walkthrough of the full CRO process, Wes McDowell breaks down each step:

FAQ

What are the steps in CRO?

The CRO process has five steps: research (find what’s broken), hypothesize (guess how to fix it), prioritize (pick the most impactful change), test (run an A/B test), and analyze (learn from the results). Then you repeat the cycle. Each step builds on the last, and skipping research, the first step, is the most common reason CRO programs fail.

How long does the CRO process take?

One full cycle takes 4-8 weeks for a solo marketer. That breaks down to 1-2 weeks of research, about a week for hypothesis and prioritization, 2-4 weeks of testing, and a few days for analysis. A team can compress this to 2-4 weeks per cycle. Expect real compound results after 3-6 months of consistent testing. An agency like Conversion.com reports that individual tests typically cost $3,000-$15,000 when done at enterprise scale, but you can run your own for the cost of a testing tool.

CRO process vs CRO strategy: what’s the difference?

The CRO process is the step-by-step workflow for one optimization cycle (what this post covers). A CRO strategy is the longer-term plan: what parts of your site to focus on, how to allocate resources, and how to grow your testing program over time. Process is “how do I run one test well.” Strategy is “what should I test over the next six months, and why.” You need both, but start with the process. You can also read about conversion optimization planning for another angle on this.

What is CRO vs CRM?

CRO (conversion rate optimization) focuses on improving your website so more visitors take action, whether that’s buying, signing up, or requesting a demo. CRM (customer relationship management) focuses on managing relationships with people who are already customers. Different goals, different tools. But they work well together: CRO brings in more customers, CRM helps you keep them.

What does CRO mean in marketing?

CRO stands for conversion rate optimization. It’s the practice of systematically improving the percentage of website visitors who do what you want them to do. If 100 people visit your site and 3 buy something, your conversion rate is 3%. CRO is how you make that number higher. Read what is CRO for the full breakdown, or explore the digital CRO space if you want to see how CRO fits into broader ecommerce CRO strategy.