What is conversion rate optimization?
Conversion rate optimization (CRO) is the systematic work of getting more website visitors to take a desired action. You measure it with the conversion rate: conversions divided by visitors, times 100.
A conversion can be:
Hard conversions (revenue)
- • A purchase or order
- • A signed contract
- • A demo booking with intent to buy
Soft conversions (leads)
- • A newsletter sign-up
- • A white paper download
- • A contact request
Why conversion rate optimization matters:
A higher conversion rate works on every visitor you already have. More traffic, by contrast, costs money, often brings visitors who aren’t a good fit, and leaves the same questions unanswered, just for more people. That’s why more traffic only pays off once the page answers those questions.
Example with made-up data: 10,000 visitors × 2% conversion rate × $500 per deal comes to $100,000. At 3% it would be $150,000, without a single extra visitor.
What is a good conversion rate?
There is no number that applies to everyone. Benchmarks from other industries don’t help much, because they mix visitors with different intentions, offers and goals. Put your rate in context with these factors instead:
| Factor | Why it shifts the rate | What you compare |
|---|---|---|
| Traffic source | Someone searching for you on purpose is further along than someone who stumbles onto an ad. | Rate per channel, not across all channels |
| Conversion goal | A download is a smaller decision than a demo or a purchase. | Rate per goal, kept separate |
| Page type | A landing page with one goal measures something different from a whole website. | Landing page against landing page |
| Offer and price | The bigger the decision, the longer it takes. | Your own rate over time |
| Next step | Lots of inquiries don’t help much if hardly any of them lead to a sales conversation. | Share of inquiries that become sales conversations |
Important in B2B:
In B2B, the quality of leads often matters more than their number. So also measure how many leads turn into sales conversations and closed deals.
Why visitors don’t convert: three reasons
It is rarely the colors or the form fields. Look for these three reasons first, then pick the fix.
An unclear value proposition
At first glance, visitors don’t understand what you offer and why it matters to them.
Signs: high bounce rate, little scrolling
Check: “Could someone from another industry tell what problem we solve after one look at the first screen?”
Unanswered objections
Visitors are interested, but the page doesn’t answer their questions, for example about effort or fit.
Signs: long time on page, but no inquiry
Check: “Which objections do we hear in first calls, and where does the page answer them?”
The wrong order
The information is there, but details come before the problem. Or the page asks visitors to get in touch before it has earned their trust.
Signs: visitors read a lot but don’t get in touch
Check: “Does the structure follow the buyer’s decision or our product structure?”
How to increase your conversion rate in five steps
If you want to raise your conversion rate, don’t start with colors or buttons. Look for the point where interested visitors get stuck. Here’s how:
- Record the baseline. Write down the conversion rate per channel, page and device, using the formula above. Without a baseline you can’t tell later whether a change made a difference.
- Pick the right page. Start where many visitors arrive and few act. A change to that page reaches the most people.
- Find the cause. Web analytics, heatmaps and session recordings show where visitors drop off and what they click. They don’t show why visitors leave. For that, ask your sales team about the objections they hear in first calls, and let personas from your audience tell you where they would drop out.
- Make a targeted change and check it first. Write a hypothesis: “If we change X, Y will happen, because Z.” Change one thing per cause, for example the headline or a CTA that names the next step, so you can later tie a result to a cause. Check the new version against the current one before it goes live.
- Measure live and write it down. If you have enough traffic, run an A/B test until the result is solid. Otherwise compare equally long periods before and after the change, with the same channels. Note what worked and why.
The next round then builds on what this one showed. That is why conversion rate optimization has no end date.
Eight conversion rate optimization techniques
Which fix makes the biggest difference for you depends on where your visitors get stuck. Check these eight first:
Make the call to action (CTA) clearer
The CTA says what happens after the click. A label that names the action and what you get from it leaves fewer questions open than “Submit”.
Shorten your forms
Every field is a small hurdle. Only ask for what you really need for the next step.
In B2B, name, email and company are often enough for a first inquiry. Cover the rest once you’re talking.
Build trust with proof
In B2B, trust is part of the decision. Show that others already trust you, with proof people can verify.
- • Customer logos, if you are allowed to show them
- • Testimonials with name and title
- • Certifications (ISO, SOC 2, GDPR documentation and so on)
- • Case studies with numbers people can follow
Check your page speed
Anyone waiting for the page isn’t reading yet. Check load times especially on phones and over real mobile connections.
Sharpen the headline and value proposition
The headline sits at the top and gets read first. It should say what you offer and what the reader gets from it.
Weak: “Welcome to Company XY”
Stronger: “Rate your suppliers, from $89 a month” (made-up example)
Use psychological triggers honestly
Triggers can make decisions easier. Only use them when they are true: a false trigger costs trust.
- • Social proof: only with real numbers and real customers
- • Scarcity: only when the offer really is limited
- • Urgency: only with a real date
Check the mobile view
On a phone, much of what sits on the first screen on desktop slides further down. Compare your conversion rate by device.
Check for large tap targets, readable type, short forms, and whether the next step is visible without scrolling.
Answer objections before anyone raises them
Every visitor brings concerns. Answer the most common ones right on the page.
- • “Is this secure?” → a section on security and certifications
- • “What does it cost?” → prices or “from $X”
- • “Is this right for us?” → examples from the reader’s industry
Conversion rate optimization before launch and with low traffic
The usual CRO tools assume that a page already has visitors. Heatmaps and session recordings analyze what real visitors do, and an A/B test splits them between two versions. A new page doesn’t have those visitors yet. Many B2B pages, pages for new products and short-lived campaign pages have too few to finish a test in a reasonable time.
How much traffic an A/B test needs: an example calculation
The number of visitors you need depends on two values: your current conversion rate and the smallest difference the test should still reliably detect. For this example we assume a conversion rate of 3% and ask whether a new version reaches 3.6%, a fifth more. Both values are assumptions for the calculation, not industry benchmarks.
How the sample size is calculated (example)
The usual setup is a two-sided test with a 5% significance level and 80% power. These correspond to z-values of 1.96 and 0.84. Their sum is squared: 2.8 × 2.8 = 7.84. Multiply that by the variance of both versions. For each version, that’s the rate times one minus the rate, and the two are added together: 0.03 × 0.97 + 0.036 × 0.964 = 0.0638. Divide the result by the squared difference between the two rates, here 0.006 × 0.006 = 0.000036.
The result is 13,896 visitors per version, 27,792 in total. The current version would get about 420 conversions out of those. At 1,000 visitors a week the test runs for about 28 weeks. If it only needs to detect a large difference, 4.5% instead of 3% (half again as much), 2,512 visitors per version are enough, about 5 weeks.
Formula for comparing two proportions per Kim 2016 (retrieved October 11, 2026). Online sample size calculators sometimes use other formulas and arrive at slightly different numbers. To run it with your own values, use the A/B test calculator, which also checks whether a finished test is statistically significant.
Before the test, you don’t know how big a difference your new version will make. So with little traffic, a test often detects only large differences within a reasonable time, or it ends without a clear result. You can still optimize, just earlier: before the page gets visitors, look for the places where readers would drop out. There are four ways to do that without traffic.
What you can do before launch
Check message match against the search
People who come from a search or an ad expect to see what they searched for on the first screen. Put the keywords you bid on or want to rank for, along with your ad copy, next to the headline and intro. Do they state the searcher’s question in the searcher’s words? For an existing page, Google Search Console shows which queries bring visitors.
Limit: you learn whether the page matches what visitors intend. It doesn’t show whether the page convinces them.
User tests with five people
You sit people from your audience in front of the page, give them a task and listen while they think aloud. Jakob Nielsen calculated that five test users find about 85% of the usability problems in an interface. He recommends several small rounds with a revision in between instead of one big one (Nielsen Norman Group, 2000, retrieved October 12, 2026).
Limit: the 85% refers to usability problems. Five people can only give you a rough sense of whether an offer is convincing. Recruiting and scheduling take time.
Expert review
Someone with CRO experience who didn’t write the page goes through it with a checklist: Is the value proposition clear on the first screen? Are the common objections answered? Can the proof be checked? Is the next step small enough? That finds obvious gaps quickly. A checklist for this: landing page optimization checklist.
Limit: it is still one person’s judgment, and even an expert knows how your audience reads only from experience.
Simulation with personas from your audience
99takes is a tool you use in Claude to check a page before launch. Your audience consists of 10 to 14 personas, researched from public customer voices where they exist and from your competitors’ copy. They read your page or two versions and say why they would act or not. That also works with a draft that isn’t online yet.
What you get for your page:
- • A score and the main objection per persona
- • Whether each persona would take the next step, and if not, why
- • Findings per section: what works, what holds readers back, what is missing
- • With two versions: which one your core audience prefers. You decide who your offer is for (the core audience) and who just reads along (the extended audience); the verdict follows the core.
- • If you like, a challenger version that 99takes writes and tests against your current one
Limit: 99takes does not measure your real conversion rate. It tells you where readers from your audience get stuck and why. If you have enough traffic, the A/B test remains the proof; the check comes first and tells you which version is worth testing. It won’t help much while your offer itself is still unclear.
Once the page is live, measure as described in step five: with an A/B test as soon as the traffic is there, otherwise by comparing equally long periods before and after the change. Other methods besides the A/B test, and what each one tells you, are covered in A/B test alternatives. How simulated personas compare with real users: synthetic users.
Check your page before launch
Personas from your audience read the page or two versions and say where and why they would drop out. In Claude, with 350 credits to start, valid for 30 days, no payment details.
Create an account and check your pageWhich CRO tools do you need?
Each kind of tool answers a different question. Most teams need more than one.
Analytics
- • Web analytics: traffic and conversions per channel and page
- • Heatmaps and recordings: where people click and scroll (see Hotjar alternatives)
A/B testing
- • Testing tools: compare two versions with real traffic
- • Requirement: enough visitors for a solid result
- • Tells you: which version wins, not why
User research
- • On-page surveys: what visitors are missing
- • Interviews and usability tests: how real users work with the page
- • 99takes: personas from your audience check versions before launch
Which tool fits when
The tool you pick depends on how many visitors the page gets and what you want to know: why readers don’t act, or what a change does to the result.
| Traffic | Why don’t readers act? | What does a change do? |
|---|---|---|
| Before launch | Message match with the search, user tests with five people, expert review, simulation with personas (99takes) | Not measurable yet, since there are no visitors |
| Low traffic | The same methods, plus on-page surveys and the objections from first calls | Web analytics and a comparison of equally long periods before and after the change, with the same channels |
| From a few thousand visitors a month (rule of thumb) | Heatmaps and session recordings, surveys; simulation to filter out weak versions before the test | An A/B test, if the required sample size is reached in a time you can wait for |
The levels are a rule of thumb, not a fixed threshold. Whether your traffic is enough for a test depends on your conversion rate and on the difference you are looking for. Work it out before you start, as in the example above or with the A/B test calculator.
Common conversion rate optimization mistakes
Changing and testing without knowing the reason
Testing buttons and colors without knowing why visitors don’t get in touch is optimizing blind. Without a hypothesis, an A/B test is a coin toss.
B2C tactics in B2B
Countdowns and “only 3 left” are familiar from online retail. Ask yourself whether someone buying expensive software believes a countdown. You don’t have to guess whether clarity or pressure works better with your buyers: put both versions in front of your audience first.
Changing everything at once
If you swap the headline, form and images in one step, you won’t know afterwards what worked.
Why is conversion rate optimization different in B2B?
B2B buying decisions work differently from consumer purchases:
| Aspect | B2C | B2B |
|---|---|---|
| Who decides | Usually one person | Often several roles |
| Length of the decision | Fairly short | Fairly long, with approvals |
| What matters | Often spontaneous; personal benefit | Carefully weighed: benefit, cost, risk |
| Traffic | Often high | Often low to medium |
| Value of a deal | Lower | Higher |
What this means for CRO:
In B2B it isn’t enough to fine-tune the CTA. You need to know who comes to your page (leadership, IT, finance?) and which objections each role brings. How to describe these roles as personas: audience research.
One landing page or the whole website?
| Situation | Approach |
|---|---|
| Lots of traffic, few inquiries on one page | Optimize that landing page |
| Few inquiries across many pages | Optimize the whole website |
| New campaign or new product | Build a dedicated landing page, for example from a template in the funnel library |
| Existing pages with an unclear message | Test the message, then revise where it matters |