Two companies buy the same clicks at the same price. One turns 2% of visitors into leads, the other 4%. The second pays half as much for each lead, and it can afford to bid higher, win more clicks and grow while the first one wonders why ads stopped working. That gap is the whole case for conversion rate optimization, usually shortened to CRO.
This guide explains what CRO is, how to measure it, the process professionals follow, what to test first and how to know a result is real. It is written for people who pay for traffic, and it is built from the searches people make, including "what is conversion rate optimization" (about 1,300 US searches a month in Semrush), "conversion rate optimization services", "conversion rate optimization agency", "conversion rate optimization tools" and "conversion rate optimization best practices". Sources are listed at the end.
What is conversion rate optimization
Conversion rate optimization is the systematic process of increasing the percentage of visitors who take a desired action. The action, called a conversion, can be a purchase, a booking, a form submission, a call, a sign up or a download. Google Ads describes a web conversion as a specific action users take on your website after interacting with your ad, and says measuring it gives insight into user behaviour and campaign performance.
CRO has two halves that are often confused. The first is research, finding out why people do not convert. The second is testing, changing something and measuring whether it helps. A redesign based on taste is not CRO, and neither is a test with no research behind it. The skill is in connecting the two.
How to calculate conversion rate
The formula is simple. Conversion rate equals conversions divided by visitors, multiplied by 100. The details matter more than the formula.
| Decision | Options | Advice |
|---|---|---|
| What counts as a conversion | Purchase, qualified lead, any form, call, click to call | Pick the one tied to revenue. Track micro conversions separately. |
| What counts as a visitor | Users, sessions or ad clicks | Use clicks for ads and sessions for pages, and do not mix them in one comparison. |
| Which traffic | All traffic or one source | Measure paid traffic separately. Branded and non branded convert very differently. |
| Which device | Desktop, mobile | Report both. A blended rate can hide a broken mobile page. |
Example. A plumber's landing page receives 1,500 ad clicks in a month and produces 45 calls and form leads. The conversion rate is 3%. If the ad spend is $3,000, the cost per lead is $66.67. Lifting the rate to 4% brings 60 leads from the same spend, a cost per lead of $50. That is a 25% cut in cost per lead without changing the ad.
What is a good conversion rate
There is no single good rate. The honest answer is that it depends on traffic source, intent, price and offer. Benchmarks give context, and your own history gives direction. LocaliQ and WordStream report 2026 Google search ad averages for each industry, which are shown below as a starting point.
| Industry (Google search ads) | Conversion rate | Cost per lead |
|---|---|---|
| All industries | 8.18% | $66.69 |
| Personal services | 12.34% | $54.60 |
| Physicians and surgeons | 12.43% | $40.04 |
| Dentists | 10.67% | $72.97 |
| Home and home improvement | 8.05% | $90.92 |
| Attorneys | 5.55% | $131.63 |
| Real estate | 3.70% | $102.51 |
| Finance and insurance | 2.64% | $74.44 |
On Facebook, the same source reports a 2026 average lead ad conversion rate of 8.54% and a cost per lead of $27.39, but the leads come from people who were not searching, so quality varies. The rates above are for ads and cannot be compared with organic, email or direct traffic. Compare yourself with your own last quarter and with your best campaign before you compare with an average.
The CRO process, a picture
Step 1, make sure conversions are measured
No test can work with bad numbers. Before anything else, confirm that each conversion fires once, on the right action, and that its value or lead quality is recorded. Test every form and call button yourself, check that thank you pages cannot be reached by accident and that the numbers in analytics match the numbers in your ad platform and your own records. Our guide to the Meta Pixel and the Google Ads conversion setup help page cover the common tracking checks.
Step 2, research why people do not convert
Use several sources, because each one hides something.
| Source | What it shows | Watch for |
|---|---|---|
| Analytics funnels | Where visitors drop out, by device and source | Averages that hide one weak segment |
| Session recordings and heatmaps | Where people hesitate, rage click or miss a button | Watching too few sessions to see a pattern |
| Customer and lost lead interviews | The words people use and the doubts they have | Asking leading questions |
| On page surveys | Why visitors did not complete a task | Low response rates and vague answers |
| Support and sales calls | Questions the page does not answer | Relying on one person's memory |
| Page speed and errors | Slow, broken or unstable pages | Testing only on a fast office network |
| Search terms and ad copy | Whether the page matches the promise of the ad | Sending every ad to one generic page |
For paid traffic, check message match first. A visitor who clicked an ad about same day boiler repair and lands on a general heating page has to work to find what they were promised. Matching the headline, offer and next step to the ad is often the largest single gain, and it costs nothing but time. Our guide to writing search ads helps keep the promise and the page aligned.
Step 3, write a hypothesis
A hypothesis links an observation to a change and an expected result. Use this form. Because we saw this evidence, we believe changing this element for this audience will move this metric, and we will know when it reaches this result.
| Part | Example |
|---|---|
| Evidence | Recordings show mobile visitors scrolling past the form, and mobile converts at half the desktop rate |
| Change | Move a short form above the fold on mobile and add a tap to call button |
| Audience | Mobile visitors from paid search |
| Metric | Lead conversion rate, with cost per lead as the guardrail |
| Expected result | A relative lift of 15% or more |
Step 4, decide what to test first
Most teams have more ideas than traffic. Score each idea on how much it could move the metric, how confident you are from the evidence and how easy it is to build, and test the highest scores first.
| Idea | Impact (1 to 5) | Confidence (1 to 5) | Ease (1 to 5) | Score |
|---|---|---|---|---|
| Match headline to the ad keyword | 4 | 5 | 5 | 14 |
| Shorten the form from seven fields to four | 4 | 4 | 5 | 13 |
| Show total price early in checkout | 5 | 4 | 3 | 12 |
| Add real reviews near the button | 3 | 3 | 4 | 10 |
| New hero image | 2 | 2 | 4 | 8 |
The scores are examples. Use them to compare ideas against each other, not to predict a lift.
What to test, ideas that tend to matter
Offer and message
- The headline and the first sentence, which should say what you offer, for whom and what happens next.
- The offer itself, such as a free quote, a call back, a fixed price, a guarantee or a bundle.
- Specifics over claims. A named result, a price range or a time frame beats a general promise.
Trust and proof
- Reviews, ratings and case studies close to the action.
- Licences, accreditations and guarantees that are true and checkable.
- Real photos of people and work instead of stock images.
Forms and steps
- Fewer fields, with the rest asked later or by phone.
- One clear primary button with specific text, such as Get my quote.
- Click to call on mobile and a visible phone number for urgent services.
Checkout and cart for stores
Baymard's 2026 survey of 1,083 US adults who shopped online found these stated reasons for abandoning a purchase, extra costs such as shipping were too high (40%), having to create an account (18%) and a checkout that was too long or complicated (17%). Those are testable. Show total cost earlier, offer guest checkout and cut fields and steps. Baymard puts the average documented abandonment rate at 70.22% across 50 studies, which shows how much revenue sits in the cart before ad spend is even counted.
Speed and stability
Slow, shifting pages lose people before they read a word. Google's web.dev defines good Core Web Vitals as a Largest Contentful Paint of 2.5 seconds or less, an Interaction to Next Paint of 200 milliseconds or less and a Cumulative Layout Shift of 0.1 or less, measured at the 75th percentile. Fix these before you test copy, because a speed problem hides the effect of everything else.
A calculator for test size and what a lift is worth
Many tests fail because they were too small to detect the change they were looking for. This calculator estimates how many visitors each version needs, using a standard two sided test for two proportions, and shows what the lift would be worth. Enter your own figures. Defaults are examples.
A/B test size and lift value calculator
With the example numbers, a page converting 3% needs about 13,900 visitors per version to detect a 20% relative lift with 95% confidence and 80% power. At 400 visitors a week that is more than a year, which is the real lesson. Low traffic pages cannot support small tests. Either test bolder changes that could produce larger lifts, combine traffic from similar pages or use qualitative research and sound judgment instead of a formal test.
How to read a test result
| Rule | Why |
|---|---|
| Set the sample size and duration first | Stopping when the result looks good inflates false winners |
| Run whole weeks | Behaviour differs between weekdays and weekends |
| Change one idea at a time where traffic is limited | Otherwise you cannot tell what worked |
| Check the guardrails | A lift in leads that cuts lead quality or sales is not a win |
| Segment carefully | Splitting results into many segments after the fact finds patterns that are noise |
| Record flat and losing tests | They prevent repeating mistakes and often hold the best insights |
Most tests do not win. A programme is judged on what it learns and on the wins that stick, not on every test producing a lift.
CRO for different business types
| Type | Main conversion | First places to look |
|---|---|---|
| Local service business | Call or quote request | Message match, mobile call button, short form, reviews, speed to reply |
| Ecommerce and Shopify | Purchase | Product page detail, shipping costs shown early, guest checkout, payment options, speed |
| B2B and SaaS | Demo or qualified lead | Clear value statement, proof, form length, pricing clarity, follow up speed |
| Lead generation from Facebook ads | Qualified lead | Form questions, instant form versus landing page, call back speed, see Facebook lead ads |
| Content and subscription sites | Sign up | Offer, placement, friction at sign up and email onboarding |
Conversion rate optimization tools
People search for CRO tools about 1,600 times a month in the US, according to Semrush. Tools support the process, they do not replace it. Pick them by the job.
| Job | What to use | Notes |
|---|---|---|
| Measurement | Google Analytics, your ad platforms and server or tag manager setups | Free and necessary. Check consent and tracking quality. |
| Behaviour | Heatmap and session recording tools | Several have free tiers. Mask sensitive fields and follow privacy law. |
| Feedback | On page surveys and post purchase questions | Ask one or two questions, not ten. |
| Testing | A/B testing platforms or built in tests in your page builder or ad platform | Choose one that handles sample size and flicker properly. |
| Speed | PageSpeed Insights and Search Console Core Web Vitals report | Use field data where available. |
| Forms and calls | Form tools with tracking, call tracking | Make sure each conversion is counted once. |
CRO services, agencies and consultants
Searches for conversion rate optimization services, agencies, companies and consultants add up to many thousands a month, so the market is crowded. Choose on process and evidence, not on promises. Ask the following questions.
| Question | A good answer |
|---|---|
| What will you do first? | Audit tracking, then research, then a ranked list of tests |
| How do you decide what to test? | By evidence and a scoring method, not by opinion |
| How do you know a result is real? | Pre set sample size, duration and guardrail metrics |
| What happens with low traffic? | Bolder changes, qualitative research and careful before and after reviews |
| Who owns the data and tests? | You do, including accounts and test records |
| What results can I verify? | Examples in your sector with the method behind them, not only percentages |
Be wary of guarantees of a specific lift, case studies with no sample size and any company that cannot explain how it measures. Our landing page and conversion optimization service works from tracking and research first, and the free audit is a low pressure first step. For the wider picture of how conversion work fits with paid media, see our guide to performance marketing.
CRO and paid advertising
Ads and conversion work should be planned together. The ad decides who arrives, and the page decides whether they act. Three practices help.
- One page per intent. Send each ad group to a page that matches its keywords and offer, not to the home page.
- Feed back conversion data. Import clean conversions into Google Ads so automated bidding learns from real leads. See our PPC guide and the cost breakdown in Google Ads cost.
- Judge by cost per qualified lead or sale. A page that doubles form fills but halves lead quality has not improved anything.
Does CRO help SEO
CRO is not an SEO tactic, but the two overlap. Google's guidance for creating helpful content asks whether a page gives a complete, original answer for the person who arrived and whether the experience is good. Faster pages, clearer layouts and pages that answer buyer questions tend to help both visitors and search. Be careful with tests on organic landing pages. Keep the page crawlable, avoid hiding content from Google and use proper redirects for split URL tests. Our guide to high bounce rates covers causes that affect both.
Common conversion rate optimization mistakes
- Testing before checking tracking. Wrong numbers lead to wrong decisions.
- Copying competitors. What works for another audience and offer may not work for yours.
- Stopping tests early. It produces false winners.
- Changing many things at once. You cannot tell what caused the change.
- Ignoring mobile. Most paid traffic arrives on phones and many pages are tuned for desktop.
- Sending all traffic to the home page. The ad promise and the page do not match.
- Measuring only form fills. Lead quality and sales are what matter.
- Treating CRO as a project. It is a habit of research, testing and learning.
A 90 day CRO plan
| Days | Focus |
|---|---|
| 1 to 15 | Audit and fix tracking, define one conversion per page, get a baseline for each device and source, fix speed and broken elements |
| 15 to 45 | Research with analytics, recordings and customer questions, write hypotheses, score and rank them, launch the first test or the first bold change |
| 45 to 90 | Read results with the rules above, roll out winners, record what you learned, run the next two or three tests and report cost per lead or sale alongside conversion rate |
Checklist
- One clear conversion defined for each key page and tracked correctly
- Baseline rates by device and traffic source
- Headline and offer match the ad or search that brought the visitor
- Mobile layout, tap to call and short forms tested on a real phone
- Core Web Vitals in the good range, measured in the field
- Proof near the action, such as reviews, results and guarantees that are true
- Hypotheses written with evidence and ranked by a method
- Sample size and duration set before a test starts
- Lead quality and sales checked, not only form fills
- Results and lessons recorded for the next test
Sources
- LocaliQ and WordStream: Search Advertising Benchmarks for Every Industry (2026). localiq.com
- LocaliQ: Facebook Ads Benchmarks (2026). localiq.com
- Baymard Institute: Cart Abandonment Rate Statistics. baymard.com
- web.dev: Web Vitals. web.dev
- Google Ads Help: Set up your web conversions. support.google.com
- Google Search Central: Creating helpful, reliable, people-first content. developers.google.com