Conversion rate optimisation (CRO) means getting more of the visitors you already have to take the action you want: enquire, book, sign up or buy. Doubling traffic is hard and expensive; improving the share of visitors who convert is often faster and cheaper. But CRO isn't about changing button colours at random. It's a disciplined process of finding problems, forming hypotheses and testing changes. Here's how it works.
The CRO process
| Step | What you do | Output |
|---|---|---|
| 1. Measure | Make sure conversions are tracked accurately | Reliable baseline |
| 2. Research | Find where and why people drop off | List of problems |
| 3. Hypothesise | Propose changes and why they'd help | Test ideas |
| 4. Prioritise | Rank ideas by impact, confidence and effort | Test roadmap |
| 5. Test | Run A/B tests or make clear fixes | Results |
| 6. Learn | Record what worked and why | Better future ideas |
Step 1: Measure properly
You can't optimise what you can't measure. Track key events such as form submissions, calls, bookings and purchases, plus important steps before them. Our GA4 beginner's guide covers setup.
Step 2: Find the problems
Quantitative research (where)
- Funnel reports: which steps lose the most people?
- Landing page conversion rates: which pages underperform?
- Device comparison: does mobile convert much worse?
- Speed data: are slow pages losing visitors?
Qualitative research (why)
- Session recordings and heatmaps
- On-page surveys: "What's stopping you from booking today?"
- Customer interviews and sales call notes
- Usability testing with real people
Step 3: Write proper hypotheses
A good hypothesis links a problem, a change and an expected result:
Because session recordings show visitors scrolling past the pricing without clicking, if we add a clear "Get a quote" button beside each price, then quote requests will increase.
Step 4: Prioritise
A simple scoring model, such as ICE, keeps prioritisation honest:
| Idea | Impact (1–10) | Confidence (1–10) | Ease (1–10) | Score |
|---|---|---|---|---|
| Shorten quote form from 9 to 4 fields | 8 | 7 | 8 | 23 |
| Add reviews beside the call to action | 6 | 6 | 9 | 21 |
| Redesign the homepage hero | 7 | 4 | 3 | 14 |
Step 5: Test or fix
When to A/B test
Test when you have enough traffic to reach a reliable result and the right answer genuinely isn't obvious.
When to just fix
If something is clearly broken or confusing, such as a form error, a slow page or a missing price, fix it. Testing obvious problems wastes time.
Running A/B tests properly
- Test one meaningful change at a time
- Decide the sample size and duration before starting
- Run tests for full weeks to avoid day-of-week effects
- Don't stop early because one version is ahead
- Look at the primary goal, not just clicks
High-impact areas to check first
| Area | Common problems | Typical fix |
|---|---|---|
| Headline | Vague, about you rather than the customer | Specific outcome for a specific audience |
| Call to action | Generic "Submit" or hidden below the fold | Clear, specific button repeated down the page |
| Forms | Too many fields, poor errors | Fewer fields, helpful validation |
| Trust | No reviews, logos or guarantees near decisions | Proof placed beside the action |
| Speed | Slow mobile pages | Optimise images and scripts |
| Pricing | Unclear or hidden | Transparent pricing or clear ranges |
| Mobile | Hard to tap, cramped | Mobile-first layout |
Our guides to form design best practices and landing page elements that convert go deeper on each.
Common CRO mistakes
- Copying "best practices" without understanding your users
- Testing tiny changes on low-traffic pages
- Ending tests early, or declaring winners without enough data
- Optimising for clicks instead of revenue or qualified leads
- Ignoring mobile visitors
- Using manipulative tactics that hurt trust; see our guide to dark patterns
Keep a learning log
Record every test: the hypothesis, what changed, the result and what you learned. Over time, this becomes your most valuable marketing asset, showing what actually persuades your customers.
A worked example: a service quote page
A home improvement company's quote page gets plenty of traffic but few requests.
| Research finding | Hypothesis | Change |
|---|---|---|
| Recordings show people scrolling past the form | The form looks like hard work | Cut fields from nine to four |
| Survey: "Not sure how much it costs" | Uncertainty about price stops enquiries | Add typical price ranges |
| Few reviews visible near the form | Lack of trust at the decision point | Add three recent reviews beside the form |
| Mobile conversion far below desktop | Form hard to use on phones | Larger fields, single column, tap-to-call option |
Rather than testing each change separately on a low-traffic page, the team made the clear fixes together, then tested the headline, where the right answer was less obvious.
Sample size basics
- Small improvements need large samples to detect reliably
- Pages with only a handful of conversions a week rarely support meaningful A/B tests
- Use an online sample size calculator before starting, based on your current conversion rate and the improvement you hope to detect
- If a test would take months, focus on research and best-practice fixes instead
Testing without enough data creates false confidence, which can be worse than not testing at all.
How we help
Our usability testing uncovers why visitors hesitate, checkout optimisation fixes the most expensive leaks in online stores, and landing page design builds pages designed to convert from the start.
The short version
Track conversions accurately, use data to find where people drop off and research to understand why, write clear hypotheses, prioritise by impact and effort, test properly or fix obvious problems, and record what you learn. CRO is a steady process, not a one-off redesign.
