Your First Steps in CRO

Your first steps in conversion rate optimization (CRO) are to set realistic expectations and then put five things in place: customer insights, quality traffic, testing tools, a testing framework and patience. CRO is not about changing website elements at random. It is a data-driven, systematic process that improves the user experience so that more visitors convert.

This guide covers where to start. If you need the definition first, read what conversion rate optimization is. If you already know the basics and want a list of things to check on your site, use the conversion rate optimization checklist.

In short, getting started looks like this:

1. Identify key performance metrics. Define what success looks like, for example higher sales or fewer visitors who leave the checkout.
2. Analyze user behavior. Use analytics, heatmaps and session recordings to find friction points.
3. Formulate hypotheses. Write testable ideas based on what the data shows.
4. Run A/B tests. Compare variations to find out which one performs best.
5. Measure and iterate. Implement what wins, learn from what loses and test again.

What is CRO not?

There are still many misconceptions about what CRO is and what it can do for you. Clear them up before you start, or your first campaign will probably end in disappointment.

1. CRO is not a list of "best practices"

If you read marketing blogs, you have seen articles on "CRO best practices" backed by single case studies. They show how a standard change, such as a new button color or a larger headline, led to a dramatic lift in conversions. This gives the impression that CRO is a list of best practices, and that if you apply them, you will beat your competitors.

Some guidelines do apply to almost every website, such as making the call to action prominent and keeping a clear visual hierarchy. But there are no fixed best practices that guarantee the result you want.

Take button color. By color psychology, green is easy to process and is associated with "go" because of traffic lights. Red is associated with stopping and with warnings. So you might conclude that a green button is a "best practice".

Yet in a test that HubSpot reported for Performable in 2011, a red button got 21% more clicks than a green one. Look closer and the reason is not the color itself. The original green button was similar to the green in the Performable logo and the rest of the page. The red button was the only red element, so it stood out.

The general principle is this: a call to action that stands out on the page gets more clicks, whatever its exact color. Read every best practice the same way, as a hint at a general principle, not as a result to copy.

2. CRO is not a magic bullet

Omniconvert case studies include results such as 60% more mobile app downloads for OLX and 20% more leads for ING. Results like these can make CRO look like a one-size-fits-all solution to every marketing problem.

In reality, CRO is only one part of your marketing, sales, product and customer success strategy. Without the other parts, conversion rate optimization will not produce results. Revenue depends on three levers:

Worked example: you sell a $10 product to 10,000 visitors a month, and 2% of them buy. 10,000 × 2% × $10 = $2,000 a month. To reach $10,000 a month, you have three options, or a mix of them:

Which lever you pull depends on your core strengths, budget and existing marketing. A business that is already good at generating traffic might focus on bringing in more qualified visitors. A business that can sell upsells might raise the average order value. For most businesses, the realistic plan is the combined row: a little more traffic, a higher price or order value, and a better conversion rate.

3. CRO is not quick

Many businesses start a CRO campaign with a lot of enthusiasm and stop after a couple of weeks without results. A well-run campaign needs two things:

• Traffic to test the variations.
• Time to send enough of that traffic to each variation.

Only when you have both can you call a variant "better" with statistical confidence. The more traffic a test gets, the faster you get a result. A page with 10 visitors a day needs weeks to collect a useful sample for even two variants, while a page with thousands of visitors a day can finish the same test in days. Calculate the sample size you need before the test starts, so you know how long it must run.

Complexity adds time too. A blogger who tests post headlines can see results within days. A large eCommerce store with hundreds of elements to test, such as product descriptions, buttons and navigation, will likely need weeks or months.

4. CRO is not consistent across websites and verticals

This is a fundamental rule of CRO: every niche and every website is different. A tactic that works on one site will not necessarily work on another.

For example, when the retailer StacksAndStacks.com added videos to its product pages, Internet Retailer reported that shoppers who watched a video were 144% more likely to add the product to their cart than shoppers who did not. From a marketer's point of view, this makes sense: a video shows the product as it really is.

Another test showed the opposite. Brookdale Living, a senior living provider, tested a landing page with a static image against a version with a video in the same place. You might expect the video to win, but the image version converted better, a lift estimated to be worth about $106,000 in extra monthly revenue.

There could be many reasons for this result. The point is that you cannot assume anything in CRO. A practice that is proven on one website can fail on another. The only solution is to build variants on sound principles and test them without assumptions.

What does the CRO process look like?

CRO is not something you do a few times and forget once you hit a goal. Your prospects' wants change and new roadblocks can appear suddenly. You need to respond to these changes quickly, so keep testing and refining for as long as your website is live. You can divide the CRO process into three stages:

1. Discovery. Understand your audience through quantitative and qualitative data, and build an optimization plan.
2. Testing. Run experiments on the hypotheses you formulated during discovery.
3. Review and analysis. Analyze the test results, implement the changes that won, and feed what you learned back into discovery.

For a step-by-step walkthrough of the testing stage, read how to create an A/B test.

What do you need for your first CRO campaign?

To maximize conversions, you must be able to form strong hypotheses from quantitative and qualitative data, create page variations based on them, run each test long enough to get actionable results, and analyze the results so you can act on them. You cannot optimize pages blindly. First find out what your audience actually cares about, then design the shortest path to that experience.

1. Customer insights

What does your audience want? Which pages does it land on? What stops it from completing a purchase? These are the key questions in CRO, and answering them is the first thing you need. Starting a campaign without understanding your audience wastes time and money. There are two types of insights to gather:

• Quantitative insights answer what, where and how much.
• Qualitative insights answer why.

Quantitative insights show where you lose visitors in your conversion funnel. With analytics data (for example, from GA4), heatmaps and session recordings, you can see what visitors do on your site: how long they stay on a page, where they come from, where they leave, and which pages are most popular. Quantitative data shows where problems occur, but not why they occur.

Think of a doctor who checks your pulse and takes your temperature. These measurements do not tell the doctor what illness you have, but they show that something is wrong. To find the cause, the doctor must ask you questions and look at your medical history. In CRO, that is qualitative data.

Qualitative insights explain the reasons behind user behavior. They give you a direct line to your visitors and answer questions such as:

• Why do people click away from the sign-up form?
• Why do so many buyers abandon their carts?
• What do customers actually want when they land on this page?

Two of the most popular methods to collect qualitative data are:

• Interviews. Talking directly to customers is the best way to find the gap between your marketing message and what customers want to hear. Interviews can show fundamental flaws in your value proposition.
• Customer surveys. Feedback at scale shows what kind of traffic you attract, what visitors want, the words they use and the doubts they have. These insights help you remove friction from your conversion process.

Keep in mind that not every idea you test will succeed, even when it is based on good insights. And you cannot run a test without traffic.

2. Quality traffic

When you test elements on your site, you want to find the winning variation in a reasonable time. To get statistically significant results, your site needs a steady flow of visitors to the page you test, usually at least a few thousand unique visitors a week. If you test several variations at once, the traffic you need rises significantly.

If you do not have this traffic, you can still run A/B tests by buying traffic. Include the cost of that traffic in your CRO budget.

Traffic quality matters as much as volume. Traffic is the start of the whole conversion funnel. If its quality is low, no amount of optimization will grow conversions. A business that gets organic traffic from a specific, high-intent search (such as "buy auto insurance in New York") will naturally convert better than one that gets generic social media traffic, even with a weaker design. So set your conversion goals by traffic source:

If you buy traffic for your tests, check that you can keep a similar traffic quality after the test ends. If you cannot, your results may not hold once the paid traffic stops.

3. Testing tools

CRO tools are now within reach of average businesses, for everything from collecting customer insights to running split tests. They fall into two broad categories.

All-in-one tools are built specifically for CRO and combine several functions. Features vary, but they usually include at least a way to collect qualitative insights with surveys, a way to run A/B and multivariate tests, and a way to analyze results and segment audiences. Omniconvert Explore is an example: it has built-in tools for surveys, tests, audience segmentation and personalization, plus a visual editor so you can change text and images and move elements without a developer. You will still need an analytics tool such as GA4 for quantitative data.

Standalone tools do one job each, and you connect them to run a campaign. For example, you could run user surveys in Typeform, compare the answers with GA4 data, build landing page variants in Leadpages, and split traffic between them with a testing tool. Many standalone tools are also useful outside CRO.

Whether you choose an all-in-one tool or standalone tools depends on your preferences, budget and marketing needs. There is no one-size-fits-all answer.

4. A testing framework

Most successful CRO campaigns follow a testing framework, a short, actionable version of the CRO process. It is described in the next section.

5. Patience

CRO is a long-term process, and many tests will not produce a winner. The last section explains how to plan for that and what to do when a test fails.

How do you build a CRO testing framework?

1. Identify your goals. Define your business goal, the marketing goal that supports it, and the conversion goal that supports the marketing goal.
2. Find the reasons you are not meeting them. Gather and analyze quantitative and qualitative data.
3. Write hypotheses. For each problem you found, propose a change that should solve it.
4. Create variants and run split tests. Test each hypothesis against the current page.

Example: an online shoe store

Suppose you run an online store that sells shoes. You set your goals like this:

• Business goal: double shoe sales on the site.
• Marketing goal: raise the conversion rate from 1% to 2%.
• Conversion goal: add larger images, improve the checkout process and reduce cart abandonment.

At this stage, the conversion goal is based on limited data and some guesswork. It can change once you collect and analyze new data. After that analysis, you connect each problem to a hypothesis:

Now you can create page variants and start testing these hypotheses. To check the rest of your site for problems worth testing, use the CRO checklist.

Why does CRO need patience, and what do you do when a test fails?

On sites with low traffic, you might spend weeks testing variants only to get a negative result. A successful CRO campaign therefore needs a lot of patience. Allow several months before you expect a clear improvement.

A failed test is not wasted. Every result, good or bad, can teach you something you did not know. For example, if you test several button colors and your current color wins, the time was not wasted. It means one of two things:

• Your current button already stood out on the page and did not need to change.
• The variants you tested were not different enough to beat it.

After a failed test, ask yourself these questions:

• Did you test based on intuition, or on insights from research?
• Did you copy a best practice from the internet without checking it fits your site?
• Does the metric you tested match your business goal?
• Was the variation bold enough for visitors to notice?
• Did you end the test too soon?

Review your test against these questions, then redesign your campaign.

Summary

• CRO does not give quick results, and it is not a magic bullet for every marketing problem.
• CRO is not a list of best practices, and results are not consistent across websites and industries.
• Every successful CRO campaign follows a process: discovery, testing, and review and analysis.
• Start by gathering insights into your audience.
• You also need quality traffic, good tools and a testing framework.
• Finally, you need patience: a CRO campaign can take weeks or months to produce results.

Frequently Asked Questions