A/B Testing in Marketing – What Should You Test First?

A/B Testing in Marketing – What Should You Test First?

A/B testing is one of the most useful practices in digital marketing when the goal is to make decisions based on real data rather than assumptions. Many businesses, however, approach it in the wrong order. They start with button colours, minor visual changes, or random ideas before identifying what actually has the strongest impact on enquiries and sales. In practice, this leads to wasted time, limited impact, and confusing conclusions. Meanwhile, Google continues to emphasise the importance of relevant, easy-to-navigate landing pages because they play a decisive role after the click.

According to NN/g: A/B Testing 101, an A/B test is a quantitative method in which two or more versions are shown to a real audience to determine which performs better against predefined business metrics. Optimizely describes the same principle similarly: traffic is randomly split between a control and a variation, after which results are evaluated statistically against the chosen objective. This is precisely what makes A/B testing valuable. It does not simply tell you which version “looks better,” but which one performs better for a specific goal.

However, not every test is a good test. Nielsen Norman Group specifically notes that A/B testing works best when clear business metrics and a specific objective are defined, while the variation should ideally differ from the original in one meaningful element. In other words, before asking “What should we test?”, you first need to know “What are we trying to improve?” This is why the topic is closely connected with What Is CRO and How Does It Help Increase Enquiries and Sales?, because A/B tests are one of the practical tools used in conversion optimization.

Start With One Clearly Measurable Objective

One of the most common mistakes is starting with ideas for changes before defining which action is most valuable to the business. Google Analytics defines a key event as an action that is particularly important to business success. This may be a submitted form, click on a phone number, consultation booking, quote request, or purchase. Once an action is marked as a key event, it can be tracked across channels, pages, and user journeys. A strong A/B test therefore begins not with design, but with accurate measurement.

Google Ads, in turn, defines a conversion action as a specific user activity that is valuable to the business and whose measurement can be used to improve campaigns. If you have not selected one primary conversion for the test, you risk comparing variations across too many metrics and ultimately not knowing which version actually performed better. This is why it is sensible to verify your tracking setup before launching an experiment. The topic is also connected with Google Tag Manager – What Is It and How Is It Used?, because without reliable event and data collection, you cannot trust the outcome of the test.

Do Not Start With Minor Cosmetic Changes

It is tempting to make the first test something small and easy, such as changing a button colour, icon size, or image placement. The problem is that these changes often have limited impact, especially when the page has more fundamental issues:

• Weak headline
• Unclear offer
• Difficult form
• Lack of trust
• Poor mobile experience

NN/g emphasises that experiments should be based on well-developed variations, while stronger variants are usually created when A/B testing is combined with UX research to identify the real causes of performance problems.

This is why starting with “cosmetics” is rarely the most sensible approach. A more practical method is to test elements directly connected to the decision to act. If visitors do not understand the offer, cannot identify a clear next step, or do not trust the brand, changing the shade of a button is unlikely to solve the problem. Your first tests should therefore focus on the strongest conversion levers rather than the easiest tasks to implement.

Start With the Value Proposition and Headline

The first thing worth testing is how the page communicates its promise to the visitor. The headline and subheading should clearly explain what you offer, who it is for, and what benefit it provides. If this is not obvious within the first few seconds, a significant part of the traffic may be lost before users even reach the CTA. Google recommends strong alignment between the advertising message and landing page intent, and in 2025 further emphasised the importance of relevant content and easy navigation after the click.

In practice, this means it often makes more sense to test different versions of the core promise than minor decorative elements. One variation might focus on a specific outcome, while another highlights the problem you solve. Even when the service itself remains identical, the way it is communicated can significantly change how the audience perceives it.

For a deeper look at how the post-click experience influences results, a logical next article is Landing Page Optimization – Which Elements Increase Conversions?.

Then Test the Offer and Call to Action

Once the core message is clear, the next natural priority is the offer and CTA. Google recommends using a specific call to action that tells users what will happen after they click. This is why buttons such as “Request a Quote,” “Book a Consultation,” or “Get an Audit” can be stronger than generic wording such as “Submit” or “Learn More.” The CTA should also match the stage of the customer journey. If you ask for too much too early, conversion rates may fall even if the page itself looks strong.

It also makes sense to test not only the button text, but the offer surrounding it. For example, you might compare “Free Consultation,” “Free Audit,” “Custom Quote,” or “Demo.” This distinction matters because users respond not only to the button, but to the complete promise behind it. The first priority is therefore not “What colour should the button be?” but “What are we offering, and how clearly are we communicating it?”

The Form Is Often the Next Strongest Lever

After the offer and CTA comes the form. This is where many landing pages lose a significant share of potential conversions. According to NN/g, careful form design has a major impact on how quickly and accurately users understand and complete forms. The same source cites research showing that forms following basic usability guidelines can reduce completion time and make successful first-attempt submissions almost twice as likely compared with forms that violate those principles.

Some of the strongest early tests therefore include:

• Shorter form vs. longer form
• Clearer field labels
• More logical field order
• Better visual organisation

NN/g recommends keeping forms concise, using a single-column layout, and placing labels close to the fields they describe. If your page is designed for enquiries, consultations, or quote requests, testing the form may produce a much stronger impact than experimenting with secondary visual details.

Trust Should Also Be Tested

People rarely submit personal information or enquiries simply because a button is visible. They do it when they feel enough trust. Nielsen Norman Group identifies four enduring factors that support trust in digital environments:

• Design quality
• Upfront transparency
• Comprehensive and up-to-date content
• Connection with the wider web environment

This means testimonials, case studies, real clients, clear service descriptions, visible contact details, and transparent explanations are not merely “extras.” They can influence a visitor’s willingness to take action.

Trust elements are therefore excellent candidates for early A/B tests. You can compare a version with reviews against one without them, specific performance evidence against generic claims, or a short case study section against a standard description. Particularly for higher-value services, trust can be the factor that ultimately influences the decision. When deciding what to test first, consider not only structure and copywriting, but also what gives visitors a reason to believe you.

Mobile Experience and Speed Should Not Be Left Until the End

Even a strong message and good form can fail if the mobile experience is poor. Google Search Central explains that page experience contributes to overall page quality, while web.dev provides specific thresholds for the main performance metrics. A good LCP is 2.5 seconds or less, a good INP is 200 milliseconds or less, and a good CLS is 0.1 or less. Google also states that strong Core Web Vitals are recommended for both user experience and Search performance.

If a significant share of your traffic comes from mobile devices, some of your early tests should therefore focus on the mobile first screen, CTA size, form position, loading speed, and visual stability. This is not purely “technical” optimization. It is a marketing test because these factors directly affect how many users reach the point of clicking or submitting a form. The page needs not only to look good, but also to respond quickly and predictably to real interactions.

Prioritise Tests by Impact, Not Convenience

The most sensible approach is to start with pages that have the highest intent and the most traffic. If one service page receives the largest share of Paid Ads traffic and has a clearly defined key event, that is usually where an experiment has the greatest potential.

Then prioritise elements closest to the conversion:

1. Headline
2. Value proposition
3. Offer
4. CTA
5. Form
6. Trust signals

Only after these should you move to secondary tests such as images, supporting copy, or decorative elements.

It is also useful to combine analytical data with behavioural observations. Microsoft Clarity provides heatmaps and session recordings that show where users scroll, where they struggle, and where friction appears. This does not replace A/B testing, but it helps you formulate stronger hypotheses instead of launching experiments randomly. The stronger process is therefore: observe behaviour first, develop a hypothesis second, and test third.

How Do You Prepare a Strong A/B Test?

A properly structured A/B test randomly distributes traffic between a control and variation and measures a predefined objective. Optimizely emphasises both random visitor allocation and the need for statistical analysis. Its sample size planning tools also highlight that tests should be designed around realistic expectations for the effect size and required amount of data instead of being stopped as soon as one version appears to move ahead.

In practice, this means several things.

First, avoid changing too many elements at once unless you intentionally want to compare two broader concepts.

Second, do not stop the test simply because one version appears to be “winning” after two days.

Third, monitor one primary KPI together with several secondary metrics. Otherwise, you may generate more clicks while reducing lead quality. If the experiment runs within Google Ads, the platform’s experiment tools can be used to compare performance over time between the control and test setup.

There is also an important practical update. Google Optimize was discontinued on September 30, 2023. Website A/B testing now generally relies on other experimentation platforms, while Google Ads Experiments remains available for campaign-level testing within the advertising platform. This distinction matters because many older resources still recommend Google Optimize even though it is no longer an active product.

Which Mistakes Should You Avoid?

One of the most dangerous mistakes is testing without a hypothesis. NN/g warns that poorly designed variations can produce weak results and misleading conclusions. A similar problem occurs when marketing teams test elements that do not address the actual reason for low conversion. If the real problem is the offer or lack of trust, testing a minor visual detail will provide little useful direction.

Another mistake is poor alignment between the ad, user expectation, and landing page. Google continues to emphasise that the landing page should match what the user searched for and what the ad promised. When this alignment is missing, even a technically sound test can be compromised by a weak starting experience.

A/B testing in marketing should therefore not be treated as an isolated process. It works best as part of a broader system involving messaging, measurement, UX, and continuous optimization.

Conclusion

A/B testing in marketing creates the most value when you start with the right elements. First, choose one clearly measurable objective. Then test the strongest parts of the conversion journey: the headline, value proposition, offer, CTA, form, trust signals, and mobile experience.

Only after those areas have been addressed does it make sense to move towards smaller adjustments.

When tests are based on a clear hypothesis, reliable data, and sufficient traffic, they become a genuine growth tool rather than a sequence of random changes.