TeaCup Lab Logo

Optimizing Experiences: How to Use A/B Testing to Improve Your Website's Conversion

Mónica Bohigas
March 6, 2024

Unqualified leads, high abandonment rates in the purchase funnel, low engagement... Does this sound familiar? These are common challenges we face daily, especially for Conversion Rate Optimization (CRO) professionals.

With A/B testing, we can make quick, data-driven decisions to address conversion and abandonment issues that significantly impact our business. In this article, we'll break down what A/B testing is, its advantages, how to implement it, and the key performance indicators (KPIs) to maximize its potential in UX Research.

What is A/B Testing?

In simple terms, A/B testing is an experimental methodology where two different versions of a website or app are presented to different users during a real visit. We then analyze which version performs better in terms of conversions.

By directing a percentage of users to version A and the rest to version B, we collect data from real visits on our website or app. The results of an A/B test allow CRO experts to make informed decisions backed by data, reducing risks.

The key to a successful A/B test lies in controlled variation and establishing clear and relevant KPIs. While one audience experiences version A (original), another is exposed to version B (variant). Subsequently, results are compared to identify which version achieved better KPIs, searching for statistically significant differences that support informed and guaranteed decisions.

Comparativa TestAB de una interfaz. Una con un botón rojo y la otra con un botón azul.

Advantages of A/B Testing

Summarizing the main benefits in 5 points:

  • Ease of Implementation: A/B testing is more agile and less invasive compared to complex methods like ethnographic studies or usability tests. Changes can be tested directly on the platform where your product runs without significant modifications.
  • Statistical Rigor: A/B testing typically provides statistically significant results, offering a numerical and objective understanding of the impact of variations based on solid quantitative evidence.
  • Continuous Optimization: A/B testing is ideal for continuous improvement of your conversion data (CRO) as you can iterate and constantly test new ideas and changes.
    Validation of Specific Hypotheses: It is effective in validating specific hypotheses about how certain elements or changes may be affecting key performance metrics.
  • Efficiency in Resources: In terms of time and resources, A/B testing can deliver results in a short period without complicated setups and participant costs. Compared to other user research methods, A/B testing is an agile and cost-effective technique.

Focusing on tangible results, A/B testing directly links to measurable and quantifiable business metrics like conversion rates or generated revenues, translating into real improvements.
While A/B testing excels in providing solid quantitative results, emphasizing continuous improvement and being resource-efficient, it is essential to consider the complementarity of different methodologies based on the goals and specific context of each UX research.

However, it's crucial to acknowledge some limitations of A/B testing:

  • Data informing the new design is necessary to decide which version to test.
  • The number of variables we can test with this technique is, in a way, limited. The more variables we introduce in design B, the harder it will be for us to attribute success or better results to a specific change.
  • A/B testing does not provide qualitative insights explaining why one version outperforms the other. If a version fails, there may be a lack of information on alternative approaches without prior research.
  • Determining the experiment's duration and declaring a winner can be challenging due to factors affecting results over time.
graph that shows test a and test b
In this image from a real test, we can see how the new version wins during the first two weeks. However, if we leave the test active for a couple of months, it becomes the worst option.


  • Some companies overuse this technique in the desire to achieve an improvement in conversion with every minor design or flow change. The reality is that small changes in design or copy usually help us improve results to a small extent, but only structural or strategic changes will bring significant improvements in conversion, in most cases. Therefore, if the goal is to impact the conversion rate significantly, an A/B test alone might not be the best tool.


How to Implement A/B Testing

illustration of the steps needed to implement an ab test


1. Define the Objective: Clearly identify what you want to improve or modify. Formulate a well-defined hypothesis with the problem, proposed solution, and metrics to measure results.

Example of hypothesis: Changing the button label to "Complete Purchase" instead of "Next" is expected to decrease the abandonment rate in the final step by helping users better understand the purchase process.

2. Create Varied Versions: Design a variant (B) that differs from the original (A), such as changes in color, layout, or text. Limit the A/B test to a single hypothesis to attribute improvements to a specific change.

Example: Version A is the button with the label "Next," and Version B is a button with the label "Complete Purchase."

3. Implementation: Use A/B testing tools like Google Optimize, Optimizely, or integrated tools in e-commerce platforms. Configure events and specific conversions to track during the A/B test.

4. Random User Assignment: Randomly assign users to version A or B to ensure equal distribution and eliminate biases.



5. Data Collection and Statistical Analysis: Wait and observe! Integrate A/B testing with Google Analytics or similar analytical tools for detailed user behavior data on both variants. If running ads, set up goal tracking on advertising platforms like Facebook Ads or Google Ads. Use statistical tools to determine if variant B significantly outperforms in achieving the objective and hypothesis.

Ensure all integrations and tracking are set up before launching the A/B test. Adhere to privacy and ethical policies when collecting and analyzing user data.

Key Performance Indicators (KPIs) for A/B Test Evaluation

In A/B testing implementation, defining one or several metrics aligned with business objectives is essential. Distinguish between:

  • Primary Metrics (KPIs): The fundamental indicator guiding the success or failure of the A/B test.
  • Secondary Metrics: Additional layers of information on underlying project goals, revealing user behavior changes.

Common KPIs for A/B tests include:

  • Conversion Rate (CR): Measures the proportion of users performing the desired action compared to total users.
    📈Calculation: CR = (Number of Conversions / Total Visitors) x 100
  • Click-Through Rate (CTR): Indicates the proportion of users clicking on a specific element compared to total users who see it.
    📈Calculation: CTR = (Clicks / Impressions) x 100
  • Dwell Time: Indicates the average duration users spend interacting with the page.
    📈Calculation: Dwell Time = Total Session Duration / Total Sessions
  • Bounce Rate: Represents the percentage of users leaving the page without any interaction.
    📈Calculation: Bounce Rate = (Bounce Sessions / Total Sessions) * 100
  • Abandonment Rate: Focuses on incomplete conversions or purchases.
    📈Calculation: Abandonment Rate = (Number of Incomplete Tasks / Total Tasks Started) * 100
  • Average Order Value (AOV) or Total Generated Revenue: Crucial in e-commerce, it reflects the average amount of a single purchase.
    📈Calculation: AOV = Total Revenue / Total Orders during A/B test
  • Scroll Depth: Measures how far a user scrolls on a webpage, revealing attractive sections and points of reduced attention.
    📈Calculation: Requires heat mapping or session recording tools like HotJar.
  • Customer Satisfaction Score (CSAT): Measures user satisfaction with a website, product, or service through a brief survey.
    📈Calculation: CSAT = (Positive Responses / Total Responses) * 100

When to Analyze A/B Test Results

The duration of each A/B test depends on factors like visit volume, desired confidence level, data variability, etc. However, guidelines for analyzing this methodology include:

  • Seek statistically significant results: Wait until sufficient data is collected to determine if performance differences between variant A and B are significant and not accidental.
  • Monitor sample size and results: Regularly monitor the sample size and test results to assess whether more time or sample is needed to achieve the desired confidence level.
  • Avoid hasty conclusions: Despite monitoring results and making calculations or charts, wait until the entire sample is collected for final conclusions.
  • Inform all web-involved teams: Notify all relevant teams about the ongoing A/B test to prevent overlaps or unforeseen changes. Be attentive to noticeable traffic changes during the test to understand their origin and take necessary measures.

In conclusion, A/B testing is a powerful tool for improving website performance, but it should be used in conjunction with other methodologies and considering its limitations. The careful definition of objectives, varied versions, and robust KPIs are crucial for successful implementation and meaningful results.

References and recommended readings and Podcasts about AB Testing:

A / B Testing: The Most Powerful Way to Turn Clicks Into Customers byDan Siroker

Trustworthy Online Controlled Experiments: A Practical Guide to A/B Testing by Ron Kohavi

The ultimate guide to A/B testing | Ronny Kohavi (Airbnb, Microsoft, Amazon)




Ready to elevate your user experience? Let's talk!
The answers you are looking for, one click away.
Get in touch ->
TeaCup Lab Logo
TeaCup Lab is a user experience consultancy agency founded in Madrid in 2016. We specialize in User Research & User testing for global brands.
Copyright 2019-2023 TeaCup Consulting SL

Contact us

TeaCup Lab 
Calle Jaén 2, 1ºG
28020, Madrid, Spain
+34 910 59 21 36