In this chapter, we walk you the typical CRO process, detailing the steps from the research stage all the way through to the testing stage.
There are several ways that you can go about implementing CRO into your SEO strategy for your website.
In this section, we’ll explore the typical steps of performing CRO as well as highlight two of the core methodologies of CRO: Analytics Based and User Based CRO.
Typical CRO Process
1. Objectives and Questions – lay the groundwork by identifying the objectives and questions of your target audience. This involves identifying the driving factors for your conversions as well as any roadblocks that are preventing users from converting.
2. Gathering Data – establish a baseline by gathering information about your website’s current conversion performance – this is both user based and analytics based.
3. Analysis of Data – What is your current conversion rate? Which are your best sources of traffic for your various calls to actions?
4. Hypothesis – Look at the baseline that you have established and identify the most pressing issues that are preventing users from converting.
5. Design – design a testing strategy to help measure the changes you intend to make on your website.
6. Testing – run the tests to measure the success of your changes against the baseline.
7. Post-test Analysis – determine whether the changes were successful or not, did the conversion rates improve?
8. Learning from Tests – apply techniques that helped increase conversions across your website.
The Importance of Research in CRO
Research is one of the most important cornerstones of CRO. Without it, everything else you do, simply has no purpose and will be ineffective. The process of gathering and analyzing data is strenuous and doing so correctly requires a lot of time and effort at the early stages of the CRO process.
Setting Relevant and Specific Goals
Before conducting your research it’s important to first identify and understand the goals that you are trying to accomplish. The goal isn’t simply just to “convert” – instead, it’s about converting to buy, converting to get more subscribers, converting to get more visitors to reach a desired landing page on your website. Identifying the outcome of your conversion goal is vital.
To understand your goals, you need to understand how users are engaging and interacting with your website – this begs the question, what is the purpose of my website? In other words, you want to formalise the existence of your website by identifying what you are trying to achieve for each page. Doing this, will help you form your conversion goals.
In doing so, you need to measure how existing users are engaging and interacting with the website i.e. analysing which pages are and aren’t converting well. This requires means to gather data.
As you’ll see below, there are two main methods of gathering data – qualitative and quantitative. The former, focuses on gathering numerical data whereas the latter involves asking real users what they think about the site. It’s important to combine both of these methods so that the amount of available data is as large and complete as possible.
This research helps you to identify friction and anxiety points within your web pages.
Friction is anything that is preventing the user from achieving their goals. For example, asking the user lots of questions upon signing up to a newsletter slows down the process and may frustrate the user.
Anxiety is anything that may result in the user second guessing whether they should place a purchase on your website or not; this is also known as purchase anxiety. If your website doesn’t look trustworthy or professional, users will be less inclined to place an order with you.
Analysis & Reporting
The final step in the research process is analysing and reporting the results of the data that you’ve collected. This is where you truly get to grasp the scope of what needs to be done in order to reach the goals that you set at the beginning. This will help you identify why certain pages may not be converting, as well as offer possible solutions as to how you can combat them.
To summarise, the research process for CRO is pivotal in ensuring the success of your CRO campaign.
Quantitative Data Analysis – The Analytical Approach
In order to establish a baseline for where your site’s current conversion rate is at, you need to gather data. One of the methods for doing this is known as quantitative data analysis i.e. The Analytics Approach. This methodology gives you hard numbers behind how people actually behave on your website.
For this, we needn’t look any further than Google Analytics which provides helpful information about conversion and engagement metrics.
Use Google Analytics to find out:
- Which web page users land on first.
- Which features or elements of your website users engage with the most, i.e. where on a page or within your site do they spend most of their time.
- What channel and referrer brought them to your website, i.e., did they come from another website?
- What devices and browsers are they using?
- Who your customers are: age, demographic, interest, gender.
- Where or during what activity do users leave the conversion funnel.
- Which traffic sources are more likely to convert.
- Which page visits lead to the most/least conversions.
- Which paths people take on your website i.e. your sales funnel.
This information will let you know which pages you should focus your efforts on improving.
There are lots of brilliant articles on how to set up your Google Analytics for Conversions such as this one from Neil Patel or the one from Google, so instead, we’ll go over the most important engagement and conversion metrics that you should focus on.
You can find most of these metrics by heading to the Audience > Overview report in Google Analytics.
1. Traffic Source – when it comes to conversion, it’s not about how many visitors you get, but how many of them convert. In addition, it’s also good to identify where visitors who convert the most are coming from.
- Direct – visitors that land on your website by directly typing your URL in their browser address bar.
- Organic Search – visitors who find your website through a search engine.
- Referral – visitors who click through a link from another site to yours (this can also be from a social media page or elsewhere).
2. Bounce Rate – The bounce rate tells you how long users stay on your website before leaving. Google Analytics provides both an overall and page specific bounce rate, this means that you can identify the pages that users are leaving the most frequently and focus on improving them. For example, if your product pages have a high bounce rate, it’s a sign that users aren’t seeing value in them and could be a reason why conversions are low.
3. New vs. Returning Visitors – the way a new user interacts with your website is different to that of a returning visitor. For first time visitors, focus on what they’re interested in when they visit the website for the first time and how you can improve that experience. Likewise, if the bounce rate is high for new users, identify why that is the case i.e. are you not providing enough information? For your returning visitors, focus on why the user may have returned, because even if they didn’t convert the first time, you made enough of an impact to encourage them to return.
4. Devices – the device a visitor uses also determines the experience they’ll have. For example, if your website isn’t optimised for mobile users, it’s very likely that your conversions for this demographic will be low. GA provides a useful breakdown of visitor and conversion statistics based on the device that they used i.e. Desktop, Mobile, Tablet.
5. Session Duration – GA tells you the average time a user has spent on your website. This can tell you useful information about their experience, for example a high session duration may indicate issues with navigation – i.e. users are spending more time trying to find what they’re looking for.
6. Pages Per Visit – In addition to session duration, GA provides an average for how many pages a user visits during their session. This metric also links to the bounce rate, i.e. if the bounce rate is high, the pages per session will likely be low. You want to ensure that users are able to find what they’re looking for quickly so that they can spend more time on the page that they’re actually interested in.
7. Exit Rate – apart from looking at the pages that users first land on, a more telling metric for conversions, is the pages that users end up leaving your site on. Pages with a higher exit rate indicate a poor user experience. If you have an online store and the cart page has a high exit rate, this means that you should look to improve the experience on this page so that more users move onto the next stage in the funnel and potentially make a purchase.
8. Geographical Location – GA tells you the number of visitors who visited your website as well as what geographical location they are in. This allows you to identify locations that have higher or lower conversion rates based on session duration, bounce rate etc and improve the pages according to that specific market.
9. Demographics – The Demographics report provides basic information about the age and gender of users. This information can be vital in determining which pages you should target for which gender or age groups. For example, if you have an online fashion store, you may find younger users tend to visit the Sales or Discount page – this may inspire you to use specific terminology or phrases such as “student discount” etc.
Identifying Drop Off Points
Google Analytics also allows you to get crucial information about drop off points i.e. identifying the pages where users are leaving your website.
Here are a few reports from GA that allows you to do this:
1. Behaviour > Site Content > All Pages Report > Sort By “Bounce Rate” > Sort Type (next to Secondary Dimension) > Select “Weighted”
This report sorts the pages on your website with the lowest bounce rate. This allows you to identify the most important pages for you to focus on.
2. Behaviour > Site Content > Exit Pages Report > Sort by Number of Exits (not Exit %, as this would display pages with 100% Exit Rate which also have extremely low page views)
This report displays the pages that have seen the most visitors leaving.
3. Audience > Users Flow
The Users Flow report allows you to visualise how users behave on the site, how they move through the pages, and most importantly where they drop off.
4. Conversions > Ecommerce > Shopping Behaviour Analysis
This report shows the visitors’ flow through the various stages of the site’s shopping experience. It displays the number of people that entered, abandoned and moved to the next stage of the sales funnel.
It’s worth noting that this report does not show what the affected pages are. In order to find this, you can create some segments based on stages like:
- No Shopping Activity – displays all of the pages where your users did not perform any action.
- No Cart Addition – shows a list of pages that users visited but decided not to add any products to the cart. If the homepage appears as one of the page’s listed, then this is likely not a serious issue as users generally will not add items to their cart from the homepage. If product pages are listed however, you will need to identify why a high percentage of users do not add any items to the cart.
- Cart Abandonment – displays all of the pages where your users left your site despite adding products to their cart.
- Checkout Abandonment – displays all of the pages where your users abandoned their purchase.
5. Checkout Behavior Analysis
The Checkout Behavior Analysis report shows the user’s movement through each step of the checkout process. For example, it illustrates how many users moved on from one step to the next, how many dropped off at each step, and how many entered the process at each step.
This is very useful if you want to identify pages where users are dropping off during the checkout process.
6. Funnel Visualisation Report
The Funnel Visualisation Report refers to the eCommerce conversion goals you’ve previously set up. It shows how users move through the steps that you’ve identified, as well as the percentage of people who move on to the next steps and people who drop off are.
Qualitative Data Analysis – The User Based Approach
Once you’ve carried out your quantitative analysis, you will have a great understanding of how users are interacting with your website. This will allow you to move onto looking into the why behind their behaviour. This user based method, which is also referred to as qualitative data analysis, helps identify your ideal customer by surveying your existing customers and users.
By actively surveying previous customers who have successfully made a purchase on your website or signed up for your service, you’ll get an insight into what motivated them to go ahead with choosing you over your competitors.
Here are some top tips on how to go about creating your customer surveys:
- Only ask the most relevant and important questions – customers’ patience will be very low, so shorter surveys tend to be more effective and are likely going to result in more responses. Aim to limit your survey to between 5 to 10 questions and if absolutely necessary, split a longer questionnaire into multiple smaller surveys for users to fill out.
- Offer some sort of incentive – providing a prize or incentive naturally increases the chances of a user filling out your survey. Incentives could be anything from gift vouchers to your store, discounted orders or free prizes.
- Ask open ended questions – although multiple choice questions are easier for users to answer, when it comes to CRO, you want to hear as many subjective answers as possible. Every user will have their own preferences and will ultimately have their own unique experience whilst using your website, so you want them to answer the questions in as much detail as possible. This is something that you won’t be able to achieve with the limitations of multiple choice questions.
There are countless surveying services available, such as SurveyMonkey, Google Forms etc.
But we won’t dwell on those for too long, instead, let’s focus on what kinds of questions you should be asking your users.
Here are some of the vital questions that you should look to include within your survey to gain a better understanding of the reasons behind why your users chose your product/service over your competitors. Please bear in mind that these will likely need to be adapted slightly to fit the product or service that you’re offering.
1. How would you describe the product/service to a colleague or friend?
2. What alternative options did you consider before choosing the product/service?
3. What made you decide to go with our product/service?
4. If anything ,what may have prevented you from signing up or making a purchase?
5. What questions were unanswered about the product/service?
6. Why did you ultimately choose to place an order or sign up?
7. How could we improve the website to persuade your friends or colleagues to choose us?
8. How would you persuade more people to choose us?
9. What are you hoping to achieve with the product/service?
10. When did you realise you needed a product or service like ours?
11. What caused you to come looking for our product/service?
12. What problem would you say our product/service alleviates?
13. What two adjectives/words would you use to describe our product/service?
Whilst customer surveys are focused on why users decided to choose your product or service, on-site surveys are designed to get a better understanding of their experience and behaviour whilst using your website.
You have the opportunity to ask questions to visitors who not only may become future customers, but also those who visited your website and for whatever reason, did not convert. This second group’s answers are the most important as these are the people who you want to impress the next time they visit.
Here are some questions you can ask to learn more about your site’s visitors:
1. Is there anything you were unable to find on this page?
2. Are there any confusing elements about this page?
3. What would you say is your biggest concern about purchasing the product/service?
4. What is the main reason that stopped you from making a purchase or signing up?
5. What else would you like to see on this page?
6. What could we have done to convince you to complete the purchase or sign up?
7. What’s the main issue we can help you solve?
8. What are the main things that you are looking for in your ideal solution to your problem?
9. What else can we place on this page to convince you to buy?
10. How would you describe your overall experience with this page?
11. How would you describe your overall experience with this website?
12. What was the first thing that you noticed about the page?
The goal for on-site surveys is to find out what prevented visitors to your website from becoming customers. Their answers will give you an idea of what exactly they’re looking for i.e. would they prefer to try your service for free a short period of time before committing, are your prices too high, was the content on your website unclear or not very descriptive?
Without testing the changes that you’ve made to your website based on the research you’ve carried out, CRO is a pointless exercise and could be detrimental in certain cases.
In this section, we’ll discuss one of the core testing methods for CRO: A/B testing.
What is A/B Testing?
A/B testing (which is sometimes also referred to as Split Testing or Bucket Testing) compares two different versions of a web page (also known as a 50/50 split or A/B split) to see which one performs better.
For example, a simple A/B test would involve sending a 50/50 traffic split between the original page and a variation to see which performs better for a given conversion goal.
This allows you to directly compare the two different variations of the page and make an informed decision on which is best based on the data collected. By empirically measuring how the changes impact the user’s experience, you can ensure that every change produces both meaningful and positive results.
Why is A/B Testing Important?
Make Careful Decisions Based on Empirical Data – A/B testing allows you to make careful business decisions about your website by collecting empirical data that backs up your hypotheses and ideas, so you aren’t randomly guessing and making changes on the fly.
You Learn More About Why Rather Than How – With a focus on hypothetical testing, A/B testing gives you a better insight on why certain changes improved conversions and enhanced the user’s behaviour and experience of your website, rather than just highlighting what those changes were. On the flip side, A/B testing also shows you why certain elements of the experience perhaps don’t work as well.
It’s A Cyclical Process – A/B testing isn’t just something that you carry out once. It’s a procedure that you can carry out again and again. If you find that a change improved conversions for one landing page, you can then perform the same (or similar) test on different pages to see if the results can be replicated. Testing should be performed consistently to simultaneously continue to improve the user’s experience and your conversion goals over time.
Granular Testing – testing a single change at a time helps you to identify which changes had a positive (or negative) impact on the users’ behaviour. Combining these changes over time will ultimately help produce the best possible results.
Google Encourages It – If any of the above reasons didn’t convince you, then this probably will! Matt Cutts, the former Head of Webspam at Google, shared this blogspot from Google which included best practices for performing A/B tests.
A/B Testing Process
A/B Testing is usually carried out right at the end of the CRO process; it’s the very last thing you do before the implementation of a specific change, after you have:
1. Collected both qualitative and quantitative data and performed very thorough research.
2. Identified the goals that you want to achieve for your conversions
3. Created some hypotheses (which need to be very specific: including what changes you want to implement, based on what data, and what outcomes you aim to achieve). Here’s a great Hypothesis Kit from CRO expert Craig Sullivan to simplify the process.
4. Prioritized the hypotheses in terms of expected impact and difficulty of implementation.
5. Create variations based on the changes that you want to make using your A/B testing software (we recommend HotJar) i.e. changing the size of a submission form button.
6. Run the experiment and collect data on how users are interacting with the different versions of the web page.
7. Analyze the results to determine which variation of your page performed the best. If the change you made performed better than the original page, then you can apply what you learnt from this experiment to other pages on your website.
8. Analyze the results to determine which variation of your page performed the best. If the change you made performed better than the original page, then you can apply what you learnt from this experiment to other pages on your website.
The Implications of A/B Testing on SEO
Google encourages the use of A/B testing and multivariate testing and has stated that performing such tests will not affect your website’s search rankings, if you do it properly.
Google have outlined some guidelines that you should follow to run an effective test without impacting your site’s search performance
- Cloaking – Cloaking is where you intentionally display a different version of a web page or content to search engines than you would to normal visitors to your site. Google takes cloaking incredibly seriously and may go as far as demoting or even removing your site from the search results. To prevent this, Google recommends that you do not display different content to Googlebot based on user-agent or IP address.
- Use rel=”canonical” – Google recommends using the rel=”canonical” (rather than the noindex) attribute on all alternate versions of the page that you are testing, to point to the original version. This will help prevent Googlebot from getting confused by multiple versions of the same page.
- Use 302 Redirects – When running a test that redirects the original URL to the variant, Google recommends that you use a 302 redirect rather than a 301 redirect which tells Googlebot that the redirect is temporary as opposed to permanent and that the original URL should be indexed instead of the test URL.
- Don’t Run Your Experiments For Too Long – Google recommends that you run your experiments for as long as necessary. Running your tests for longer than needed may be perceived as an attempt to deceive Googlebot. You should update your site by removing all of the test variants as soon as the experiment has finished.
Small differences like the colour of text, size of fonts, placement of a button etc will have very little impact, but following the above guidelines should prevent any issues to your site’s positions in the SERPs – after all, it’s better to be safe than sorry!
Common Pitfalls to Avoid With A/B Testing
Going On A Hunch – A/B testing starts with a hypothesis, so if this element of your experiment is incorrect, it will drastically impact the results. Likewise, the hypothesis should not be defined by a hunch or personal preference but instead should be defined by the outcome that you want to achieve.
Testing Too Many Elements At Once – A/B testing is most effective when you’re testing a single change at a time. Testing too many variations together will make it difficult to pinpoint exactly which change(s) impacted the results – regardless of whether the results were positive or negative.
Unbalanced Traffic – Each variation of your experiment should receive the same (or as close to the same) traffic – otherwise the data collected will naturally not be representative or accurate. By extension, you will be unable to determine which variant performed and converted the best.
Incorrect Duration – finding the right balance fow how long to run a test is key. Running an experiment for too long or too short a period can also result in inaccurate data. This is because one version of your URL may instantly show positive results whereas the other simply may take a little longer, so cutting the experiment short would not account for this.
Likewise, experiments that run for too long not only defy Google’s guidelines (which we highlighted above), but are simply pointless – there’s no magic number on when you should stop, it’s simply a matter of knowing when you’ve collected enough data to back up your hypothesis.