In this chapter, you are going to learn about Conversion research or diagnosis. When you begin to optimize your website for higher conversion, it can be challenging where to start and how to measure the improvements. Every optimization or website redesign project should start with conversion research.
Think of it in this way: if you go to a doctor and say, “I have a stomach ache”, and the doctor says, “Sure, let’s get you operated tomorrow.” You’d call the doctor crazy, right? You’d expect the doctor to do some analysis, take some tests, and identify the problems first.
The same way, when you work on a website, it’s important to identify all the problems, and then only work on them. But I can’t tell you how many companies are guilty of designing their websites without really knowing what worked in the past, and whether their new design is going to convert better or not.
Peep Laja - the founder of CXL Institute on when Conversion Research should be conducted.
If you’ve got at least one acquisition channel working and bringing you traffic, conversion optimization is the easiest and cheapest way to grow your business, without spending too many resources.
The goal of growth marketing is not to acquire more customers, but to grow your businesses the fastest. While building your product, you consciously let several conversion gaps in order to move fast. Now is the time to fix that!
Note: If you are a new business, or a new site, you may not need to get into the conversion research right now. Just go and talk to customers, survey them (read the last chapter again) and acquire more customers first. Because there’s not much to optimize yet.
Before you start analyzing data, make sure you’re gathering all the data that can be acted upon. You don’t have to measure everything and fall into the analysis paralysis situation. However, you also don’t wanna be restricted due to the lack of data.
🛡️ You are reading “Conversion Research: A Game of Growth (Chapter 3)” – a series of articles on growth marketing. To read from the beginning, start here 👉 Fundamentals of Growth Marketing: A Game of Growth (Chapter 1)
How to conduct conversion research?
I highly recommend you to use the ResearchXL model by CXL, made popular by the founder, Peep Laja. It consists of six types of analysis:
- Heuristics analysis
- Technical analysis
- Digital analysis
- Qualitative Surveys
- User testing
- Mouse tracking analysis
This is an experience-based assessment of your internet properties i.e. website, mobile site or mobile app as a user.
Are there some concerns over the user experience? What are the key customer journeys?
Put yourself in the customers’ shoes and go through what they’d normally experience.
- Make note of URL structure and build a picture of site navigation flows
- Identify key points of interests for analytics and note down any suspicious URL
- Test all the key devices and browsers
- Fill out all the lead capture forms, go through the payment gateway and signup forms to see if they’re working properly. You’d be surprised.
Go through the journey from your inbound or outbound channels, to landing page, to checkout page (in case of ecommerce) and evaluate each page based on these criteria:
- Is it clear what this page is about?
- What’s being offered?
- Is there a clear, single call-to-action?
- Do you understand the purpose of this page?
- Is the webpage what visitors thought it is?
- Are pre-click and post-click messages and images the same?
So many companies make the mistakes of running a Facebook or PPC ad with a specific messaging and guide this traffic to their homepage where they don’t see the specific message.
3. Incentive to take action
- Is there a clear incentive or motivation for users to take action?
- Do they have enough information to take the desired action?
- Is the copy persuasive?
- Can you increase the motivation with more persuasion techniques like urgency, trust, scarcity or social proof?
- Is it easy for them to take the desired action?
- Are there any long forms to fill, or sensitive information being asked?
- Are pages slow to load? That’s a major friction and users will just drop off.
- Is the design or information easy to find? Too many hidden tabs or critical information below-the-fold might be points of friction you should address.
- Any cheesy, stock images, spammy looking pages, typos and poor spellings, or poor readability errors?
5. Distracting elements
- Are there any distracting elements that are taking the focus away from the desired action?
- Blinking banners or sliders, are there too many CTAs?
- During checkout, are there other navigation elements that could be removed?
6. Customer Journey issues
- Are you asking for their commitment too early in the customer journey or their buying phase?
- Do they have to give their asking for email or phone number without getting anything in return?
- Are you asking for a credit card before they could sign up for your product?
A technical analysis is important to find if there are any specific browser or device or technical aspects that are hindering conversion.
With technical analysis, you wanna mainly answer these key questions:
- Does the site work on all main browsers? Are there any browsers that are converting less as compared to others?
- Does the site work on all main devices? Are there any devices that are converting less as compared to others?
- Does the site load slower in certain key geographies?
Let’s take a look at this sample data set of conversion rates and revenue from different internet browser versions (in this case, Internet Explorer)
We can see that IE8 converts much lesser than the others – IE7, IE9, IE10. What could be the possible reason?
IE8 accounts for $20,904 of revenues per month at a conversion rate of 4.6%. If you can improve the conversion rate from 4.6% to 6.5%, that’s an extra $8500 per month (~$100K per year) with just a small technical fix. Is that worth fixing?
Site speed matters! If your site takes 15 seconds to load, your users are just going to lose interest and simply quit. And just working on speeding up your site can improve your conversion rate and purchases significantly.
The third type of analysis we do is Digital Analysis. With digital analysis, we are trying to understand 3 things:
- Drop-off points
- Correlate behaviors with the outcomes
- Is the data verified and correct?
Drop-off points analysis can be things like:
- Where do people leave the site?
- What are the types of users that are dropping the most and why?
- Where is the user flow stuck?
- Where is the money leaking out from?
With behaviour-outcome correlation, you are trying to find out what are the behaviours that users who buy the most exhibit? For example, the users who buy more stuff on your site, also search more.
A behaviour can be the path they follow to become a customer. For example, people who visit the product page, then watch the customer success video, then go to the pricing page, then sign up for the “Standard” plan have a higher CLV than those who signed up directly from the product page.
The only way we can measure such behaviors is if we are measuring everything on your site i.e. every page, every CTA, every form, every video, etc.
Next, we wanna be sure that the data you’re collecting is correct and verified. You can do that by checking the previous analytics configuration. Are some forms measured twice or are there any anomalies in the data, etc.
Next, you should do qualitative surveys to understand your existing visitors’ behaviors. There are two types of surveys you want to do:
i) On-site polls
These are the polls that you put on your key pages where you want to increase conversion. With these polls you are trying to get answers to only one question.
For example, you see that people are visiting your product page, but aren’t buying anything or adding the product to the cart. You can show them a poll and ask, “Is there any information that you want to know about this product?” If they click yes, then show open an input box for them to write in more details.
Such polls can be on any key pages like the Checkout page or Payment page to know what’s holding them back from moving to the next step.
Tip: Only show one poll for each user, otherwise you’re adding unnecessary friction to that user’s experience.
ii) Post Purchase polls
After someone has made a purchase from you, here is an opportunity for you to know more about your real buyers, and their shopping experience.
Do not ask immediately after they made the purchase since they might want to just finish the task of buying from your website. But you can get this survey after several hours or days over email while they still freshly remember it.
You ask 8-10 questions about what made them buy your product, what product features or benefits were important to them, what doubts they had, did they compare your product to any other competitor product, etc.
You can also ask them if they would like to get a better service (free or paid) – e.g. faster delivery time, personal counselling session, an upgrade, etc.
User testing is observing your users interact with your website or product. Ideally, you’d want to invite your actual customers or audience to participate in this study. Sometimes, it is not possible (due to lack of resources, time or money) but try to find people similar to your target audience.
In this testing, you give them two types of tasks: a broad task, and a specific task.
A broad task can be – find a product that you want to buy from this website. This way you will understand how most people interact with your site when they don’t know what to buy.
A specific task can be – find a light colored couch that can be converted into a bed for your home which is below $1000 and has a delivery period of less than 3 weeks. By observing your test users find the product when they work with constraints, you learn how your actual users might be looking for specific information.
You can ask your test users to describe what they are doing, but more than relying on their words, focus on their actions. Oftentimes, people describe their experience completely differently than what their actions tell. They may have experienced a slow site, confusing UX and they spend several minutes trying to figure out how to take an action, but they might describe their experience as perfectly smooth because they’re trying to be nicer.
Mouse Tracking Analysis
With tools like Hotjar, Capturly, etc. you can view how your visitors browse through your website. There are three types of behavior you can see:
- Clicking or tapping behavior
- Scroll behavior
- Mouse hover behavior
Click maps or heat maps are good to observe broadly to know if there are any particular CTAs that people aren’t clicking on. You can also get this information via Google analytics (refer digital analysis section) but such maps can be a good way to visualize or visually present your analysis.
Scroll behavior is understanding how far users are scrolling your pages so that you don’t have useful information miss out simply because it was located several scrolls down.
You can safely ignore mouse hover behavior because it is not an actual indication of where people are looking. It’s just where they subconsciously move their mouse. For example, while reading an article or a product description, nobody hovers their mouse from left to right.
What to do after you’re done with your conversion diagnosis?
After you have conducted your diagnosis, and have found (hopefully) tons of problems, issues, and minor obstacles to fix, you have to prioritise these issues. The way you can prioritise is on a scale of 1 to 5:
- 1 being minor usability issues which might not have huge impact but still are important, and
- 5 being a major usability, conversion or persuasion issue that is blocking most of the users from taking actions on your site and can have a high impact.
After prioritising the issues, you need to categorise them into various categories:
Issues related to incorrect data gathering, problems with analytics tools or complete absence of analytics or data gathering.
These are issues for which you understand the problem and there are several obvious solutions, but we need to test different solutions to know which one works the best.
Issues for which the problem is known, but you don’t know the solution. You need to figure out the solution that might work. You may need to brainstorm with rest of the team, and
These are the issues for which the problem is not clear. You need to investigate further to understand the issue first.
Just do it
At last, these are the issues for which you don’t need to spend much time. These are very small and simple issues that you should just do it. It doesn’t take much time.
This is how you conduct a Conversion Research before you start optimizing your website. Now is the time to open your priority issues spreadsheet, and just start fixing these issues one by one.
Note that a conversion research should also be done after you fix all the issues to come up with a fresh list of issues and ideas to optimize your website.
Because optimization should never stop!
This article summarizes the Conversion Research course by CXL Institute which goes into more details on how to design surveys, how to conduct surveys, what other questions and issues you can find from talking to your customer support team, exit surveys, live chat, etc.
There are other interesting concepts discussed in this course such as:
- the health check of Google Analytics
- Funnels and Goal Flows
- Content report
- Conversion research reports
If you wanna learn more about these concepts, please comment below or drop me a message on Linkedin.
🛡️ You just finished reading Conversion Research: A Game of Growth (Chapter 3) – a series of articles on growth marketing. To read the next article, click here 👉 A/B Testing: A Game of Growth (Chapter 4)