The art and science of conversion rate optimization is a data-driven approach to figuring out what customers actually want. It’s a research and data-driven process that seeks to best match customers’ expectations, and you can see what’s working by running A/B tests or analyzing data.
This article will cover 6 of my favorite conversion optimization research methods for discovering what your eCommerce customers really want, no more guesswork needed.
Also, you could run through these step-by-step, and that would be really helpful, but you don’t need to. You can pick and choose methods based on your business and needs.
eCommerce User Research: Learning to Ask Better Questions
What separates the best marketers from the average marketers isn’t necessarily more knowledge or experience. Rather, the best marketers ask the best questions.
It’s easy nowadays to find an answer rather quickly, and a correct answer is never far away. What matters, then, is that you’re finding the answers to the right questions. Starting with the right question in mind, you can then move to collect the proper data and decide on a strategy moving forward.
What are some good questions to ask in the context of ecommerce conversion research? Obviously, that’s going to depend on your particular situation, but some of my favorite fishing trips start with one of these questions:
- What types of words do our customers use to describe our products?
- Are there any bottlenecks that cause friction in our buying process?
- What fears or doubts do customers have, and how can we resolve them before they decide to buy or bounce?
- What are our customers’ biggest desires and motivations, and how does our product(s) fit into the picture?
- Is our website navigation intuitive?
- What do our best customers have in common, and how do they behave differently than other customer cohorts (or those who bounce from the site)?
- Which products are commonly purchased together and who buys them? What do they have in common?
Essentially, you want to a frame a question that will give you a testable hypothesis, so that you can run experiments to potentially improve your website. In eCommerce, there are generally three levers you can pull to improve your revenue:
The Three Levers of eCommerce Revenue Growth
- Increase conversion rates (or overall traffic)
- Increase the average order value
- Increase the purchase frequency
Conversion optimization largely deals with the first two of these, though the insights you gain in the process will help with everything.
So let’s dive into my favorite six data gathering techniques.
1. Ecommerce Analytics Analysis
I’m a huge fan of digital analytics. Nowadays, it’s so easy to set things up and collect valuable information, that there isn’t much of an excuse not to do so.
Google Analytics, the gold standard of web analytics, has continued to build out their suite of ecommerce-specific features.
At this point, you can configure things to know just about anything you’d want to know about visitors. Tying this into your Adwords account or your Google Optimize account also gives you the ability to target very specific user segments. But we’re getting ahead of ourselves…
At the most basic level, you should have set up goals, key behavioral events, and enhanced ecommerce.
Goals tend to be quite obvious in the context of ecommerce. It’s usually a) a product purchase or b) an email list sign up. You can configure both quite easily (here’s a guide on how to do so). If you have goals set up, you can get basic funnel visualizations…
Event tracking is a little more complicated because it involves more prudence and strategic thinking. You need to ask yourself, “which behaviors matter on my site?”
This tends to be things like promotional video interactions, clicks to product images, slider views, etc. Anything you want to track and make decisions off of can be included here. If you want to learn more about event tracking, here’s a super comprehensive guide. As an example, on my personal blog I track scroll depth as an event:
Google Analytics’ ecommerce features is where things get interesting. There are two types of ecommerce implementations in Google Analytics:
- Standard ecommerce reports, where you can get the basics like product and transaction information, average order value, ecommerce conversion rate, and other basic data.
- Enhanced ecommerce, where you get all the standard features plus more advanced information such as seeing when customers add items to their cars, seeing when they start the checkout process, and building more advanced funnel visualizations.
If you have a proper enhanced ecommerce setup, you can see really cool horizontal funnels:
Via Econsultancy. Image Source
With a proper enhanced ecommerce setup, you can also get super granular product performance data:
Google analytics can also track average quantity and number of unique purchases
Covering all of Google Analytics for ecommerce would require a book, so I’ll stop here. But basically, know that you can learn about just anything you want with the right Google Analytics implementation.
The only thing GA has trouble helping you with is qualitative insights. But for that, we can utilize customer surveys…
2. Customer Surveys
Most people use customer surveys incorrectly (or as Justin Rondeau puts it, usually, “surveys suck.”). They ask either a) leading questions or b) questions that have no real application or ability to make decisions off of.
In other words, when many marketers send out customer surveys, they’re wasting a whole lot of time and getting data that’s neither useful nor actionable.
However, when done judiciously and with optimization goals in mind, they can give you tons of insights, especially of the voice of customer variety.
Everyone always wants to know what survey questions to use, and it’s a hard question to answer, because everyone’s situation is unique. I won’t give you a cop out answer, though, and instead I’ll list my favorite questions for a CRO focused survey (h/t Svitlana Graves):
- What can you tell us about yourself? can also be framed as In 1 sentence, describe yourself.
- What are you using [the product] for? What problem does it solve for you?
- How is your life better thanks to [the product]?
- What made you buy [the product]? What convinced you that it was a good decision?
- What doubts or hesitations did you have before buying?
- What questions did you have that you couldn’t find answers to?
- Did you consider alternatives? How many websites did you visit before buying from us? which ones?
- What was your biggest challenge, frustration or problem during your visit to our website?
- Anything else you would like to tell us?
Here’s a survey sent out for a CRO client (name blurred out):
Part of user research is asking the right questions
I like to send surveys programmatically, usually immediately after a purchase as well as another survey if someone signs up for my email list. Depending on the client, I usually put the survey on the 3rd or 4th email, and I use it to draw insights about motivations and competitive insights.
Surveys can also be used programmatically to track customer satisfaction over time, though for the purposes of this article, I’ll gloss over that use case and link out to a guide here on the topic instead.
3. Email Marketing Analytics
Remember the three levers of ecommerce growth?
- More customers (more traffic or higher conversion rate)
- Higher average order value
- More purchases per customer
(Note that referral marketing can actually help you move the needle on all three of these levers).
As such, there are tons of insights to be gained from your email analytics.
- Which subject lines get higher than average CTR?
- Which email campaigns have been most effective?
- Are you properly timing your email marketing intervals?
- Are you optimize your behaviorally triggered emails? (such as abandoned cart and remarketing emails)
- Are there certain segments that respond better or worse to particular campaigns?
- Can you personalize emails for greater results?
The problem with instruction here is that the analytics you can get really depend on the tool you’re using, though most marketing tools give you the basics: open rate, CTR, conversion rates, etc.
Some tools, usually the marketing automation suites like HubSpot or Intercom, will give you much greater customer data. You can also get good website data if you simply tag your email links with UTM parameters (which many marketers forget to do).
What to improve: 50% open rate or 18% click rate
Again, though, just as with digital website analytics, you need to start with good business questions and then seek to answer them with the information you can feasibly gather.
4. On-Site Polls
Another go-to conversion research tool for me: on-site polls.
Customer surveys are a great way to gather insights from those you already know (and who who already know you). But if only ~3% of website visitors become customers, you’re missing out on a whole lot of data (and you’re engaging in a form of survivorship bias if you only listen to those who successfully converted).
On-site polls give you the ability to poll anonymous website visitors. This can be great for tons of purposes:
- Crowdsourcing blog topic ideas and questions to answer
- Finding user experience bottlenecks on a cart page
- Learning more about the psychographics of your general audience
- Finding voice of customer insights to inform your product copywriting
Via Qualaroo Image Source
Again, everyone always asks about good polls questions (and you’re probably tired of the “it depends what you’re trying to learn” response). While I hate to link to a gated guide, Qualaroo has a really great one that you should check out if you really want to dive deep into on-site polls.
Also, my favorite poll & feedback tools are:
5. Live Chat & Support Transcripts
Here’s the thing about CRO research: we always want to add on research methods, and we often forget about the piles of passive data that we’re already sitting on.
As an analogy, it’s like in B2B business – how much data and insights do your sales team and customer support team have, just by nature of talking to customers all day?
In ecommerce, if you have live chat, you have tons and tons of qualitative insights. How you analyze that is a question to be answer, but know that there is likely a ton of actionable information there.
You can go as simple as looking through a sample of transcripts and flagging UX issues, or you can go as complicated as conducting a statistical sentiment analysis to track happiness over time and in key cohorts.
I’ve written about this in the past, but when I was at LawnStarter, we often combed through live chat transcripts for product insights as well as for ways we could improve our website experience and messaging.
Customer queries can reveal their user intent
Even if you don’t have live chat, it’s likely you have some sort of ticketing or help desk software. This, too, is a source of amazing customer data. Who complains? What are their problems? Do issues repeatedly crop up? Can you solve them or create a programmatic support solution?
Finally! The gold standard of data-driven marketing and product decision making.
All of the insights you’ve gathered in the previous research steps could culminate here, in a prioritized roadmap, in the form of experiment ideas. That way, you can validate each decision as you move forward. I like to prioritize ideas based on impact (the opportunity size) and ease (how complex the test is to set up and analyze) and put them in a big spreadsheet:
Experiments, or A/B tests, give us a much more scientific way to make decisions. They mitigate risk and downside while simultaneously encouraging us to try wilder and more creative ideas.
What you learn in your other research methods can be used to inform A/B test hypotheses. For example, let’s imagine you have digital analytics funnel data that suggests too many visitors are dropping off during the checkout. To diagnose the issue, you set up an on-site poll using Qualaroo, and you find out that pretty much all of the answers point to people being confused about “coupon codes.”
Now, the right solution could be one of many things. But your hypothesis could be stated as:
“We believe visitors are not completing their checkout because there is confusion and a lack of clarity regarding our promotional coupon policies.”
This, then, could lead to several A/B tests to see which potential solution could provide a lift in conversion rates. One such test could be removing a visible and prominent field for coupon codes, and instead showing a text link that says “Have a coupon? Click here.”
There are many options, btu the point is that the test hypothesis arose from data, and you’re using experiments to find a data-informed solution to your conversion problem.
Perhaps the prominent coupon code is triggering FOMO and lowering conversion rates (image source)
Now, A/B testing is not for everyone. If you don’t have a ton of traffic, it’s very unlikely you’ll be able to run tests, let alone get any solid ROI from a testing program. It’s a simple matter of mathematics; you need a proper sample size to conduct valid experiments.
There’s no magic answer as to the number of traffic or conversions you need, though you can play around with a sample size calculator to see if it’s right for you.
6 Steps to Customer Research for eCommerce Conversion Rate Optimisation
There are tons of ways to learn about your customers, some of which are highly rigorous and quantitative (a/b testing) and some of which are more open ended and qualitative (customer surveys).
The customer data method you use should depend on what answers you’re trying to get. If you want to know what headline works better, an A/B test is one of the best ways to find that out. If you want to know if your headline is confusing in the first place, a poll or survey might actually help you find that out.
Generally speaking, the more you know about your customer, the better – though there is a point of diminishing returns with conversion research. At a certain point, you can’t learn a ton more from conducting additional surveys or pouring over data, you simply have to act.
But it’s all a balance, and unfortunately, the most companies are operating under too little information, not too much. That makes asking the right questions even more important
This article outlines a minimum viable version of ecommerce user research. You could go much deeper, but these techniques will get you on your feet and making better, more data-backed decisions. That, in turn, will lead to high conversion rates and a better ROI on your efforts.
Alex Birkett works on user acquisition growth at HubSpot. He’s based in Austin, Texas, but travels roughly six months of the year.
When he’s not working on growth and optimization, he enjoys martial arts, yoga, snowboarding, and occasionally writing on his personal blog.