Conversion Rate Optimisation (CRO) is the spectre at the feast for marketers. In the online world, it’s more important than pretty graphics, elegant writing or even beautifully crafted products. Getting customers to do what you want is an absolutely essential but overlooked skill. We’ve put together the following guide to help you.
Good CRO means understanding human psychology and using rigorous testing and analysis to confirm/disprove different strategies. There are no hard and fast rules for improving conversions, but there is a mindset you should adopt:
Conversions don’t mean success. It’s vital to understand this simple truth. As a metric, the pure conversion rate of customers into sales, leads or signups can only tell you so much. Were these sales profitable? Did these leads later purchase a product? Did these signups do anything but add to your expansive Christmas list?
If you were so inclined, you could boost your conversion rate in minutes by slashing prices or giving away luxury goods with every email signup. This practice would probably fare most businesses quite well until inevitable bankruptcy two weeks in.
Effective CRO is about making the signup/purchase process irresistibly seamless without compromising your brand or service. Psychology, design and copy all play a role. Most importantly, CRO is data-driven. Testing and analysing teaches marketers the most about their audiences. Because every case is unique, every site has to discover its own strategy using ideas listed below.
What does conversion really mean for businesses?
“Making the path to purchase as simple as possible” – Juan Diego Calle
What do you want visitors to do on your site? This is your conversion event. From making a purchase to joining the mailing list, your conversion rate is how frequently visitors trigger this event.
Say you were running a fashion website. If 2,000 visitors made 20 purchases, your conversion rate would be 1%. If 2,000 visitors made 10 purchases, your conversion rate would be 0.5%. Simple.
But what does this really mean? While conversion rates are quite a bare-bones approach to analysis, they can tell businesses a lot—especially if they have the ability to analyse and contextualise results.
Conversion rates are useful because they measure what happens when people are already on your website. Sure, a good paid search campaign might bring a lot more viable customers to you, but ultimately, it’s what your website does with them that creates results.
Consistently improving conversions should be your primary goal. How to do this with testing and tracking is something we’ll explore later on.
Increasing your conversion rate from 1% to 1.2% means 20% more revenue. What’s so great about this? Good CRO saves money in practically every other area of marketing:
• If you’re spending £5,000 on paid search to send 50,000 new visitors to your website, but only 0.5% convert—this creates 250 customers.
• If you increase your conversion rate to 2.5%, a spend of only £1,000 for 10,000 new visitors would yield the same results—250 customers.
Optimisation makes other campaigns more effective. Imagine approaching a restaurant with an impressive exterior and adverts in every other paper, only to find out it was dirty and unappealing inside. You probably wouldn’t stick around to find out if the food was good.
Improving conversions gives you the freedom to explore other aspects of marketing, safe in the knowledge you can capitalise on the returns.
The first steps of Conversion Rate Optimisation
Okay, so a big deal has been made about there being no rules to CRO. This isn’t entirely true. There are some basic frameworks. Although testing, re-testing and implementing changes is what supercharges optimisation, companies do have to start from somewhere.
This is where heuristic analysis comes into play. Without initial data to compare, we have to optimise using common sense. But, pure guesswork isn’t enough:
Good heuristic analysis uses a structured approach. By learning from experienced sources, it’s possible to save time by taking shortcuts. Implementing changes with a good track record of success gives us an excellent starting base for future development. Looking at things from a customer’s viewpoint is invaluable:
MarketingExperiments created a well-known formula for conversion heuristics:
• Conversions = 4(user motivation) + 3(clarity of value proposition) + 2(incentive – friction) – 2(anxiety of sharing information)
• Translated: users will convert when incentives are clear and strong enough to outweigh the friction of the ordering process and their anxiety about sharing personal information.
We have to remove the psychological barriers to conversion (friction, anxiety) and boost positive incentive (user motivation, value proposition).
Let’s transform this from marketing jargon into something applicable:
Heuristic analysis framework
• User motivation: is your website’s content relevant to the user? Consider your demographic—who’s buying your product? It’s important your copy and design convey a clear message (more on this later).
• Clarity of value proposition: is your offer, service or product obvious? Your defining feature or selling point should be clear, whether this is quick delivery or quality.
• Incentive: how is value being communicated? This takes serious appraisal. Sales copy should create a sense of urgency, relevant information should be easily accessible and the perks of your brand should take center stage.
• Friction: is the look, feel and process of buying conducive to conversion? Confusing design without focus, slow page loading times and offensively long checkout forms can be the death knell for your conversion event.
• Anxiety: why wouldn’t someone trust your website? Look for normal signs of authenticity (payment gateways, SSL certification, working phone numbers).
What next? Usability testing involves tracking an individual’s actions and thoughts while using your website. This offers unparalleled insights into psychology and problems that might not be clear elsewhere.
Usability testing reveals the most serious issues that designers miss. Working on a project can often make us blind to problems. Having seen our work thousands of times, we can never capture a ‘first impression’.
Scavenger hunts are a popular way to put websites to the test. By setting users specific tasks we can see whether our designs work functionally. If a user is simply looking for information, it follows that the easier it is for them to find this, the more time they’ll be willing to spend ‘being converted’ elsewhere.
Whether you’re using a heatmap to track mouse movements (or finger taps on mobile devices), a scroll heatmap to see how long users engage with content, or software to record a user’s session, we’re ultimately looking for the same data: what’s stopping people from converting and how can design be optimised.
Let’s look at how to implement this further:
How do I optimise design?
“Simplicity is the ultimate sophistication” – Leonardo Da Vinci
Design should be the cornerstone of your optimisation efforts. Unlike other elements on a page, design is the only thing with instant impact.
Our brains can decide whether we like the look of a website in 50 ms—according to research from 2006. So, what have we learnt from over a decade of responsive design?
Visual appeal matters. 38% of consumers will leave because of a bad design. Because we process visuals so quickly, it’s unlikely people will stick around long enough for good copy or product features to save bad design.
Some quick wins
Basic symbols of trust can immediately make customers more receptive.
Product reviews, testimonials, phone numbers, returns options and SSL certification are all fish in a barrel when it comes to optimisation. They make people feel relaxed and clearly signpost your website as a shop. Most shops look the way they do not because their design is the best, but because they remind people of other shops.
• Example: when we see websites with an HTTPS connection (the green padlock symbol next to the URL bar) we automatically recall the countless other websites with this kind of authentication (Google, Farfetch, YouTube, gov.uk).
Using elements like this in a harmonious way is a recipe for relaxed consumers:
Think about the whole
Visual hierarchy plays an important role in design.
Your Call-to-Action (CTA) is usually a button which initiates the conversion process (examples: “Sign Up”, “Learn More” and “Buy Now”). Located on any page of your website, it’s the place you want users to engage most.
Common sense dictates that we should make CTAs as large and as bright as possible. While this is better than the alternative—one measly link getting lost in a sea of images and text—it can be overkill.
Having a visual hierarchy means balancing elements so little is needed to emphasise your CTAs.
• Example: If your website is primarily black text on a white background, even a small grey button is going to pop.
An individual element only stands out because of all the other elements on the page (either distracting from or drawing attention to it). Make your CTA stand out by slightly boosting its size or colour while muting everything else. This will help you avoid the pitfalls of loud, overcrowded websites.
Follow one vision
Having a unified look helps customers ‘buy-in’ mentally. Online, this unity means you can guide people towards your CTA without disrupting flow—making purchasing decisions seem organic rather than forced.
Use a small number of fonts and colours. Good design, like a film score, should rest comfortably in the background. People know it’s there and appreciate it, but it shouldn’t distract from the main content.
Visual appearance is the most important factor for 93% of consumers when looking at new products. This is why online shopping works so well. By letting your product take center stage with large images, you’re tapping into your audience’s innate buying habits.
Don’t let overly clever web design talk you out of a sale.
Let white space do the work
White space is simply a part of your website with nothing going on. When text is centered, 3 columns are created: 1 column in the center for text and 2 blank columns on either side—this is white space.
White space is what describes the page. In those first 50 ms, we use blank areas to navigate areas of text and colour.
White spaces come in two flavours:
• Macro: the space between main elements on the page (e.g. the space between the header and the main body of text)
• Micro: the space between smaller, individual elements (e.g. the space between images and their text captions)
Micro white space improves readability by spacing things out. This makes people want to read what you’ve written. Macro white space can be used to signpost visual hierarchy. A well-structured web page will lead visitors through to the checkout process.
The more white space you have, the quieter and clearer your site will become. Customers will hear what you need them to, rather than an onslaught of competing noises.
How do I optimise copy?
“You have got to hit them where they live in the heart or in the head” – John Caples
We’ve written a more in-depth article about good copywriting here, but, let’s look at how to tie things into heuristic analysis. If you were to approach your site, how would the text sway you?
The average visitor will only read 20% of your website. Why? If the content doesn’t speak directly to them, they skim read.
Creating engaging copy means knowing who your customer is. Before expert analysis or testing, and without a lot of user data, we can make some general assumptions.
Firstly, everybody hates long sentences. Secondly, overly complex language makes people think you’re stupid (erudite vernacular utilised irrespective of necessity). Thirdly, people like reading things that sound like a human wrote them.
You know your product and your customer, therefore you know people’s motivation for visiting. You also know what your value proposition is. Use language that emphasises this in a way that doesn’t break character (e.g. sophisticated fashion brands shouldn’t be promoting “Cheap Deals”).
Write simply using common language. If navigation is signposted strangely, this will create huge friction. What’s more, an inability to engage emotionally with your content causes anxiety—how can people trust someone whose writing lacks empathy or human touches?
… one last check
Once your website looks good and reads well, you have to make sure it feels right. Technical problems are the quickest way to lose customers, but the easiest issues to solve. Slow loading times, broken links and missing images will decimate your potential customer pool.
How to analyse your customers
“Lean isn’t stab in the dark. It’s stab in the dark AND then listen carefully for screaming” – Alistair Croll
In terms of The Lean Startup cycle, a book by Eric Ries many consider the best prototype for small-medium businesses, we’ve covered the ‘build’ phase—largely through heuristic analysis and standard conventions. What about ‘measure’ and ‘learn’?
This is the heart of Conversion Rate Optimisation. By analysing the data, we know where to focus our attention.
Quantitative data from analytics tools lets us know what our customer’s journey looks like: what they’re doing, why they’re leaving, where they’re coming from. Automated tools give us the best results because they can process millions of users while filtering data automatically.
We have to make everything relevant trackable. This is where tag managers and session tracking comes into play.
Using a tag manager
A tag is a snippet of HTML code that sends information when pages load or users perform a certain action. This way, third-party tools such as Google Analytics can record user sessions in detail.
Google Tag Manager lets companies deploy tags without changing their source code. From within the manager, triggers and variables control how and when tags fire. By using variables, it’s possible to define specific criteria for firing the tag.
• Example: a trigger can fire a tag when users remove items from their shopping cart while the variable sends more information about the user (referral URL, which HTML elements were present on page..etc).
Analysing the data
Using this data, we could find out which kind of customers were failing to convert—and where. Because the data is quantitative, it tells us how changes affect conversions (whether this is the conversion itself, or steps in the process).
Once set up, your analytics tool, in conjunction with tag managers, should track the entire buying process from landing page to checkout or exit page.
Set up a funnel to visualise traffic. This will clearly show where users are dropping off before conversion.
• Example: 200,000 users look at a product, 100,000 add it to cart, but only 1,000 checkout. From this basic data, you know that your conversion issues are in the checkout process.
By segmenting data it’s possible to discover trends. Say IE11 was sending you lots of traffic, but converting at only 0.3%, while Chrome users were converting at 2%. This would immediately set you off in the right direction: there are likely technical issues with IE11.
How to perform A/B tests
“A pinch of probability is worth a pound of perhaps” – James Thurber
Now you have the data to find and resolve problems, it’s time to let changes loose in the real world. Testing is the final step—the ‘learn’ phase of the Lean Startup cycle. This is what makes good CRO so powerful: the lack of guesswork.
A/B testing means implementing and testing live changes on a number of users. Software splits users into two streams: one where changes are visible and one where they are not. This means you can compare data from the largest number of genuine customers.
From the smallest addition to complete design overhauls, it’s now possible to measure the results with a data control.
Who is A/B testing for?
A/B testing is ideal for websites with an existing customer base and the resources to test frequently. You’ll still need heuristic analysis and an expert eye to pinpoint issues, but A/B testing means you can ensure every change is moving things in the right direction conversion-wise.
Small, incremental changes over time reap serious benefits.
Real life example: EA’s website promoted the release of SimCity by letting people pre-order. Obviously, hype was the name of the game. A large banner across the top of the page offering a cashback incentive wasn’t having a positive effect on sales, so an A/B test was performed:
• The version of the website without the offer message generated 43.4% more purchases
Customers were already excited for the game, so further incentives actually dissuaded people from purchasing.
How does it work?
Just like analytics, in order to perform A/B testing effectively, you’re going to need a program to handle the data and implementation for you—typically with as little obstruction to existing code as possible.
Why is this? Changes need to be quantifiable, quick to implement and seamless. For larger sites, downtime and bad user experiences can lose thousands of valued customers.
The standard A/B testing process works as follows:
• Variations of the same landing page are created (normally the original control page and an A and B version of the hypothesis) while the testing software generates a script to display these to different users at random—recording results
• ⅓ of visitors will see the control page, ⅓ will see the A version and ⅓ will see the B.
• The testing software will process data from this (often using tag managers to record conversions, basket abandonment and bounce rate)
• Once the test reaches statistical significance, it will stop. If the winning variation was either the new A or B versions, these will then be implemented as a permanent fixture into the site.
• Following this, you should reflect on the winning version and whether it matched the hypothesis and plan the next test. You should always have a test running or be planning the follow up.
We now have clear information contrasting the site with and without changes
The right tools for the job
With tools like Google Analytics ‘Content Experiments’, Optimizely and the recent Google Optimize, testing sits on top of existing structures.
Google Analytics Content Experiments:
• Contained within the internet’s most popular free data analysis software, Content Experiments lets users set up split-test pages
• After creating variations of the same page, plug URLs into GA and grab the resulting script for your original page’s header tags
• Large quantities of raw data are then pumped back into GA for analysis
The only serious downside of Google Analytics is the sharp learning curve for inexperienced users. Content Experiments is built into GA, giving you a huge but complex pool of data which can be bewildering to amateur data analysts.
• Much simpler WYSIWYG interface while cheaper than other non-GA options
• Intuitive system with option to pay for more advanced features later on
• Possible to create variant pages easily from editor while defining goals
Although Optimizely has a smaller learning curve, the pay-per-user testing model and features only accessible through Gold Plan (like multivariate testing) can be inhibitive for smaller companies. However, basic testing is extremely powerful and ideal for businesses learning data analysis. Overwhelmingly the most popular tool in this industry.
• Unique in its use of Bayesian method (as opposed to frequentist): updates predictions using new data—able to hypothesize whether page variations will have positive/negative impact on conversions and to what extent for the first time
• Displays results in Google Data Studio for easier comprehension
• Fully integrated with Analytics: simple implementation
Released this year, Google’s newest addition is designed as a successor to Content Experiments and a free version of Optimize 360 (which costs around $150,000). Considerably easier than Experiments and with all the functionality of premium testing software, Optimize could have a significant impact on the testing marketing in the near future.
Choosing the right tool enables regular user testing: the standout feature of successful ecommerce companies.
Putting things into practice
Good design should let users do what they want easily while guiding them towards the checkout.
In order to achieve this, you will always need to focus on what works for your site, and your site alone. Heuristic analysis based on existing concepts is useful, but ultimately, testing is the only effective way to increase conversions—bar incredible blind luck.
Conversion Rate Optimisation groups several strategies and ideas about ecommerce under one umbrella term. As this happens, software companies increasingly look to small businesses for inspiration. Effective CRO is still a complex undertaking, but in the right hands, it’s moving from intangible art to repeatable science day-by-day.