Make the switch from Google Optimize 360. Learn why A/B is better with AI.

AI Conversion Rate Optimization: A Complete Guide to Getting Your Best ROI

If you’re hoping to build or grow a successful business online, then getting visitors to your website is only part of the challenge. You also need to make sure they take the action you want them to when they get there.

See, more traffic isn’t always the answer when it comes to growing your revenue. After all, SEO is a minefield, content marketing is a constant battle, and PPC advertising costs rise year-over-year.

But learning how to squeeze the most out of the traffic you are getting can be a very cost-effective route to higher profits. Attracting potential customers is expensive, so you want to give yourself the best chance of realizing that potential. This is where conversion rate optimization with AI comes in, allowing you to maximize what you’ve got, helping you to capture more leads, decrease bounce rate, increase efficiency, grow basket size, get more orders—basically helping you to boost whatever key metric you are keen on increasing.

Free ebook

A/B Testing Revolutionized:
How to leverage AI and ML to continually improve CX

Get your ebook

Evolv AI’s conversion rate optimization (CRO) helps serve the best experience to each audience, every time. Our AI-driven solution takes the guesswork out of typical CRO and helps enterprise companies achieve sustained growth.

But let’s step back and take a look at a quick conversion rate optimization example. If you run an online store, then your key conversion rate is the percentage of people who visit your site and complete a purchase. Say you spend $100 to attract 1,000 people to your site every day. If 50 of those visitors each buy something from you, that gives you a 5% conversion rate (50/1,000 = 5%). Now if you make changes to your site that encourage some more of those 1,000 visitors to buy from you, that means you’ve already increased your revenue, without actually having to grow your traffic.

Say you can grow that conversion rate from 5% to 5.25%—a relatively modest increase of 5%. Even this level of improvement can have incredible ongoing value. A 5% rise would mean an extra two customers a day—that’s potentially 14 more sales a week, 56 sales a month, nearly 700 sales a year.

Now imagine if you’re able to improve your conversion rate by 10%, 20%, even 30% or more?

You can quickly start to see how high-impact improvements can mean thousands of dollars in extra revenue every month.

There you have conversion rate optimization in a nutshell. You’ve made your website work smarter and harder, increasing your revenue in the process. Now let’s take a deeper dive into CRO, to explore what it is, how to do it and how it can help take your business to another level.

Getting started with conversion rate optimization: the basics

Let’s start with the basics—how do you work out your website’s conversion rate? This all depends on what you quantify as a conversion for your site. Conversions, in simple terms, are the desired actions that you want a person to take when they visit your website. A common type of conversion is a sale. If you run an ecommerce website, one where you are selling goods and taking payments directly from your site, then your key conversion is when you make a sale and collect the cash.

But not all sites sell goods directly. Instead, there are other actions used as measures of success. Another typical conversion would be lead generation—this could be for a B2B site that asks interested visitors to submit their details through a form, ready for somebody to call them back. SaaS websites often like to get new leads into their funnel by getting them to sign up for a free trial. So they would record free trial signups as an initial conversion, but then also record how many of these free trial users turn into paying customers.

The formula for working out your conversion rate is quite simple:

(Number of sales / Number of visitors to your site) x 100 = Conversion Rate

So for a website selling shoes, if you have 1,000 visitors in a month and 50 of those visitors buy a pair of shoes in that month, then your conversion rate is:

(50/1,000) x 100 = 5%

Now, if you were to increase this conversion rate from 5% to 6% by making the journey to purchase on your website more user-friendly, then you would bring in an extra 10 sales for every 1,000 visitors, which over a year would be an extra 120 sales.

Luckily, you don’t need to do the math every time you want to measure this. Set up your event tracking and conversion tracking in your analytics platform and this will tell you the conversion rate and allow you to drill down into the data even more. For example, you will be able to compare conversion rates across different devices, different browsers, different landing pages and so on. It is this sort of insight that will help you to identify pain points, blockers and problem areas that are stopping your site visitors from converting. And where there are problems, you can come up with potential solutions—this is how you can generate hypotheses and come up with the tweaks you want to test to try and improve your conversion rates.

Drilling down into the data is a crucial part of the CRO process. It is only through proper research and data analysis that you can eradicate the need for subjective opinions. Attempting to do CRO without data is just guesswork.

What is conversion rate optimization?

As the phrase suggests, conversion rate optimization is all about optimizing your site, so that it performs as well as possible to help you achieve your business goals. Conversion rate measures the proportion of visitors who ‘convert’—how many make a purchase, fill in a form, download a white paper, or whatever the key action is on your site that you want them to take. In general terms, if you can grow your conversion rate then it is great news for your site and your business.

But the process of conversion rate optimization has come to encompass many more measures of success than just your website’s conversion rate. While conversion rate itself is an important metric for most sites, your main goal should be to utilize your online presence to improve the overall performance of the business. This could be through attracting more conversions, encouraging higher value sales, learning more about your target market, reducing costs, or a combination of all of these.

A true optimizer’s work should not end just at boosting the conversion rate; you also need to take a more holistic view of the impact upon the wider business. After all, if you reduce your prices by 75%, the chances are you will get a better conversion rate. But that doesn’t mean you will make more money, as the new prices may be below your cost. For a lead generation site, flooding your sales teams with low-quality inquiries may cause more harm than good if they are unable to properly service the high-value leads. So, despite the name, conversion rate optimization is actually about much more than just conversion rate.

True optimizers need to blend skills of business analysis, marketing, web design, user experience, customer satisfaction and customer service. You need to know that your changes will bring long-term benefits to the business and to do that you need to be able to properly drill down into the data.

The key ingredients to a solid CRO process

The most important thing to remember with CRO is that it’s a process. It is not a one-size-fits-all formula to send your sales soaring. But when approached systematically and with an effective method of measuring success, it can drive long-term, sustainable improvements to your business goals. The key ingredients to this process are research, hypotheses, testing, and implementation. It’s a cycle that can be repeated on an ongoing basis.


If you want to make a success of your conversion optimization program, then it’s essential that you have solid research behind it. It might seem tempting to jump straight into changing elements on your page, redesigning landing pages, blindly copying what other sites do, or running A/B tests based on subjective opinions.

But what works for one site may have the opposite impact on another. To really have a long-term impact, you should take the time to understand your customers and—most importantly—understand their pain points. What is it about your website or your business that is stopping them from converting in the first place? What are the barriers preventing them from signing up or buying from you? Get to understand this and then you can come up with solutions from there. Solving these pain points before is like the secret sauce to website success.

Some common sources of insight when carrying out CRO research:

1. Website analytics

Google Analytics, the free analytics platform from the search engine giant, is the best known and most widely used tool out there with up to 30 million sites or more currently using it. Other options include Adobe Analytics, Woopra, Clicky, Piwik. But whatever platform you use, the important thing is you delve into your reports and use the data to help you identify patterns, problem areas and the pages where your site is leaking money or losing would-be customers.

Here are some good starting points when using analytics to identify CRO opportunities:

  • Compare metrics such as bounce rates of different landing pages to identify which are your best and worst performing pages. This helps you to prioritize where you should focus your CRO efforts.
  • Check your site speed and see how this is impacting conversions. Slow load times are a conversion killer—40% of people will abandon a site that takes more than 3 seconds to load.
  • View your funnel visualization reports to see what stage of the journey you are losing visitors. You’ll see what pages or sections are causing users to abandon the path to conversion, then you can drill down to find out what is causing them to leave.
  • Compare performance across different browsers. If your site has bugs stopping it working properly on certain browsers, such as Internet Explorer, then this could mean a proportion of your visitors are not able to convert even if they wanted to. If this affects enough visitors to justify fixing the bugs, then you could see a quick uplift in conversions from visitors using that particular browser.
  • Segment your visitors by device. This doesn’t just mean desktop versus mobile, but drill down into specific handsets and machines as well. Similar to the browser issue, if you see that users on a Samsung mobile convert at a significantly lower rate than iPhone users, for example, then you could have a compatibility issue. Fix that and you’ve suddenly boosted your chances of securing more sales from Samsung users.

2. Heatmaps, mouse tracking and form analytics

These are all different forms of web analytics, but give you a visual representation of how real people are using your site. Heatmaps represent how people interact with different elements on your site, for example, using colors to show which buttons or links are most often clicked or pressed. Mouse tracking allows you to see the mouse or finger movements that your site visitors use to move around the page, while form analytics can show you actually which form fields are being filled in or abandoned. The more you learn about the behavioral patterns of people on your site, the easier it is to identify the barriers, pages or elements that are causing hesitation, confusion or deterring them from progressing any further towards conversion.

Here are some examples of the sort of insights these visual analytics tools could give you:

  • Showing you which elements your visitors are clicking on instead of clicking on your preferred call to action button
  • Identify the particular field on your sign-up form that is causing users to give up and abandon
  • Demonstrating whether or not users are scrolling far enough to see your call to action, or paying enough attention to your key messages.

3. Customer surveys and polls

You are not your customer. The only way to really understand what they want is to get out there and speak to them. Focus groups and customer surveys have existed for brick and mortar businesses for decades. Luckily, the tools available to carry out this sort of research online has made it affordable and accessible for businesses of any size.

The voice of the customer is essential to understanding what they want from your site, what is stopping them from converting or how your business can help them. Surveys and polls are a form of ‘qualitative research’ (as opposed to ‘quantitative research’); research that explores the sort of opinions, motivations and real human insights that you could never get from raw data or numbers alone.

You could use exit surveys (surveys that pop-up as you’re about to leave a page), live chat boxes or follow-up emails to ask customers a couple of simple questions about their experience and what stopped them from completing their purchase. These nuggets of information can be invaluable. For example, you might see numerous customers mentioning that unexpected delivery costs on the checkout page put them off, so they left your site to find somewhere that offers free delivery. If this is enough of an issue, then it gives you a business case to go to whichever department in your company is responsible for shipping costs and deliveries and see what other options you could cost-effectively offer. If the expected increase in profits from the extra sales outweighs the additional cost of offering the free shipping then it makes business sense to try it. For example, men’s shoe retailer 2BigFeet reported a 50% increase in conversion rate overnight when they introduced free shipping. How much could a 50% uplift in conversion rate be worth to your business?

4. User testing

User testing is perhaps the most direct and immediate way to see how real people negotiate their way around your site. It is important to observe what they’re doing, how they’re interacting, what their reactions are—but also to encourage them to think aloud and vocalize their thought process.

User testing often involves scenario-based questions, where you set the user tasks and ask them neutral questions to try and gauge realistic reactions to a typical customer interaction with your site. For example, if you run an ecommerce site that sells electrical equipment, you could set them a task of finding a particular brand of speaker and adding it to their cart, or even asking them to complete the purchase using a dummy or pre-paid credit card so they experience the full buyer funnel.

These live tests can be done remotely online, with various sites available to recruit users who will carry out your tasks while recording their reactions using their computer, for you to watch back later. User testing is also often done in person—you can do it simply and inexpensively in a coffee shop or a meeting room, while you make notes and record the test on your mobile phone. Or, if your budget allows, there are more sophisticated set-ups such as a testing lab, where observers can watch through one-way glass, while every movement, sound and button click is recorded. Advanced tools such as eye-tracking cameras can be thrown into the mix for even more advanced user insights. But however complex you want to get with your equipment, user testing is a simple concept—you want to observe how your users interact with your site, learning more about what does and doesn’t help them to achieve your desired goal.

Develop your hypotheses

Hopefully, your research has thrown up a long list of potential issues and areas for improvement on your site. Your research should also include a solid understanding of how your site is performing now, what your key metrics look like now and what the business drivers are behind wanting to improve them. Establishing this baseline is important for being able to measure the impact of your CRO strategy and will also help you to prioritize what to tackle first.

Once you have your list of issues and problems, you can then analyze them in more detail, identifying recurring themes and the biggest issues, or the issues that are having the biggest impact on your business goals. Once you know what you want to solve, you need to come up with a hypothesis for how to do it—you can do this for each and every problem that you’ve identified. A hypothesis is basically a way of saying, ‘If I change X, then X will happen, because of X’. It is a short, sharp way for you to narrow down what you will change to fix the issue identified, what you expect to happen, and the reasoning behind it. Solid hypotheses are essential to a successful CRO strategy. Testing is only as good as the hypothesis behind it; otherwise, you are just throwing mud at the wall and waiting to see what sticks.

To return to the free shipping example, say you identified through your site analytics that you have a high exit rate from the checkout page and you’re aiming to increase the number of people that complete purchases once they’ve added items to their cart. Your user research showed that unexpected shipping costs were a recurring cause for complaint in surveys. Your theory is that if you remove the shipping costs then more customers will complete their purchase. So your hypothesis could be: ‘If we offer free shipping on orders over $25, then the number of completed purchases will go up, because customers will not be put off by unexpecting shipping costs.’

You should come up with a hypothesis for each issue that you think is having a negative impact on your conversions. From there, you would brainstorm ideas for different designs, new wording, alternative layouts, perhaps even new pricing structures; then you can test these against the original version to find out what moves the dial in the right direction.

A/B and multivariate testing

Testing is at the heart of the CRO process. In fact, if you asked many online marketers to describe CRO, then there’s a good chance most of them would mention A/B or multivariate testing. A/B testing is often a starting point for those getting started with conversion rate optimization. On a basic level, it allows you to test different designs of your key web pages, using statistical analysis to measure which design had the biggest impact on conversions. Running two variations at the same time is A/B testing—for example, you test two different photos on one of your key landing pages. Half of your visitors will see one version of the page, the other half will see the second version with a different photo.

At the end of the test, once you have attracted enough visitors, you will hopefully learn which photo has the biggest positive impact on sales so you can then roll out this winning design permanently for all your visitors. It gives you a high level of confidence that your choice of photo is optimal for delivering the best results possible. That is because a winner is only declared once the results have reached ‘statistical significance’—this where the action or conversion you are measuring has been taken by enough people for you to be statistically confident that the changes are not just down to chance. Your testing platforms will calculate this for you based on the number of visitors exposed to the test and the detectable difference in results between the variations.

Now let’s assume there are other elements on the page that can have an impact on sales, beyond just the photo. You might also want to test different copy, different call to action buttons, new headlines, different layouts. Running all of these as A/B tests would take a long time, and also wouldn’t allow you to easily learn which combinations of elements work well together. Headline A may perform better than Headline B, while Photo A gets better results than Photo B. But does Photo A do even better if combined with Headline B, rather than Headline A?

This is where multivariate testing (MVT) comes in. MVT involves testing multiple versions of a page at the same time, splitting the audience across each different version and again measuring which variation has the biggest impact. It allows you to test more design variations and learn more quickly than simple A/B tests, but the caveat is that you need high levels of traffic to achieve reliable results. For this reason, multivariate testing has often seemed out of reach for sites without huge numbers of visitors—the more variations you include, the longer it takes for a test to reach statistical significance, because the volume of traffic is being spread more thinly between different versions of the page.

Evolv AI solves this problem and opens up the potential of MVT on a much larger scale, by allowing you to test more and test quicker. It uses artificial intelligence to quickly learn what combinations of design elements work together on a page and, unlike with A/B testing, it gives you the chance to test hundreds, even thousands, of combinations all at the same time. The more you can test in a shorter space of time, the greater opportunities you have to find big wins through your CRO strategy.

How CRO can boost your business

The clever thing about conversion rate optimization is that it makes all of your marketing efforts more efficient. It is taking what you already have (traffic) and making it more effective. Incremental changes can have a huge impact on your business, with an ROI that can easily outweigh throwing more dollars into traffic acquisition.

With this in mind, there are various ways that CRO can add value to a business. But the important thing is that you focus on adding value in the right places. Is the process actually improving a business’s bottom line? If you are just focusing on showing an improved conversion rate in your Google Analytics reports, without being able to prove the value of these conversions, then your conversion rate becomes little more than a vanity metric.

And as we’ve mentioned, there are many other metrics that an effective CRO strategy can improve, beyond the headline conversion rate. There is customer lifetime value. Customer retention. Basket size or order value. Profit margin. PPC cost per inquiry. Reduction in advertising costs. Even a reduction in staffing costs. Here are some of the ways that CRO can help you grow your business:

1. Increasing the number of conversions

Let’s start with a key one. As we’ve seen, improving your conversion rate means a higher percentage of visitors to your site will convert. For an ecommerce site, that means more of the people coming to your store will turn into paying customers, adding products to their cart and completing the purchase.

There are endless volumes of research out there about the psychology of customers—what is it that makes a person want to buy from you? Luckily, CRO means you no longer have to guess. You can narrow down the elements that are important to your potential customers and test what works best for you.

The factors that have an effect on the buyer journey can seem endless, so your job as an optimizer is to identify the areas where you can have the biggest impact. For example, changing the design of a page, using more persuasive headline copy, improving your product imagery, adding more detail or removing unnecessary detail in your product descriptions—these are all examples of the sorts of things that ecommerce sites are testing on a regular basis.

By measuring the impact of your changes, you can see which version will get more customers ‘through the door’. Improving the number of conversions through CRO can have a huge ongoing impact on a business. Even incremental changes can really add up. If your optimization efforts generate a 3% increase in the number of people converting on your site in a month, what would that mean to your business? Now, if you can add on 3% every month then that adds up to a 40% improvement over the course of a year.

2. Getting the right leads through the door

As strange as it may sound, there are times when the best CRO strategy can actually involve reducing the number of conversions and lowering your conversion rate.

Imagine you operate a B2B lead generation site. A typical sales process might involve a potential customer filling in a ‘request a call back’ form. A member of the sales team then gets in touch to arrange a demo or a face-to-face meeting and works on converting this lead into a paying customer. But if you are selling a high-value product or service then you may get a lot of irrelevant traffic to your site, visitors who will never have the budget to become a paying customer.

In this case, effective optimization may be actually pre-qualifying leads before they fill in the form, so as not to clog up your sales pipeline with a load of junk leads. You want to remove the requests from ‘never-would-be’ customers, that only waste the time of your sales team, giving them less time to concentrate on the high-value leads. Some typical examples of pre-qualification on lead generation sites that you may have seen include asking a potential client to select their budget, specifically asking for a ‘work email’ address (and rejecting forms from generic email addresses such as Hotmail or Gmail) or asking how many employees your business has.

If you have been split testing different designs on your lead generation funnel, for example, then you will want to know the value of the leads that each variation generates, as well as the conversion rate. Evolv AI allows you to sync your offline sales data directly with your reporting dashboard. This helps you make more informed decisions. For example, one variation might generate 10 leads from 100 visitors, worth $50 each. Another variation might generate five leads from 100 visitors, worth $300 each. If you were taking conversion rate alone at face value, then the first variation would be declared the winner—a 10% conversion rate beats a 5% conversion rate, right? But when you plug the order value into the data as well, you’ll see that the second variation actually generated $1,500 for your business, compared to $500 from the first, higher converting variation.

3. Growing basket size

This leads us onto another benefit that CRO can bring beyond simply the baseline conversion rate. As well as looking to increase the your number of customers, another essential strategy for growing a business is increasing the order value from your customers. This could be through encouraging them to buy higher order products, add more products to their cart, or purchase additional related services.

Recommending ‘add-on’ products is a tactic widely used by Amazon and numerous other ecommerce sites. This doesn’t necessarily improve the headline conversion rate of your site, but it can grow the average order value of customers. You’re happy if a customer buys a product from you, but you’d be even happier if they buy two products. There are a variety of tactics that you have no doubt encountered when shopping online, designed to encourage you to buy more than just one item.

For example, when you view a product or add an item to your cart, you will often see ‘related products’ pop up. Right from within the cart you can present customers with other relevant items that they may also want to buy. Often this tactic uses social proofing by showing you what other people like you have bought, and by including it as part of the sales funnel you are already progressing through, it has removed any barriers and made it as easy as possible. Similarly, offering ‘add-ons’ that can ‘enhance’ your purchase is a common tactic. This might include insurance for a holiday you’re about to book, a protective case for a piece of tech equipment you’re about to spend a significant amount of money on, or an enhanced delivery option so you can get your item at a specific time.

These are all variations on an age-old sales tactic known as ‘upselling’. You’re already well on your way to securing a sale, now you want to try and persuade this customer to spend even more—and you’re making it as easy as possible for them to do so. Growing basket size can be a tricky optimization strategy, as it often involves testing changes across multiple steps or pages in your sales funnel. Your hypothesis for getting a customer to add a single product to their cart may involve running tests to optimize a standalone product page. But what is the best point to make additional product recommendations? A pop-up on the product page? Messaging on the checkout page? A design feature on the payment confirmation section?

But running tests across multiple pages has often been a tricky exercise using traditional A/B testing software, as it can be difficult to tell which changes have made the difference and your data can become polluted when trying to track potentially conflicting actions across multi-step funnels, impacting the validity of your results. But Evolv AI solves this, by using evolutionary algorithms to allow you to test hundreds of variations across multiple pages, all at the same time. Evolv AI learns how the changes to different pages impact each other, analyzing results much quicker and identifying full-funnel, holistic customer experiences that can have the biggest impact on order sizes.

4. Improving your return on ad spend

For many businesses with an online presence, online advertising is one of the most significant costs. Big businesses spend millions of dollars a month on PPC campaigns, while even smaller ecommerce sites can easily spend thousands during the busy season. As Google continues to evolve its pay-per-click model, search engine marketing gets more and more expensive.

The removal of sidebar ads in 2016, for example, meant getting into the top 4 ad positions became more important than ever if you wanted your business’s message to show above the fold to potential customers. More competition inevitably means more expense, higher costs per click and often higher costs per conversion. But the key to optimizing your ad spend doesn’t always just lie with improving the keywords you bid on or tweaking your ad copy. Landing page optimization is crucial to an efficient ad campaign.

If you have spent the money on driving clicks through to your site, then you are just wasting that money if visitors land on your site then bounce straight off it again. But the more of those visitors you convert into customers, the more return you will get for your advertising costs.

Conversion rate optimization allows you to test different design changes, measure the impact on your PPC traffic and discover the optimum variations that will stand the best chance of converting those costly visitors into paying customers. The logic here is clear—if more of your paid traffic buys from you site, then you have got more bang for your buck and made your ad spend more efficient.

Getting an ongoing return from your optimization strategy: Introducing a culture of optimization in your business

CRO can revolutionize your design process—iterative design can now be based on data instead of gut feelings. Choosing a design because the boss likes it is not a valid business reason. But choosing something because the data shows that it resonates more with your actual customers will have a much more positive impact on your bottom line. CRO has given rise to a new era of data-driven design, adding a robust underpinning to a discipline that has often in the past been at the mercy of whims, fads and subjectivity.

CRO continues to grow in its popularity, but it is much more than just a buzz term. It is a high-impact method of data-driven marketing, a process that can help inform decisions that allow a business to grow. It is also inspiring a cultural change at many businesses, whereby egos and opinions go out the window and new ideas are encouraged and rated based on their measurable impact on the metrics that matter.

Not all your tests will be winners—but that’s ok. No test is a wasted test, so long as you still learn from the results. If you haven’t successfully conquered a problem with your first test, then you analyze the results, hypothesize and iterate—test more solutions until you have a winner. The volume and velocity of tests that Evolv AI enables you to run is a game-changer in the world of conversion rate optimization. Not testing enough and not learning fast enough can make your CRO program inefficient—not testing is a huge missed opportunity. But Evolv AI lets you fail faster and win quicker, by using artificial intelligence to quickly learn what does and doesn’t work. The evolutionary algorithms allow you to test more variations at the same time and get to winning results quicker than ever before. More winning tests, means more improvements to your conversion rate.

There is no such thing as a perfect website, but the more you learn and the more you test, the more chance you have of getting closer to your own version of perfection.