The ROI of Conversion Rate Optimization for e-commerce

In e-commerce, an overall conversion rate of 1% is considered average. But that means 99% of the people who come through the door leave without purchasing anything. Based on Compass’ study of e-commerce stores totaling $61 Million in monthly revenue, 25% of the top performing stores reach a 2-3% conversion rate. This means that the average store can double or even triple their performance just by optimizing conversion rates .  

A conversion rate is the percentage of users who take a desired action. For example, your site is visited by 100,000 people during the month of November. During that month, 2,000 users bought something from the site. Thus, the site’s conversion rate is 2,000/100,000 = 2%.

Benchmark Your Store’s Conversion rate

Click on the link below and find out for free.

Most people wonder how they should optimize their conversion or not, because they don’t know what a good conversion rate is for their type of product/price. As a result, most people fail to get the full knock-on effect that conversion improvements can have on their businesses.

Let me give you a real example: An e-commerce website we know came to the conclusion that product pages with at least one review in them performed twice as good as the ones that didn’t. They then looked at their reviews page and realized that with a simple change they’d increase the number of reviews they received by 5x.

That was it. One targeted improvement and nearly double the sales from their long tail pages, producing a monthly revenue increase of $100k while costing only $15k to make the changes. And profits will grow even more because the cost of implementing this change is not increasing.

The moral of this story is not to “go out and improve your reviews page.” The moral of this story is that this company used data to find a major bottleneck for a key metric and then worked relentlessly to improve it.

As you can see in the graph below (built using data from  Compass), while the average conversion rate in e-commerce is below 2% (blue line), the top performing companies (orange line) can get conversion rates of more than 5%, way better than the average company. It’s important to note, however, that as the graph shows, conversion can vary heavily depending on the price of the product.

Conversion Rate vs. Average Transaction Value

Data shows that 30%, 50%, 100%, 300% improvements are entirely possible. But for that to happen, you need to be serious and systematic about analyzing data. At Compass we strongly believe that businesses can make extraordinarily positive impacts if they use data to:

  1. Learn how they stack up against their peers
  2. Focus on the area with the highest opportunity
  3. Make targeted changes
  4. Understand if what they’re doing is working
  5. Repeat the process

Once you’ve learned how the best performers in your segment are doing and start making impactful changes in areas that present the highest opportunities, your goal should be to to be your own benchmark. That’s the time where you can work hard on always having better numbers than what you had last month.

In this article we’ll look at how shoppers behave in the conversion funnel. We’ll then understand how to use data to be smart about what to improve in order to improve conversions. Finally we’ll learn the process that can help you drive consistent improvement over time.

1. Understanding shopping behavior

Before we go any further, we need to first understand shopping behavior. This includes the behavior of visitors who might never shop with you, no matter how much you optimize your site.

1.1 The myth of perfect conversion

Tim Ash introduces “The Myth of Perfect Conversion” in his book Landing Page Optimization.  According to Tim, there are three types of visitors to your website:

  1. Noes – Those who won’t ever take the desired action
  2. Yesses – Those who will always take the desired action
  3. Maybes – Those who may take the desired action

Optimizing your site for conversions is very much like a politician campaigning for votes. You should concentrate on the Maybes and ignore the Noes (they won’t buy from you, no matter how hard you try) and the Yesses (they’ve already decided to buy from you, no matter how clunky and buggy your website may be). It’s the undecided ones that make a difference.

Landing Page Optimization

The group of undecideds contains a wide variety of people. For some of them (the Yes-Maybes) one small improvement might get them to convert. Others might need a significant effort in persuasion and hand-holding to come around. While the remainder will forever be out of your reach. That means, the maximum conversion rate for your site is limited to capture the rest ofMaybe-Maybes.

Chet Holmes, in his popular book “Ultimate Sales Machine”, estimates that only 3% of customers are in the market to buy most products at any given time. The rest will fall into four distinct phases of the buying cycle:

  • 7% are open to buying your product, but not actively pursuing it
  • 30% are neutral, meaning they haven’t even thought about your products (but may be interested)
  • 30% think they are satisfied with their existing solution (but not completely unreachable)
  • 30% are completely uninterested in your product or service

A great way of optimizing conversions for the Maybe-Maybes and Yes-Maybes is to use email marketing automation. According to John McIntyre, founder of ReEngager in this post for Shopify, Email is now the second best performing channel for e-commerce businesses, behind search. He makes a good point in saying that email series can improve conversions in different stages of that funnel, for example:

  • Abandoned Cart email series is a great tactic to improve conversions among the 3% of people who are ready to buy from you, but only need an extra push.
  • Welcome email series can work very well in bringing the 7% of users who are not actively pursuing your product.
  • And Email nurture series can help increase the likelihood that your brand will be the choice for those who are neutral and even those who are satisfied with a different solution.

This is important because it will help you organize your acquisition efforts strategically. Some channels tend to play an early and assisting role in the typical sale, others are more likely to be the last interaction before a purchase. This will differ from company to company, so you need to plan and track the channels that will bring your customers from a Maybe-Maybe stage to a Yes stage.

1.2 Different types of buying decision behavior

Another aspect to have in mind when optimizing for conversion is understanding how the buying decision works for your type of business. There are great differences between buying a toothpaste, a packaged holiday, a personal computer or a car.

The type of decision process your customers need to go through to buy your product will affect conversion rates. So by understanding buying behavior you’ll be able to optimize your conversions accordingly.

For example, if the buying process is complex, your conversion rates will be lower. So you need to make sure that the customer spends as much time as possible on your site (and will return to finalize the purchase, if they didn’t already). If it’s a simple buying process, your aim should be to get them through the funnel as fast as possible.  

What makes a difference in purchase decision behavior is the level of involvement needed with the product before buying. There are four main types of involvement:

  1. Complex Buying Behavior
  2. Dissonance-reducing Buying Behavior
  3. Variety-seeking Buying Behavior
  4. Habitual Buying Behavior

1. Complex Buying Behavior

This type of purchase requires a high degree of engagement with significant differences between brands and their offerings. This is often the case when the product price is high, there are low levels of certainty about the purchase, low quality of after sale service and so on. Good examples of this are buying mobile phones or laptops. Both are expensive, complex and are sold by a variety of brands. That makes consumers feel uncomfortable when deciding what to buy.

In this scenario, the consumer journey, from being aware of a product, to buying it, becomes highly complex (see graph below).

Consumer journey graph

This poses the challenge of attribution. What acquisition channel was responsible for this sale? There are many different points of interactions that get consumers closer to buying a product, which also makes the conversion optimization process very complex, with companies having to optimize many touchpoints with their consumers to improve conversion rates.

2. Dissonance-reducing Buying Behavior

Just like Complex Buying Behavior, consumers with Dissonance-Reducing Buying Behavior need high amounts of involvement. However, buyers in this behavioral situation perceive very few differences among the brands they are selecting products from. The keyword here is perceive. There may be many real differences between the different brands, however the buyer’s beliefs about the other brands are that there are very similar or essentially the same.

After the purchase, consumers may face dissonance post-purchase behavior, also known as “buyer’s remorse”. Post-Purchase Dissonance will begin once a consumer begins to “notice” any disadvantages of their purchase, and begin to hear “good” things about the other products they did not buy. If you think that this might the case, you need to work on communicating the value of the product after the sale has been made.  

3. Variety-seeking Buying Behavior

This occurs when there is low involvement, but the consumer perceives significant differences between the brand options they have. In variety seeking situations consumers tend to do a lot of brand switching. There is no real brand loyalty. Good examples of this are cookies and crackers. 

In this case, consumers may pick another brand purely out of boredom or to just try something different. The marketing strategy might differ for the market leader versus the competitors trying to grab market share. Leaders should encourage habitual buying – dominating shelf space and keeping shelves stocked, and running frequent reminder advertising. Marketers should encourage variety seekers to buy by using lower prices, special deals, coupons, samples, and ads that have messaging that give reasons for trying something new.

4. Habitual Buying Behavior

This happens where consumers have low involvement in a purchase, and perceive very few significant differences between brands. Products in this category are those of everyday use, such as toilet paper, salt and black pepper.

In these scenarios the buyer behavior doesn’t go through the normal belief-attitude-behavior sequence. Because consumers are buying based on brand familiarity, it’s useful to have your marketing department use ad repetition to build brand familiarity, emphasizing only a few key points. You should also use more visual symbols and imagery in your website and adverts, because they can easily be remembered by the consumer and associated with the brand.

2. Using data to be smart about optimizing conversions

If you are interested in fixing your conversion rates yourself or if just want to know more about it before hiring someone, this section will give you a good understanding of how some of the experts in the industry work on conversion optimization.

It’s worth remembering that working on optimization makes sense for businesses that already have an acceptable conversion rate (you can learn about what is an acceptable conversion rate for your business using Compass’ dashboard). If you don’t, then you have more fundamental problems, such as attracting the wrong customer or having unrealistic product pricing.

Sean Ellis, founder of Qualaroo says in this article on Optimizely that you should start by first finding your best opportunities and then understand your visitors needs, so that you can get insights into your own products. The first is done with quantitative data analysis, while the latter is done via qualitative data from customers.

According to Sean, you should begin your quantitative research by using your analytics to uncover the following:

  • Top 5 highest bounce rate pages
  • Top 5 abandonment points in your funnel
  • Top 5 most valuable pages to your business

A high bounce rate means that your visitors aren’t finding what they’re looking for, or are frustrated by not being able to take an action that they want to.

We interviewed Ryan Koonce for this article. He is the Founder and Managing Director of Whopping Ideas / SaaS Management Group

Ryan’s tip to find abandonment points is to look at places on your site where you lose the most traffic is: “Take a look at user pathing reports in your analytics. Look at a pathing report that goes backwards from your final conversion goal to understand what the steps were that many users took prior to reaching that page. You can start to understand what the reasons were that they ended up at that final conversion point.”

Ryan also suggests a great way of using data to work on the areas of your site that will generate the highest impact on your revenues:

“Use the 80/20 rule that says 20% of your products make 80% of the revenue. So you should be focusing on optimizing conversions for 20% of the products. And in fact, you probably should focus on 5% of the products, because they probably make 50% of revenue. So you want to use data to find out what your best sellers are and who is buying those products. Go and get better lookalike audiences to put in front of those products so that you can convert those consumers into sales.”

3. The conversion optimization process

Unfortunately, there are no silver bullets we can give you that will magically make your visitors convert more. Instead of looking for that one tactic that will make your conversion magically improve, the focus of your optimization process should be on learning.

There is a proven method that has been consistently helping companies find the best converting tactics for their business. It is called High Tempo Testing. The idea is to speed up the learning process of your company by helping you run as many experiments as possible in a given period.

This is a system pioneered by Facebook’s growth team and adopted by many high growth stage startups as well as larger companies like Twitter, Uber, Airbnb, Hubspot, Mint.com and others. The idea behind high tempo testing is simple. The more tests you run, the more you’ll learn what works and the faster you’ll improve your numbers. See below a chart from Twitter (put together by Sean Ellis) that illustrates this point:

Twitter Accelerated Testing

The process borrows from the scientific method and the agile methodology of software development. It helps you design experiments, implement fast learning cycles and do more tests in a short period of time.

This type of testing can be applied to any part of the customer journey, but it is especially suitable to conversion rates optimization. It will help you optimize for each specific item you have on your site.


The scientific method applied to conversion rate optimization

Roughly speaking, the scientific method of discovery is conducted in five stages:

  1. Ask a Question.
  2. Do Background Research.
  3. Construct a Hypothesis.
  4. Test Your Hypothesis by Doing an Experiment.
  5. Analyze Your Data and Draw a Conclusion.

Online businesses have borrowed some of these precepts to elaborate a process that, as opposed to traditional management ideas, looks more like a science than an art.

In a fast moving world such as the online market, fast-paced learning is crucial for any company’s survival. In such an environment, High Tempo Testing provides a scientific framework to test ideas (hypothesis), originated by observations and descriptions of phenomena (data), via carefully designed experiments (new features, split testing, etc.).

As the name “high tempo” suggests, speed of execution is a crucial part of the process. Because the consumer environment changes so quickly, “the cost of being wrong is less than the cost of doing nothing (Seth Godin)”. It’s therefore more important to do more tests in a given period than it is to take the time to formulate the perfect experiment.

Differently from the scientific method (where the quality of the experiment is more important), conversion rate optimization is all about speed. That doesn’t mean there aren’t any standards in testing, but it means that more learning happens outside “the lab”, i.e.: on the live site and not in the meeting room of your office. The more tests you run, the more quickly you’ll learn what works.  

Backlog of ideas

The first thing you need to do is to gather new ideas to be tested in a document that helps you prioritize by topic, urgency and the amount of work it will take to carry each test. It is better to have a big list of ideas than to stop and brainstorm solutions every time you need to come up with a new test.

This is why we suggest building a backlog of ideas that are ready to be tested. It will make the entire process faster. This involves talking to people in your entire organization (and especially the customer support team) on a regular basis. Encourage everyone to give you ideas and celebrate the good and bad ones they have.

While gathering these ideas, find a way to catalog and organize them. I personally use Asana, but Pipefy and GrowthHackers.com have been working on promising products for this. What’s important here is the process. Just make sure your ideas are organized by:  

  • Idea name: This needs to be as clear as possible for people to understand them.
  • Lever: Where is this idea going to create impact? In our case: conversion rates is the obvious candidate, but you can run tests to lower bounce rates or increase page speed, for example.
  • ICE score: This is where you give a weight to help prioritize the ideas, ranging from 1 to 10.
  • I = Likely Impact of the idea on the metric you’re testing
  • C = Confidence you have that this idea will work
  • E = Ease of implementation

Weekly meeting       

The meeting is the most important aspect of the High Tempo Testing method. It’s where you’ll be presenting the results from the previous week, assimilating the learnings you’ve had during the process and defining the tests you’ll be running next.

Because it’s so important, there is a lot of work that you (as the meeting organizer) will need to do beforehand:

  • Throughout the week: Assess the progress of the tests and make sure everything is being implemented in time.
  • 3 days before the meeting: Inspire the team to come up with more ideas to pitch in the next meeting, based on the focus area.
  • The day before: Review the data from the tests and the overall company data and select ideas from the team that will be presented in the meeting.

Running tests

The most common type of test you’ll be running is A/B testing, which tests one hypothesis against another or against a control/incumbent tactic. Tests with multiple variations being tested simultaneously are called Multivariant tests.

While this might depend on the hypothesis you’re trying to prove, as a rule it’s recommendable to favor simple A/B tests instead of Multivariant tests. The reason for this is that Multivariant tests can get extremely complex and might affect your ability to analyze the results.

For the same reason, it’s recommendable to test one specific assumption at a time. So when testing two variations of a landing page, for example, try to make one change to the variant page and test it, instead of making multiple changes. This will help you know precisely what factors are driving improvements.

A/B tests are preferable when testing hypothesis because you’ll have more confidence that external factors didn’t affect the results. This is usually what happens when you change something on your website without doing any A/B testing (i.e.: watch compare data from different intervals).

A common question that arrives when you don’t A/B test is: Did the conversion go up because of the change we made or for some other reason?

According to the Kissmetrics blog, the right way of doing A/B tests is:

  1. Decide the minimum improvement you care about. (Do you care if a variant results in an improvement of less than 10%?)
  2. Determine how many samples you need in order to know with certainty that the variant is better than the original.
  3. Start your test but DO NOT look at the results until you have the number of examples you determined you need in step 2.
  4. Set a certainty of improvement that you want to use to determine if the variant is better (usually 95%).
  5. After you have seen the observations decided in step 2, then put your results into a t-test (or other favorite significance test) and see if your confidence is greater than the threshold set in step 4.
  6. If the results of step 5 indicate that your variant is better, go with it. Otherwise, keep the original.

For all of this to work you need to be concerned about sample sizes. According to Ryan, “Sample size is one of the most important and misunderstood elements in the testing area. Let’s say, somebody has 200 clicks, of which 2 became a conversion and they’re trying to make a decision based on that. You don’t have enough data yet.”

Small businesses can (and should) be running tests as well. It’s more challenging for them because they don’t have as many people coming to their website each week. But they just need to run the tests for longer periods of time, until they have a sample that is large enough to base their decisions on.

Another important thing to consider is segmentation in traffic. As Ryan puts it: “Where did the traffic come from? If 100 clicks came from an irrelevant advertisement in a magazine or some mention on a twitter post, of course the conversion rate is going to be low. So there are two pieces to this. One is understanding that a test needs to match the intent of the traffic and another the sample size needs to be great enough so that you don’t make uneducated decisions.” 

4. Conclusion

Start by identifying whether the investment into conversion rate optimization makes sense. Compare your data with your peers to learn if your conversions are below the top performers (or the average) in your segment. You can use the Compass dashboard to analyze that.

Then try to focus on Yes-Maybe customers. They are the 7% who are open to buy your product, and just need a little extra push to convert. Begin by using data to learn what are your best selling products and run tests to optimize every step of the customer journey towards buying them. Keep an organized process to help you be as scientific as possible about the tests and the data you are collecting, so that you won’t make uneducated decisions.  

Sadly, there isn’t a recipe for optimizing conversion rates. Each company needs to watch out for different metrics and understand how each segment of their customer behaves. But you can implement a process that can help you identify bottlenecks and optimize for your needs. If you stay disciplined with your process, you’ll eventually see your conversion numbers rise, which will in turn reflect on sales and revenue.