Conversion Rate Optimization : Definition (not what you think)
It's not about quick wins!
I'll be transparent with you: the majority of guides / articles on CRO [.highlight-text-default]are wrong[.highlight-text-default].
CRO is not about adding a popup that asks for an email, having clear and bold CTAs, or displaying reviews on a product page. These are good practices (or not), but they are not CRO.
The reality is that these tips are generalized and don't fit your branding, audience, or business. In fact, most of these tips may simply not work on your store.
More importantly, they won't have any impact in the long run. Let's say that adding a popup increases your conversion rate today, that's great congratulations. But tomorrow, what else are you going to do to improve your conversion rate?
In this guide, I'm going to teach you a data driven method that will allow you to continuously improve your conversion rates. The goal is to give you a concrete action plan to implement a [.highlight-text-default]long-term CRO strategy[.highlight-text-default] on your store that will allow you to improve your results and user experience a little more every day.
Each percentage of improvement gained on your tests will turn into several dozen on your conversion rates in total: the snowball effect.
The CRO Loop
First of all, to set up a CRO strategy on an ecommerce site I use this method: [.highlight-text-default]The CRO Loop[.highlight-text-default].
It is divided into 5 steps, each of them having a specific objective:
- Analyze: identify deficiencies, problems on your website
- Hypothesis & Solution: understand why these problems exist and how to solve them
- Prioritization: define which actions to implement in priority
- Testing: set up an experiment to obtain data on our hypothesis
- Learns: analyze the results to draw conclusions
I call this method the CRO Loop because your goal is to perform continuous testing. Each test will give you results that will open up new ideas for tests to set up, and so on: it's an iteration loop.
1. Analyse
The first step is to identify performance gaps on your website but also to identify problems of your potential customers. You have 2 types of analysis to perform: [.highlight-text-default]Quantitative and Qualitative analysis[.highlight-text-default].
Quantitative analysis
The quantitative analysis is based on your data. We will try to identify anomalies or performance gaps on your site at different points. This analysis is quantitative because we base it on the amount of data collected. The more data you have, the more relevant your analysis will be.
3 questions arise:
Which tools to analyze your data?
To analyze your data, I recommend you to use the most common analytics tools like Google Analytics, or Mixpanel. Other tools like Hotjar can help you go even further in data analysis by directly analyzing the behavior of your visitors on your site.
If you don't have an analytics tool implemented today, I recommend you to install one as soon as possible! Without a tool, it will be difficult for you to go further in your CRO strategy.
What data to analyze?
This is the question that comes up most often when I teach CRO. With hundreds of different data sets being collected, it can be complicated to know where to start in data analysis.
I recommend starting with a Macro analysis : the idea is to analyze your sales funnel at each step and observe the visitor drops.
A classic ecommerce funnel is Homepage > Collections > Product > Cart > Checkout. Of course, you need to adapt it to your store, and especially identify several funnels: your visitors may arrive on a product page first, not go through the cart if you have quick checkout buttons, etc.
[.highlight-text-box]Macro analysis is crucial, because optimizations at the sales funnel level are the ones with the most impact on your conversion rate.[.highlight-text-box]
Once this first analysis is done or you have sufficiently optimized your funnel in the past, you can move on to the Micro analysis: this time, you will segment your analysis further by going deeper into pages, audiences, etc.
Here are a few ideas to explore:
- the most visited pages
- time spent per page
- geographical segmentation
- old vs new users segmentation
- most viewed / purchased products
- most viewed collections
- Scroll depth
- devices
[.highlight-text-box]When implementing a CRO strategy, you will always start with the macro phase. As your results improve, your analysis will become more and more refined to find new areas of improvement.[.highlight-text-box]
How to analyze your data
Your objective is to identify anomalies or performance gaps. Depending on your niche, your conversion rates can differ enormously, so I advise you to use your own data as a reference. You'll have to use common sense if you start from 0 to ask yourself if a data is good or not.
Some examples of performance gaps:
- a 10% add-to-cart rate with a 1% purchase rate
- average time spent on your store less than 5 minutes
- 40% drop between adding to cart and viewing cart
Some examples of anomalies:
- the delivery page is the 3rd most visited page
- one collection has 5x less visits than the others
- conversion rate Homepage > product > purchase is 50% lower than other funnels
Also take the time to observe your good performances, these will help you to define tests:
- which products make your best sales?
- Which categories are the most consulted?
- Which medium converts the best?
Keep this data aside for the rest of the process.
Qualitative analysis
Analyzing data is good, but it is not enough for an optimal CRO strategy.
The problem with data is that it doesn't necessarily reflect the needs / problems of your visitors. To identify them, there is only one solution: get feedbacks.
Customer support
If there is one place that is in constant contact with your customers, it is [.highlight-text-default]your support department[.highlight-text-default]. Make sure you report recurring problems concerning your products, the use of the store, etc. A gold mine for your CRO strategy.
Product reviews
Whether it's about your products or those of your competitors, you must absolutely observe customer reviews! Very often, they share their frustrations and problems encountered with their purchases.
Here is a very concrete example:
This customer finds that a competitor's yoga mat she purchased doesn't smell right.
[.highlight-text-box]This point can be exploited on your product page by specifying that your raw materials are odorless for example.[.highlight-text-box]
Communities
To go even further, consult the communities around your niche to identify more and more opportunities to exploit: Facebook groups, talks on Reddit, questions on Quora,...
At this stage, you have identified a lot of ways of improvement to improve your conversion rates, well done!
Let's move on to the next step.
2. Hypothesis & Solution
This is my favorite step. The one where you have to understand why these problems exist and define a solution to solve them. This is the time to be creative!
Hypothesis: why?
You need to determine why you have performance gaps on your site (quantitative analysis). For each anomaly identified, make a hypothesis about [.highlight-text-default]the origin of the problem[.highlight-text-default].
Here is an example:
- Your conversion rate from "Add to cart" to "Place order" is 10%.
Hypothesis on this anomaly:
- the shopping cart page is difficult to access (icon too small / not clear)
or
- the shopping cart page is not reassuring enough
or
- the visitor would like to know the delivery costs before the checkout
[.highlight-text-box]These are 3 examples of assumptions, you need to define only one. To do this, use common sense, but above all, don't be afraid to "make a mistake" with your hypothesis. You can carry out endless tests.[.highlight-text-box]
Regarding your findings during the qualitative analysis (customer feedback), the hypothesis to define concerns why he encounters a problem on a product.
If we take the example of the yoga mat that does not smell good, identify the reason that causes this problem (example: the raw materials used to make the mat).
Solution: solve the problems
To summarize, we have :
- identified problems via the Analysis step
- made assumptions about the source of each problem
Now we need to define solutions to solve these problems in order to improve our performance.
Again, we need to use common sense to define the most appropriate solution.
However, there are some important points to consider:
- the solution must be sufficiently differentiating to obtain results.
Example : changing the font size from 12 to 14 will not help. - the solution must not corrupt your previous tests: if you have modified an element following a successful test in the past, be careful not to affect it with new tests indirectly (unless it is voluntary of course).
- the solution must be consistent with your branding: we are not looking for quick wins but for changes to be implemented in the long term. When you define a solution, keep in mind that it must fit naturally on your site.
Finally, I recommend that you start with a Macro solution (similar to the analysis step):
- Macro solutions: these are significant changes that can bring important results. These are solutions that impact your sales funnel, your branding, your copywriting, your pricing, etc.
- Micro solutions: these are less impactful changes like changing a title, image, color, adding additional text, etc.
[.highlight-text-box]Always start with the Macro tests as these are the ones that will get you big results quickly. Of course it's not always easy for an established brand to change its branding for example. On the other hand if you have just recently launched, testing a completely different branding can give you a new direction to your online store.[.highlight-text-box]
3. Prioritization
A large part of the work has been done, congratulations! We will soon be able to launch the tests, but before we do so, you'll have to prioritize your tests to optimize your time: a test can last several weeks (I'll come back to this point later in the guide), it would be a shame to launch the less relevant ones first.
To prioritize, I use a well known method called ICE. This method allows you to prioritize your tests according to 3 factors:
- Impact: how much do you think the solution to test will impact your store's results?
- Confidence: Do you think this test will work? Do not hesitate to use previous tests to avoid biasing this factor.
- Ease: Is this solution complicated to implement? Changing a title will be easier than adding a popup for example.
→ For each of these points, put a score out of 10.
→ You will get a score ranging from 3 to 30 points (10 + 10 + 10)
Then sort your tests in descending order: the higher the score, the quicker the test should be implemented. This is a simple but effective method to prioritize your solutions to test.
[.highlight-text-box]I got something for you! I made a Google Sheet template to build your CRO strategy. Feel free to use it.[.highlight-text-box]
4. Testing
It's time to evaluate if the solutions you have defined will improve your conversion rates or not. We are going to test these solutions, more precisely to carry out A/B testing.
In concrete terms, we will compare the current version of your store with a test version containing the solution to be tested: this is called an A/B test.
What tools should I use to perform an A/B test?
There are many tools to do A/B testing, but I highly recommend Google Optimize: it is free and integrates perfectly with Google Analytics which will allow you to refine your analysis. I recommend this tutorial if you want to get started with the tool.
If you are already at an advanced stage of your CRO strategy, tools like AB Tasty, Optimizely or Kameloon will help you go even further.
What objectives to track?
When creating a test on Google Optimize, you need to define an objective: what is the goal of your test? I recommend that you create an objective on Google Analytics that corresponds to the statistic you want to improve and select it on Google Optimize.
If your solution aims to improve the conversion rate on your Checkout, track the "Purchase" objective, etc. In any case, I recommend tracking the add to cart and purchase as secondary objectives because you can have nice surprises (example at the end of this guide).
Can I do several tests at the same time?
You can, just make sure they don't impact each other. I'll give you an example not to follow:
You run a test on the product page, and another on the shopping cart page. If you track the purchase conversion during the first test, the results will necessarily be biased because the visitor will go through the shopping cart page, which is itself undergoing a test.
[.highlight-text-default]Remember: beware of overlapping tests[.highlight-text-default]
How long should an A/B test last?
Generally, we will focus on the volume of conversions and not the duration of the test. For example, if you made 10 conversions over 2 weeks, it is not enough to draw conclusions. I will come back in detail on the volume of conversions needed in the next part.
On the other hand, if your store generates a lot of traffic and consequently a lot of conversions, I recommend that you leave the test for 3 to 4 weeks. Your visitors, especially recurring ones, will tend to be surprised by the test version that they do not knowThis can bias the results.
Usually the results will smoothed out to have similar performances after 4 weeks. If the test had been stopped at only 2 weeks or earlier, you take the risk to have irrelevant and misleading results.
[.highlight-text-default]Remember: leave a test in place for at least 3-4 weeks.[.highlight-text-default]
5. Learns
It's time to analyze your results!
But first, we need to make sure we have enough data to draw conclusions.
Is your A/B Test Statistically Significant?
In the previous section, I recommended that you wait a certain amount of time before stopping your test. But you also have to wait for a certain amount of Data before your test result is significant.
I won't go into details but concretely to minimize the risk of false positive, the result of your test must have a confidence index higher than 90% minimum (95% ideally).
Don't panic, tools like this one allow you to know in a few seconds if you have enough Data to have a significant result.
If it is not the case, continue your test for a while.
Drawing conclusions from your tests
When a test ends, you can face 3 situations:
- The test is neutral: the test and original versions have similar performance
- The test is winning: the test version has significantly better results
- The test is losing: the original version has significantly better results
On the other hand, look carefully at the performance of the final objectives (purchases, additions to cart,...), you may find yourself in these situations:
- The test is neutral / losing BUT your final conversions increase significantly
- The test is neutral / winning BUT your final conversions drop significantly
In these two cases, the test results are opposite to the performance of the final objectives. This can lead to further improvements, lessons learned, etc.
Here is a concrete example:
During the support of a client, we conducted a test on the homepage to reduce the bounce rate. The test consisted in displaying blocks redirecting to the product categories at the top of the page (the original version displayed ALL the products without distinction).
The result? The bounce rate did not improve, there were still as many visitors leaving the site.
BUT we did see an increase in final conversions: more additions to the cart, and more purchases.
How do we explain this? Our Data analysis showed that directing visitors to categories leads them more easily to the products they are interested in, and therefore increases performance.
This is typically the case of "the test is neutral / losing BUT your final conversions increase significantly".
This is why it is important to track the entire funnel regardless of the initial objective of the test.
So when to implement a test?
Only implement a test if it improves the whole funnel and not only the objective of the test.
And then what do we do?
The great thing about this method is that [.highlight-text-default]you can iterate your CRO strategy[.highlight-text-default] over and over again.
Each result obtained during a test will give you new lessons allowing you to continuously identify ways of improvement, and therefore new tests.
Thank you very much for taking the time to read this guide, I hope it will help you improve the performance of your ecommerce stores.
Feel free to share in comments your analysis and results if you want my opinion 🙂
If you want to get the next guides in early access, join the newsletter.
.png)
.jpg)
