Why Conversion Rate Optimization & Data Science should be together

If you’re heading any marketing website or an owner of business which generates revenue through website then your primary goal is to increase the conversions on the website and you’re probably doing Conversion Rate Optimization without realising it.

What is Conversion Rate Optimization (CRO)

Let’s first understand the concept of Conversion rate. Every website will have some purpose or goal. E-commerce website would want users to purchase product, a Lead generation website would want users to fill the lead form on its website.

If your website receives 1000 users per month and 100 make a purchase or fill up a lead form then Conversion Rate is 10% and your aim is to increase this percentage. Hence the term – Optimization.

How Conversion Rate Optimization works

Well, there are many ways to do this. Crux of this process is very simple. Study the user behaviour across the conversion funnel and see where the user is dropping off without filling up the form (or making purchase). Make changes to the funnel and see if user converts with this change.

Ex: Any typical website will have the following structure:

Home Page ā†’ Product Listing Page ā†’ Product Description Page ā†’ Conversion

With the help of Analytics tools like Google Analytics or Adobe Analytics you can get the number of users who are visiting each of the pages (or funnel). Through data if you find that most of the users are dropping off from Product Listing Page and not moving to Product Description page, then you could conclude that there’s some issue with that page.

After thorough investigation you find that the ‘button’ which takes user to next page is not clearly visible and hence users are not moving ahead. You then come up with a recommendation to change the colour of the button (also referred as CTA – Click To Action).

This recommendation has to be tested before actually making the changes. You can run A/B test where the change is shown to 50% of users and Original version is shown to remaining 50% of users. You then measure the performance. If the change is performing well in terms of conversion rate the then it can be shown to 100% of users.

What are the Challenges with CRO

  1. It is a bit of a traditional technique where you just look at page level numbers. Conversion rate can have effect due to various other factors on the website. Ex : What users have seen on home page can have significant impact on the conversions even though they visit two more pages before they actually convert. In the normal analyses you might not give weightage to home page but focus on other intermediate pages.
  2. It is difficult to assess the impact of different elements across the website which will help in conversion. There can be hundreds of pages and thousands of buttons ( or CTAs) on your website. Users can visit your website from different devices and different channels (Social, Search, Direct etc). They can visit multiple combination of pages before they convert. You can’t incorporate the effect of all of these in your regular analyses.
  3. The tools you use for CRO, typically Google Analytics or Adobe Analytics do not provide the data at user level. This means that – the tools give aggregated data for users but you might want to study the behaviour of a specific user. Also, Tools do not give a lot of flexibility in terms of data slicing you need for the analyses. There are standard reports which you could use to get numbers which you can’t alter much. The solution here is to use tools that provide click-level or user level data (Ex: Google BigQuery)
  4. In a regular CRO analysis it is impossible to guess the “Willingness” of a user to convert. For ex: If a user has visited your website and has done some activity, then you can gauge if he’s gonna fill the form or make a purchase. If you find that he’s not gonna convert then you could take some actions while he’s browsing – like, providing offers – and make him convert. The Willingness to convert is not easy to get using normal analysis.

Conversion Rate Optimization + Data Science

Data science is a technique that is gaining popularity across industries including CRO.

With the data science it is possible to look at website level data and find out the impact of each page and element on the conversion. There are many proven mathematical techniques to achieve this.

Data science techniques rely mostly on session or user level data for building models which is more efficient compared to regular CRO techniques.

Another advantage with Data Science techniques is finding users willingness to convert, which is given as a score. This can be done live while user is browsing on the website or can be done offline. Live score would help in converting users while they are on website. Offline score has few other advantages.

You can look at historical scores and compare that with the actual conversions. If the score is decreasing then you can say that user experience on the website is getting worse. On the other hand if the score is increasing and Conversion rate is decreasing then you can think that there’s some issue at the end of the journey with conversion form or purchase form.


We’ve understood about Conversion Rate and how it is used in the business environment. We’ve also seen the some of challenges with CRO and how Data Science can help in overcoming those challenges.

Hope this has helped you to gain some insights on the working of both CRO and Data Science. I’ll write more about this in future posts.