Conversion Rate optimization -getting started

Today most of the marketing team are structured to drive traffic on the website to convert this into sales & revenue. Once this process starts to deliver the result, organizations adopt this understanding that online business is a place where they spend $1 to make $2 or $3 or even $10. But despite their excellent marketing effort, sales remain stagnant (because if it’s the case, then everyone is doing the same.)

That’s the reason Conversion Optimization comes into the picture because “Not all visitors” are created the equal. Also, as the name suggests, CRO is the ongoing, data-driven process of continually discovering and quantifying the most effective experience for your customers.

There are 3 Insane benefits of having CRO as a core business function:-

  • It’s not just a process of improving CR -its a process aimed towards optimizing different element of website in and out involved in conversion funnel including landing pages, UX design, on-site & off-site conversion.
  • As mentioned above, CRO helps in optimizing the website for higher effectiveness and higher conversion. i.e., Improve website visitors experience, landing pages, defining the right content, reducing the exit and bound rate.
  • Lastly, the important it helps in improving conversion rates & drives more Average order value by retaining your existing customer with personalization (like personalizing homepage design, interested categories, or preferred payment methods.)

The below diagram represents why conversion optimization is essential and where you can apply in your customer journey.Conversion Optimization Approach

Ready to take the first step towards #CRO – check out the below framework but why you need a framework?

The problem with the conversion rate optimization is that it looks easy; many of us having some digital experience can take a look of the website and quickly find problems (assuming these are the conversion blockers). But if you directly implement these ideas and not follow the proper conversion optimization framework, then it can impact the major customer journey. Therefore it is crucial to consider the following points:-

  • To protect the overall customer experience
  • Need to test/measure everything
  • Testing can take time
  • Not every test is a winner – Failing fast is a good thing
  • There is no fail or pass, but it’s a process to move temporary tests onto your website product plan as its a continuous process of data collection.
  • Analytics should be fully integrated with the process.
  • Need lots of ideas – based on research and data
  • The process should be developed and combined with an agile & quick, and also balances in place
  • Customers hate popups
  • What are the rules for certain design patterns? Overlays; Toaster; Alerts etc
  • Define spots and pages that marketing can update
  • Promotions may conflict and overlap
  • Ensure all ideas are well thought through with ROI

Conversion Research

Starting the Conversion Framework

The conversion optimization framework is deployed on a hundred website and many industry pioneers are strongly recommending it as it helps to remove the guesswork and increase conversion rates.

Step-1:- Conversion Research 

Conversion research is the first step in CRO process as it is the strategic approach focused on identifying/interpreting data to find possible points of friction in a conversion funnel (sales funnel) and ultimately allows you to actions against that problem statement.

Primarily there are 2 steps of data gathering and analysis- Quantitative/Qualitative way. Followed by this you can create a master sheet with all the issues that then turn into action items along with the expected uplift:-

Qualitative Analysis:-

  1. Qualitative analysis helps you to understand or establish what users like or dislike about your website, mobile app or your services? How easy it is for users to perform the core task? Would they recommend your business to others, if no then why?, What are their recommendation for improving your website?

You need to gather all these insights from users through surveys, usability/customer interviews, heatmap tools which will keep your finger on the pulse of your audience. Below are the primary ways which help you to establish these:-

  • Surveys: –Helps companies to listen to the voice of customers and make informed objective decisions based on what you know your users want.
  • Information Architecture: – Leverage different research methods within the same study for greater confidence in taxonomy like navigation structure
  • Heat maps: – Helps you to understand what users want, care about and interact with on your site by visually representing their clicks, taps and scrolling behavior.

Quantitive Analysis:-

  • Quant ITive analysis is the practice which focusses on collecting data on user behavior, understanding these numbers and interpreting with actionable metrics.
  • In a nutshell, quantitative research helps identify certain behavioral patterns and leaks in your funnel. It refers to hard facts and measurable data that you can use to quantify and uncover your visitors’ online behavior patterns. Tools like Google Analytics data/Adobe analytics, helps you understand the big picture of what’s happening in “every room” of your website.
  • These data will show you the ‘WHAT’ – what’s happening and where. And you can gain deeper insights into how users from your Ads campaigns engage with identify leaks to create custom site experiences using the audiences you’ve already created in Analytics.

Below are the examples which you can consider as the metrics:-

  • What visitors are doing on your site: – What activities they are performing on the site, which categories they are most engage in, where do they exist?
  • Which visitors convert the most or the least: – for example which is the best converting channel for your website and why?
  • What is the customer preferred journey?
  • What are the Leaks in the funnel:- You can find answers like HP to Cart conversion, HP.
  • Success rate & basket completion rate: – Registration success rate and basket completion rate.
  • Revenue participation of site section:-  like (Homepage, category, product page, or cart pages)
  • Types of website users: – New visitors, returning visitor, loyal customer, Registered but not purchased, Lapsed Customer
  • High entry points and exit pages?

Step 2 Hypothesis & prioritize the idea

A hypothesis is nothing but a prediction that you can create before running a test. It defines clearly what is being changed, what you believe the outcome will and why you think? Because the experiment will either prove or disprove your hypothesis.

With the information that you have found through Quant ITive or Qualitative data analysis, you can start brainstorming hypothesis for your test. Please don’t rely on competitor’s data – you need to always complement your findings with your data.

Then comes the prioritization with the uplift modeling

Prioritizing test is one the strategic approach as strong prioritization process helps conversion optimization research with the answer: “here’s what we’ll test, in what order, and here’s why.”

  • How big a difference: – that we can expect to see with the proposed change compared to the status quo?
  • Product: – Different areas of your product perform?
  • Funnel: – How each part of your funnel converts which will help you decide of an effect you’d need to see for the new change to be worth it.
  • Technical: – How much development work is required to graduate the test?
  • How strategically important is it?
  • Does this feature support plans?
  • What is the size of the audience or action are we optimizing for?
  • Consider a retest?

Uplift modeling:-

Uplift modeling also is known as incremental modeling, true lift modeling, or net-lift modeling is a predictive modeling technique that directly models the incremental impact of a treatment (such as a direct marketing action) on an individual’s behavior. To follow we have to apply in two ways:-

  • Power Calculation:- We can do a power calculation to find out how big of a sample size we need for our test. The point of power calculations is to find out what sample size we need for our A/B test.
  • Statistical Confidence: – How many views or users or form submissions or other interactions we need in each group to achieve the necessary power for our test.

Then we can finally start our test! Time to wait for those events to roll in.

Idealist and scoring

  • Based on the test idea hypothesis, uplift, and scoring model –we built a scoring model based on the impact on each of its one.
  • Also, each item goes through a scoring approach to understand the potential impact of executing this activity.
  • Further, it will allow you to ‘edge case’ ideas and instead focus on high potential activities backed by data.
  • Develop dedicated meetings each month with key stakeholders to review all of these ideas

Step 3:- Architect the data

Since conversion optimization is a continuous process of data collection and learns more about the customers, that’s the reason the success of ab test activity should not be measured based on pass or fail. But to architect the AB test in such a way through which you can capture the customer journey and enhance it for better effectiveness.

That’s the reason for designing a proper process flow is important in the AB test which can be achieved in the following ways:-

Customer Journey Mapping

: – The concept behind journey mapping is to apply maximum data insights.

  • Successful journey mapping means mastering the moment and maintaining the attention of your customers at every step in their buying journey.
  • “With journey mapping, you can evaluate the factors contributing to whether people are getting through the process you outlined for them and then determine the reasons why they are not converting
  • Kindly Note:- Before you can start developing an effective customer journey map based on insights from your data, you have to be able to access your data in an integrated fashion.

Customer journey analysis: – Each step of a customer with a process flow which helps to understand the visualization in a better way.

Functional document: – You can create a functional document outlining customer journey, why do you think you change is important, how you are tracking your test, how big is sample size. A functional document is nothing but a process document which defines what the problem statement is and what KPI you are measuring. Kindly request separately if you are looking a reference for a functional doc.

Step 4:- Execute the Campaign

Now, since we have brainstorm through data collected from Qualitative and quant ITive analysis, which then converted into well research hypothesis with quantified goals, it’s the time to execute these campaigns. Type of CRO campaign you can consider for

  • A/B test: –  A/B test allows you to compare two or more versions of your Web site content to see which best lifts your conversions, sales or registrations.
  • Multivariate test:- With Multivariate, you can Test many elements and variations. Multivariate less requires less traffic and fewer combinations than A/B tests require.
  • Auto target:- Auto-Target removes the guesswork and serves the most tailored experience to each visitor based on his or her customer journey and profile (for ex:- the behavior of previous visitors with similar profiles.) 
  • Personalization:- Personalization you can track and recommend your users as they move through your site for the sake of optimization and personalization. 
  • Recommendation: –
  • Recommendations activities display content/products that might interest your customers based on their previous activity or other algorithms. Recommendations help direct customers to relevant items they might otherwise not know about.

Once you’ve decided the type of campaign that you want to run, you also need to consider for how much time you need to run your test to achieve statistical significance and how much sample size that you need to consider.

Step 5:- Measuring the AB test data

There are multiple factors you need to consider while measuring AB test performance. Like a good research hypothesis, statistical significance above 90% of performance wouldn’t be conclusive to declare the winner and push to live.

Below mention, a few of the factors will help you to put a focus on these as the risk of deciding on bad data will be minimal.

Statistical Significance

What percentage of the time users are willing to be fooled into seeing an effect by random chance –this is called significance level (& more precisely null hypothesis)

Here are the great significance calculator tools you can use

  • Use this A/B Significance Test:-To compare two ads or landing pages and determine which performed better based on conversion rate.
  • This PCC Ad Testing Tool allows you to input more variables for a more concise analysis of click-through rate and conversion rate.
  • If you want to stick to Excel, I recommend checking out this video, which gives you more data surrounding I2C metrics for your campaigns!

Post-Test Segmentation

Post-test segmentation is vital to gain insights and maximizing revenue because statistical significance will only tell us the difference between variations and control but not with the users, as users are different and what resonates with one person, doesn’t work with another one.

Types of Segmentation Strategy you need to consider

  • ‘ ‘What segments have true value to our business?’ (Returning visitor or new users)
  • ‘What segments are actionable?’ (Desktop v/s Mobile)
  • ‘Which are most likely to be impacted by the test we are running?’ (Category X v/s Category Y)
  • ‘ What level of correction you should do in the account for the treatment of multiple observations

Validity Threat

  • One version one truth: – Always integrate with your testing tool with analytics to see if the revenue numbers match up.
  • Conduct quality assurance for every operating system and every device.
  • If you have a mobile app and a website, then it is recommended to run a test separately for each of devices types.
  • Stop your test only when you’ve sampled corrected sample size.
  • Data fishing into account and adjust your significance level to compensate for the number of tests you’re running.
  • Run test as long as necessary until your sample size has reached.
  • It is also recommended to run a test for full incremented weeks, which will help you to include the data from every day of the week and every time of the day.
  •  If you’ve run any campaign during Holiday season or any specific festival campaign, then the data you have collected is only relevant to that season itself.
  • Look at your annual data and identify anomalies (in traffic and conversions). Account for this when running your tests.

What’s next?

There are a lot of “best practices” out there, but ultimately, you need to find out what your customers respond to, and what drives results for your business. Here are three follow-up actions to get started with CRO today:

  1. Use the three formulas to start the CRO conversation.
  2. Leverage the PIE framework to help prioritize your strategy.
  3. Make CRO someone’s responsibility.

What CRO strategies does your business leverage? Share with us in the comments below.

Now you’ve got everything you need to double your conversion rate: data, insights, and a prioritized testing map.

Good luck!


Leave a reply

Your email address will not be published. Required fields are marked *


©[2018] Growth Valleys All rights reserved. Privacy Policy & Terms of Services.


We are closer than you think..!!


Log in with your credentials

Forgot your details?