Note: We put together a detailed presentation where we go over every step of the CRO Process. Click here to download it so you can use it as a reference in your future CRO projects.

Quick question: What do you want to get out of this post?

Before diving into the process and how to get started with our beginners guide to conversion rate optimization (CRO), it’s important to set the right expectations.

People tend to brag about their wins, that’s why we see headlines such as “How I Increased Our Conversion Rate By %200 By Changing One Button”, and this can sometimes be misleading.

Was the button supposed to be there but because of their negligence it wasn’t, thus causing visitors not to convert? Or maybe their conversions grew 200% from 1 conversion to 2 conversions? In most cases, they won’t disclose the actual numbers.

What worked for them, might not work for you

Huge increases in conversion rates are possible of course, and we were responsible for quite a number of those ourselves (and shared some of them here). But these tests aren’t always repeatable.

You can’t run the same test we ran on Avon and expect a 69% increase in conversion rate. And definitely don’t look at what Amazon is doing and expect it to work for your e-commerce site because it might not.

Not only is this a bad way to go about optimizing your site, but some tests might actually hurt your conversions.

My first CRO Experience: Adding $4 Million In Sales

I remember getting started with CRO for beginner course back in 2011, while I was working on my car insurance website which I have co-founded. We were facing a dilemma: We had a highly desired product and a channel with virtually no competition, but very low conversion rates.

You see, we were the first company to sell car insurance online, and as a result of our many marketing efforts we had a lot of traffic coming in and we also ranked #1 on Google for the term “cheap car insurance”. Things were going great on the top of the funnel – however, people weren’t converting.

We soon discovered why. But before I share the exact tactic we should go over the process first.

Making A Solid Commitment

As I mentioned here, Conversion rate optimization isn’t a short-term tactic. It should be treated like a long-term process and adopted as a mindset. It’s just like going to the gym – you don’t go to 3-4 sessions expecting to lose 20 pounds.

Training your muscles is about adapting to heavier and heavier weights while optimizing your conversions is about uncovering more valuable (and actionable) insights.

There will be a learning curve and there will be failed tests. Everything is difficult in the beginning, but the more realistic your expectations are the easier time you will have (and the more motivated you’ll be to continue).

You can expect things like:

  • Failed tests: It’s either a sign you should work on your hypothesis or a new insight which you didn’t expect.
  • Tests that take a long time to reach statistical significance: Means you probably chose too small of an audience.
  • Slow execution: Perhaps other people in your organization don’t view CRO as a priority, which is why you need to be the advocate for change

And so on. In every event there is a learning opportunity; a lesson in disguise.

Getting Started With Conversion Rate  Optimization

My advice for anyone starting out would be: Setting goals is important, but having a solid process if far more important. As James Clear explains when talking about Goals vs. Systems:

In a similar fashion, if you’re a CRO Specialist your goal is to increase sales by 20%, but your system is your CRO process. And it starts with:

  1. The Research Phase: Discover What Really Matters

Imagine going to the doctor about a headache. He wouldn’t just give you a pill to alleviate the pain (which is a symptom) and send you off like that. He would ask you questions and do some investigation first, in order to find out what the underlying cause might be.

Similarly, if you have a ‘conversion problem’ you start by asking the right questions.

Step 1: What’s this business all about?

The first question should be about your specific business type: Is it retention-driven? How mature is it? And what are the primary metrics that you focus on?

E-commerce shoppers have a different mindset than SaaS customers. The SaaS customer is more business savvy and might not be the sole decision maker, so they need more convincing (and in some cases longer copy).

Here are a few questions you should start with, in order to understand the business:

  • What type of business is this?
  • How mature is your business?
  • What products/services are you selling? Retention-driven or acquisition-driven? (that means either how mature it is or/and if it is selling products without repetitive buying cycles)
  • Who’s the buyer persona? What type of emotions are triggering the buying decisions?
  • Which are your main points of difference? Are you matching the points of parity in your market?
  • How are you segmenting your customers? Using any RFM model? The lifetime value?
  • Have you made conversion rate optimization before?
  • What’s the level of freedom you will have in order to run experiments?

After finding out this information, you will be able to clearly understand if your mission is to increase the customer acquisition rate, the retention rate or if you simply have no mission because the business is not yet in the phase where it makes sense to start doing conversion rate optimization.

Step 2: Web Analytics Audit

Someone once said that “If you didn’t measure it, it didn’t happen.”

That’s the first step in your conversion rate optimization efforts, it’s measuring the important metrics like Purchases, AOV, Customer Lifetime Value, Cart Abandonment and so on. Before you make any hypotheses, make sure your data is clean, accurate and more importantly… Check that you have all the right data.

This is the technical part of the Web Analytics Audit. The second part is analyzing your data in order to uncover hidden opportunities. Among other things, it involves looking at your sales performance for different products and different traffic segments. Not all visitors were born equal, which is why we aim to offer a more personalized experience for our different visitor segments.

Because the research phase is what guides your hypothesis, later on, keep an eye out for your most relevant visitor segments.

Segmentation is a crucial part of running a proper web analytics audit.


[su_quote cite=”Avinash Kaushik – Author, Digital Marketing Evangelist – Google, Co-founder – Market Motive.”]Whether you do online, offline or online analysis, or just like to randomly play with data, insights arrive faster with segmentation.In fact, I’ve gone so far as to say: “All data in aggregate is crap.” That’s certainly a bit melodramatic, but beyond the most bare bones “ahh, I see something is happening,” you can’t get anywhere with aggregate data.[/su_quote]

You might want to check his oldies but goldies blog, especially this article about users, sequences, cohorts or this one, about web analytics segmentation.

If you want to skip the line, before starting to optimize, make sure you understand your segments of visitors, based on acquisition, behavior, and results. Then, apply the Pareto Principle, get the insights and address the opportunities.

Segment #1: Acquisition

From which sources, locations and keywords are you getting 80% of your results?
Possible outcomes: traffic optimization, insights to run personalization on your website, etc.

Segment #2: Behavior

How are the users which have the best conversion rates behaving? How many pageviews are they seeing before converting, from which landing pages are they starting their journey?

Possible outcomes: discover the hidden gems on the visitors’ journeys, find the most important micro-conversions.

Segment #3: Top pages

Pages with the most traffic or highest exit rate.

Possible outcomes: improve UX for most visited pages.

beginners guide to conversion optimization

Step 3: Funnel Analysis: Where are they dropping?

A funnel is a series of steps you want your users to take, such as signing up or purchasing a product. You set a series of events in your Analytics tool such as the one below, and you will get a visual representation of how many people complete your funnel and how many drop off (and at which stage).

How you design your funnel depends, again, on your business type. It’s up to you which actions are most important, but here are some examples:

  • E-commerce: All Sessions > Viewed Product Page > Viewed Shopping Cart > Checkout > Order Complete
  • SaaS: All Sessions > Pricing Page > Registration Page > Registration Complete
  • Agency: All Sessions > Service Page > Contact Page > Inquiry Form

When conducting a funnel analysis, you might find that there’s a huge drop off in, say, people who view your product pages. Going back to our Doctor/Patient analogy, this page is like a clogged artery, and it’s an indicator that you need to dig in to find out why that’s happening there and how you can optimize your product page to promote a more healthy flow.

The issue might be with the diet (a specific segment of visitors isn’t converting), or it might be with the patient (the product page isn’t well optimized). So take both into consideration when doing your analysis.

Step 4: Heatmaps, Session Recordings & User Testing: Are there any roadblocks stopping them?

One of the ways to find out why visitors aren’t converting is to analyze their behavior, specifically on the page with the lower conversion rates. Session recordings are a great way to do that, although time intensive.

You can record 100 or 1,000 sessions and go through each one in order to find out if there are any issues or missing triggers on your page that are preventing customers from taking action. It might be a technical problem, bad design or a lack of persuasiveness on your behalf.

Heatmaps track a user’s mouse movement across your page so you can see where the most frequented elements on your site are and which sections get ignored.

You can also conduct several user tests where you bring in people from your target audience to your office and ask them to take a specific action on your site, such as “Use the search bar to find a blue t-shirt and then purchase it via credit card”. While you let them do this independently, you’ll ask them questions to help you understand what they’re thinking as they browse your site. You can also ask them questions after the test is over.

Source: Nielson Norman Group

You typically don’t need to run many user tests. As you can see in the graph above, the more tests you run the fewer usability problems you find and your time is better spent moving on and solving the problems you already found.

Step 5: Surveys: What are your customers’ pain points?

Surveys are a great way to uncover customer insights and they’re easy to implement. Their flexibility means that you can ask questions based on different scenarios, such as:

  • On a user’s exit intent
  • Based on answers to previous questions
  • Immediately after the page loads (or moments after)
  • When the user scrolls to a certain percentage of the page

And you should consider your timing, depending on the context and the question. You won’t trigger a survey that asks visitors “Why are you leaving?” right after the page has loaded, and if you have a survey on a blog post page, you might want to wait before displaying your survey until the user has scrolled down a bit.

When we ran International Ecommerce Day, we used the following survey to guide our landing page optimization efforts. It proved to be very insightful, since many visitors gave us feedback on why they weren’t sure about attending the event, and we were able to improve our copy to better answer those questions for them and for future visitors to come.

As a side note, make sure you have a way to contact your survey respondents afterwards. Omniconvert Surveys let you ask visitors for their contact information after they’ve answered your question(s), such as email address and phone number. Thanks to this feature, we were able to follow up with our survey respondents and answer their questions, thus converting many of them even after they left our website.

To create your first survey, follow this step-by-step tutorial that shows you how to create a survey using Omniconvert.

  1. Insights & Hypothesis Generation

Now that you gathered enough data and insights about your visitors, it’s time to form your hypothesis. A hypothesis looks like this:


And it can be defined as “A statement based on limited evidence that can be proved or disproved and is used as a starting point for further investigation.”

Remember what I said earlier about committing to the process? That last sentence explains why. It’s also important to have a strong hypothesis because that’s the basis of your experiments. You can’t test things on your side based on pure ‘feeling’. It can feel good (because it’s easy) but you’re going to end up with wasted time and money weeks later.

During this stage, you will make a list of hypothesis and experiment ideas. An experiment (hypothesis) should take into account the following:

  • Where: On what page(s) will the experiment run
  • What: What will you test
  • Who: Who will be exposed to this test

If you can’t conduct a proper audit with a strong hypothesis or are short on time and would rather have a team of specialists handle this for you, talk to one of our experts about conducting a Conversion Audit.

  1. How To Prioritize Your Experiments

Now that your have your experiment ideas, obviously you can’t run all of them at the same time! You will have to prioritize. The way you do this is by looking at several factors, such as:

  • Impact: Is it going to have a major impact? The larger the visitor segment involved and the more significant the change, the higher the impact will be. Stay away from things like “testing the button size or color” which don’t make a difference.
  • Cost: How much will it cost to run this experiment? Some experiments require that multiple people get involved (including outside collaborators), which can also mean more time investment. Prioritize the most time and cost-effective experiments.
  • Ease of implementation: Is the experiment too technical, thus requiring that developers and designers get involved? The easier it is to implement (provided that it’s impactful and cost-effective as well) the higher it should be on your priority list.
  1. Implementation

Finally, it’s time to implement your first test. There’s on thing I didn’t mention in the previous section which should also be taken into account: Make sure your experiment idea is doable!

Yes, sounds obvious, but if you’re really ambitious and want to test things that require making backend changes to how your website works, it might be too heavy (if not impossible considering the circumstances) to implement. You should now consider your CRO stack which might include the following features:

  • A/B Testing
  • Personalization
  • Surveys
  • Heatmaps
  • Session recordings
  • Advanced conditions and triggers

The last part applies to things like timing and segmentation, such as the different triggers we mentioned for surveys (on-exit, on-load, on-scroll) and the audience segmentation criteria such as whether that segment consists of people who took a specific action, answered specific surveys, is from a specific location and more.

Omniconvert has over 40 segmentation criteria to choose from and lets you run A/B Tests, Surveys, and Personalization all from the same dashboard. Give us a try here (it’s free, no credit card required).

  1. Check your results, learn and adjust

Excited yet? Before you stop any experiment, there’s one final thing to take into account: When to stop your experiments. When deciding whether to stop your experiment, there are three main things you should take into account:

  • Statistical Significance: Did your test reach statistical significance of 95% or higher?
  • Sample Size: Is your sample size large enough to consider a variation a winner?
  • Time: Did you run the test long enoughto reach a solid conclusion?

Peep Laja wrote a detailed post on When To Stop Your A/B Tests, I recommend you check it out. But here’s the gist of it: Statistical significance is important, but not if your sample size is too small. As Peep explains:

[su_quote cite=”Peep Laja, founder of ConversionXL”]“It’s human to scream “yeah!” and want to stop the test, and roll the treatment out live. Many who do discover later (if they bother to check) that even though their test got like +20% uplift, it didn’t have any impact on the business. Because there was no actual lift – it was imaginary.”[/su_quote]

While your test might show a 99% statistical significance, you should also look at your sample size. How many people were exposed to each variation and how many converted? The smaller the sample size, the larger the difference in conversion rates between your variations will have to be (and this doesn’t always mean it’s reliable).

Once you’ve run your tests, it’s time to analyze the results and reach a conclusion. Was the new variation a winner or did it lose to the control? Why is that? And how can this insight guide our future optimization efforts?


To summarize the process outlined in this post, here are the main steps:

  1. Do your research
  2. Form your hypotheses
  3. Prioritize your hypotheses
  4. Implement your tests
  5. Analyze your results and learn from them
  6. Repeat from step #1 (and remember what you learned in step #5)

Since we’re continuously learning ourselves, I’d love to hear your feedback and any questions you might have about this beginners guide to website conversion optimization. I also made a more digestible presentation about the CRO Process which you can use for future reference, click the banner below to get access instantly:

beginners guide to conversion optimization