Friday, April 7, 2017

Testing and Optimization

In this chapter, we will discuss how to do CRO testing and optimization. We will also talk about the various funnel stages and how to create tables.

How to Do It?

For testing and optimization, we have to first decide where to test and how to prioritize our goals and then optimize them accordingly. Your testing should revolve around customer centric perspective.
Fetch the data for a website to optimize it. Build a complete stack of information for the optimization process. Then redesign the website with a structured and systematic approach by keeping us well informed by data insights. Set objectives and KPIs.

Prioritize Testing Opportunities

Testing opportunities should be prioritized according to the values. We may not tackle all the values at once. Thus we need to set our priorities keeping in mind the gaps on Real Data and the time taken to achieve an outcome.

Funnel Stages

When talking about funnel stages, there are generally three falls to clear a funnel. The persuasive stage, informational and transactional stages with persuasion being the first one then followed by the other two.
Persuasion Stage − This is the entry point of your customer. It may be a Landing page or other pages. Here you have to make sure it makes the visitor feel he is on the right platform. It should scream you have almost everything he is looking for.
Informational Stage − It is here you have to provide a solution to visitors’ queries. Once the visitor is satisfied, he jumps to actionable buttons.
Transactional Stage − Once the customers decide to convert, he moves to this stage. It is just a step back towards the GOAL. This stage usually includes a conversion form. Optimizing transaction stage means touching up the form, reducing or adding tabs, making interactions and anxiety free purchasing or converting.
Now which step should optimize first is totally up to you. It hardly makes a difference provided it is done keenly and with a customer-centric approach.

Potential Pages

Potential pages are the foremost prioritization pages. Although every page has a scope to improve, potential pages are pages of utmost importance to you. As you cannot test all the pages in one go, you need to prioritize your potential pages separately. Take help from web analytics, statistics and you can mix it with intuition.
Let us elaborate it a little more. For instance, if you are an ecommerce company, the cart abandonment is your major concern. Analytics reports cannot justify the exact reasons like late shipping. To tackle this issue you need to go through product pages, category pages, headers footers, etc.
Tip − Look for problem indicators. In this case look for high exit rate. Gather the data, use it as an input and get a closer view of problem identifiers.

Importance Pages

Importance pages are the pages with most traffic. They need to optimize in order to relish the streaming traffic converting to GOAL. Here you can run experiments. An experiment here is rewarding as your test gets completed in a short span giving you a boost to work more. These pages are ideal to put your message, your keypoints. Make sure when you are analyzing traffic swarming pages, you are counting unique pageviews in order to avoid duplication.
Your top landing pages are definitely your important pages. Optimize them so that they look appealing, eye catching and engaging.

Ease Pages

Let us define it minutely. When you are able to get the same conversion rate from two events, the one consuming lesser time and implementation efforts are called as ease pages. If you apply the same kind of efforts only on these ease pages you are indirectly doubling your conversions.
Once you are done accumulating the information about potential, importance and ease pages, you may club them into a pie opportunity table.

Create Tables

Fetching the statistics from all three pages, you can create a table to list the priority. Let us assume that you are an ecommerce company. The home page, category page, product, cart, wish list, etc. are in your focus to be optimized. The following data is taken randomly to make you understand the priority.
Page Type Potential Importance Ease Priority
Home Page 10 10 8 9.3
Category page 8 10 9 9.0
Product page 8 10 9 9.0
Cart Page 10 9 7 8.7
Wishlist page 8 9 6 7.7
Blog temp Page 10 10 4 8
Product Payment page 8 10 5 7.7
From the above table you can easily find what is the priority. Keep in mind that there are no standard rules to prioritize. They are unique for every business.

Test Hypothesis

In this section, we will discuss the definition of what Test Hypothesis in CRO means. The dictionary meaning of hypothesis is “A tentative assumption made in order to draw out and test its logical or empirical consequences.” It is an assumption on which your test variants are based on. Hypothesis is the question you ask your visitors while testing. It defines why the problem occurs.
Once you have collected enough data about what is hindering your conversions, you have your ‘negatives’ in hand. Now you have to flip these cons into pros. Question arises, How???? Start with isolating what to keep and what to discard.
Once you are done collecting the insights for low conversion rate, you aim to conclude attributes to be changed in order to improve conversions. To do this, first list the following three things −
  • What you want to alter?
  • What do you intend to change it into?
  • What outcome you will get?
Say, changing a to b will get you c.

Creating a Powerful Hypothesis

Creating a testable and good hypothesis is crucial in conversion rate optimization. Else, you will be testing Random data without a focus, that will leave you trying in vain.
Once you have the hypothesis, you need to test it. This will let you know, whether it gives you the desired conversion or not.
To get great results, you should consider the following four crucial factors.
These are −
  • Value Propositions
  • Relevancy
  • Clarity
  • Urgency
Apart from the above factors, Hindrance, disturbance, distraction and anxiety reverses the rate of optimization.
Apart from the above factors, Hindrance, disturbance, distraction and anxiety reverses the rate of optimization.
Is your Hypothesis test correct? To know it, please try and check the following pointers −
  • Have you clearly identified the questions you would like to test?
  • The hypothesis is a proposed statement, clear and testable.
  • Your value propositions are clear to you.
  • Do research on similar hypothesis starting to clear your doubts.
The test-hypothesis is centered around the following three things −
  • Presumed problem
  • Proposed solution
  • Expected result
Once you are done writing hypothesis, test it and get answers for your questions. Derive a conclusion. This conclusion takes you to experiment designing. See the results, if it comes out positive, you have achieved what you were anticipating. Keep repeating for better results. If the result is negative research again.

Test Setting

In this section, we will discuss about the settings which are needed for testing in CRO.

Know your Testing Technology

Any website testing is not a complex process. It is simply creating different versions of your webpages. Show them to your visitors and track which one gives you better results. Presenting your audience with the desired page gives you massive gains. To get your test running, you need a testing tool that will drop cookies on your visitors machine. These cookies let you trace what percentage of visitors take action and on which variant of your webpage.

Test Types

There are different types of tests you can run. All are suited for a specific purpose. Though you can run multiple tests on one problem, duplication is more likely. Here we’ll be discussing four major tests.
  • A/B Testing − It is commonly used with two or more variants. We compare two variants to see which one is performing better. Here, two webpages are shown to same visitors and a trace is to be measured about which one gives more conversion. A/B Testing when done consistently can make a huge difference in conversion rates.
    All you need is to do controlled tests and gather empirical data. Almost anything can be tested using A/B Testing methods, headlines, testimonials, actionables, call-to-action buttons, images, links, etc. It needs patience and time, yet it yields fruitful results.
  • MVT Testing (Multivariate Testing) − It is a technique to draw the best out of combinations of ‘changeables’ to determine which one performs better. Let us clarify it further in detail.
    When you are performing an MVT test, you test combinations of elements inside one webpage to see how they interact with one another and which combination is best suited for a better user experience. Whereas in A/B Testing you test them on different webpages.
  • Split Path Testing − It splits the traffic into different linear paths containing multiple webpages. Instead of showing variations within or out of a single webpage, it shows different paths.
    For example, if you are an ecommerce company and you are optimizing conversions using this method you may follow a six step check-out process or four steps or just three.
    Using this split-path testing, you test which among the three gives a desired conversion result. It can be considered while testing checkout paths, multi-page forms, and product recommendation tab. Split-path testing may not be an easy job because you are testing paths consisting of multiple pages. The threat that occurs is people do not tend to surf in a linear way. The site navigation is not always linear. This may hamper your optimization results.
    Besides these flaws, it is an innovative test that raises an opportunity for higher lift in conversion numbers.
  • Site-wide Testing − This is a special testing that proves to be useful while testing the design and the layout elements throughout the website. It is usually done to test the product description page. You may test by swapping columns left and right. You may change the placement of image, where the product description would stay, etc.
    You need to alter all the wizards to be changed throughout the website. Site-wide test is best when you want to try a different order of links in your navigation or new product detail page layout throughout the site. The results would definitely be immense if done systematically and you will be able to maintain the consistency of visitors on your website.

Testing Tools

The market is flooded with various testing tools. You need to pick the best ones which are suitable for your business. Keep in mind the technology and features of those tools. The cost is to be decided by you completely and of course it should be flexible enough to adapt to different testable variants.
The following tools and more if used with right strategy, will lead you far beyond conversion goals and maximize ROI (Return On Investment).

A/B Split Testing

A/B Testing lets you test different variants of your webpage. You can determine which is better when it comes to user experience. It sounds simple but it certainly drives you beyond conversion goals and maximizes the ROI.
Some of these A/B Testing tools are as follows −
  • Unbounce
  • VWO
  • Five second test
  • Google Analytics experiments
  • Convert experiments
  • Maxymiser
  • KISSmetrics
  • A/Bingo
  • Adobe target
  • AB Tasty

Analytics

It is a great tool to provide you Analytics track and reports on day-to-day happening details of your website. Analytics tools like Google Analytics, kissmetrics, Mixpanel, etc. inform you about traffic sources, user behavior, bounce and exit rates, cohort analysis, goal tracking, best performing keywords, segmentation, etc. It lets you track minute things, which were going unnoticed making things easier for you. You can decide what to keep and what to erase.
Though most people feel Google Analytics is sufficient. Yet there are plenty to save your business. Some of these are −
  • Google Analytics
  • Google website optimizer
  • Openweb analytics
  • Pwik
  • Site meter
  • Stat counter
  • KISSMetrics
  • Mixpanel
  • Segment.io
  • Chartbeat
  • Clicky
  • RJ Metrics
  • Woopra

Surveys

Analytics masters giving insights but users can themselves help you with their requirements. Surveys are best tools to know a user’s need and their experience. Feedback is immensely important to work for the betterment of your product.
Though open-ended questions are the basic means of carrying user surveys, you can also try your luck by A/B Split Testing that can be done by running page level survey on the main and test pages.
You can conduct surveys by using any one of the following tools −
  • Qualaroo
  • Survey Monkey
  • SurveyGizmo
  • PollDaddy
  • Survey.io

User Testing

There are times when users cannot evaluate their problem. User testing softwares are the answer to this problem. They educate you how the user interacts with your website.
You need to figure-out what is appealing to users, where they click, what they look at?
For this HeatMapping and trackClick Density performs very well. They help you in articulating hypothesis for testing. Apart from these two, you can also use the following tools for testing
  • CrazyEgg
  • Click Tale
  • Cage
  • Loop11

Concept Testing Tools

These tools allow you to wireframe your website. You can create diagrams, flowchart, sitemap, etc. for a clear picture of the website and to find out the trouble spots.
  • Balsamiq
  • Cacoo
  • Browsershots

No comments:

Post a Comment