Testing Right, Leads to the Desired Conversion Optimization - Review
Hello, LinkedIn Community,
Its been an another great week of learning @ CXL institute’s growth marketing mini degree. This week, I started with the second track of the mini degree focusing on growth experiments.
The workshop on Research and Testing by Peep Laja focused on conversion rate optimization (CRO), which is a systematic method of identifying potential opportunities for growth and developing data-driven strategies to capitalize on those opportunities. The main aim was to identify the problems present on websites and use a hypothesis-based approach to fix them. The only way to achieve success for a business is to test effective changes and figure out what matters to users. Research plays a central role in identifying how our customers think and how these issues can be resolved. By fixing these issues, a business can boost its conversions. Every optimization process starts with framing the right questions. What are they buying? How are they shopping? It’s important to challenge assumptions and biases that may hinder this process.
Research and Testing?
What does a good optimization process look like and what does it do? A good optimization process should tell you where the problems are, what exactly are these problems and why it is a problem. With this information, you turn issues into test hypotheses and then prioritize tests and instant fixes.?
So, to initiate, how do we find out what the problems are??
1.???Research: User research and conversion research, must be the foundation for everything.?
2.???Ask the right questions in your research.
3.???Your data must be set up to answer these questions.
4.???Be aware of your biases
There are 3 things we should measure to make sure our optimization is being done right:?
1.???Test more effective changes: to our website, landing pages, wherever the traffic is landing on.?
2.???Reduce the duration of optimization: we need to generate good test ideas fast to reduce cost.
3. ?Improve the speed of experimentation: the more tests we run, the faster we reap the benefits.
A way of doing your research right and gathering the right kind of data for the optimization process is using the “ResearchXL Process” which is a framework created by Peep Laja for?identifying??problems or issues your website has and turning those into hypotheses that you can test.
This process helps you gather 6 types of data to help you make optimization decisions and build effective winning tests.?
In the course on?Conversion Research?Peep focused on the six steps of the Research XL model, essential in carrying out effective conversion rate optimization. These include:
1.????Heuristic analysis
·?????An experienced based test of a website page by page on desktop, cellphone
·?????Focuses on Clarity, Friction, Anxiety and Distraction
2.????Technical analysis
·?????Should work on the device and browser the costumer is on. Check for conversion comparison in browsers and bugs
·?????Site speed on different browsers and devices.
3.????Digital Analytics
·?????Where is the problem?
·?????What type of users are dropping?
·?????What behaviors correlate with more purchases or retention?
·?????Measure everything that needs to be measured
·?????Use google analytics to measure site data
4.????Qualitative Research
·?????Two types of surveys: on-site polls and email surveys post purchase
·?????Come up with polls on key pages to see in analytics where there is a huge drop off
·?????Ask what is hindering users from purchasing a product on the site or the doubts they have?
·?????In email surveys ask your successful costumers 8-10 open ended questions about their shopping experience
5.????User testing
·?????Recruit people from the target audience if possible
·?????Ask them to perform complete tasks, like making a purchase
领英推荐
·?????Monitor whether they are able to complete the task
6.????Mouse tracking
·?????Use mouse tracking, scroll maps and heat maps to monitor what actions users perform n the website
·?????Record and view videos of what people are doing on the website to assess user behavior
After this process, it is important to identify issues that are costing the business most in terms of revenue and conversion and fix those issues on the website. Brainstorming, evaluating hypothesis and evaluating solutions is key at this stage. Ultimately solving issues comes down to how well you know your costumer and are aware of their behavioral pattern. These insights can be gained from numerous qualitative and quantitative data sources available to businesses. However, experiential learning such as site-walk throughs may help gain valuable insights as to what elements are distractions to users and deter conversions. Resolving issues on the site involve experimentation and drawing insights from data gained through usability testing and costumer surveys.
Steps of Conversion Optimization
1.???Conduct Research
2.???Build Hypotheses
3.???Create Treatment
4.???Test Treatment
5.???Analyze Results?
6. Follow-Up Experiments
?Every problem you find, categorize it into 4 types of issues:
1.???Instrumentation issues (something is not being measured or is incorrect).
2.???Just Do It issues (things you can just fix, you don’t need to test it, for example, font size too small, or faded color).
3.???Test/Hypothesize (even if we know what the problem is, we don’t know what the optimal solution is, that’s why we need to test).
4.???Investigate (Maybe you don’t know if the problem you uncovered is really a problem).
Testing the Right Problem is the Right Solution.
?
When is the test over? In order to declare a test done you need 3 things:
1.???Enough sample sizes (there are calculators for this).
2.???Multiple business cycles (always 7-day cycles).
3.???Statistical significance (only after analyzing sample size and multiple business cycles).
Metrics of a testing program
1.???The number of variants tested. If you have the traffic, it's much better to run an ABCD test than it is to run an A/B test. The faster we test, the more we test, the more we learn.
2.???Win rate (Percentage of tests that provide a win). You always want to improve this metric.
3.???Avg. uplift per successful experiment.?
?A/B testing
Now after all the research and data you gathered, you can start thinking about building the actual A/B testing. A/B testing is splitting the traffic of a website between a control and a variation. Here’s an example of a simple A/B test.
If you have enough traffic, you can also create tests with more variations, which are called A/B/n tests. But be reminded, for this kind of multivariate test you need a lot of traffic. A/B testing is more popular because besides being simpler to run, you are also testing more significant changes, with potentially more impact. Ton Wesseling doesn’t recommend doing more variations because they have a lot of implications and a higher chance of false positives. Start with the basics!
So, for A/B testing you need a hypothesis: a hypothesis defines why you believe a problem occurs. A hypothesis is very important because it gets everyone in the team aligned to achieve the same goals. To form a hypothesis, you need a problem, a proposed solution and a predicted outcome.?
Now, when designing you’re A/B test, you should consider:
To sum it up, some think that Conversion Rate Optimization is about testing the color and size of CTA buttons. But it’s much more than this. When you consider how much time we all spend online and how high our expectations are of the online experience, it's obvious that CRO is actually much more far-reaching.?Overall, conversion is really about enhancing relevancy and reducing the friction – being able to effectively match what customers want and expect.?
Stay tuned folks for next week!