Across several presentations at Google Conversions in Dublin, several speakers shared insights and best practices for conversion rate optimization. Here's a few highlights:
Confirmation Bias - Michael Aagaard
- In the 18th century, tobacco smoke was considered very good for your heart and lungs. In particular tobacco enemas were quite popular so much that they were placed along the banks of the river Thames to help drowning victims. This is an example of confirmation bias at work.
- Confirmation biases is our tendency to accept evidence we agree with at face value and dismiss information we don't agree with unless the evidence is overwhelming. Confirmation biases limits our ability to seek out and uncover the truth.
- Torturing data: if you torture any data long enough it will confess to anything. High levels of correlation between things don't imply causation. We have to be careful to not see what we want in data.
- Stopping A/B tests when they show the impact we want is an example of confirmation bias. Instead, let them run for an appropriate amount of time. Over time, tests are likely to show much less effects.
- How to overcome confirmation bias: accept the fact that you could be wrong, seek out a different perspective. Find people who talk to customers/users. They have a bias toward end users.
- Don't test your ideas, do detective work to find out what customers need and how they talk about it. Then your A/B test is simply the final test at the end to see if you did your detective wrok well.
CRO - Lina Hansson
- Celebrate the discovery of weak spots. Don't take it as failure but instead be happy when you find something that can be improved.
- The biggest missed opportunity in conversion rate optimization is usability testing. Move away from opinions and instead use user testing to identify issues.
- A common pain point across retail sites is find-ability: both search and browse. When we move to mobile, many sites remove their top categories list in order to fit on smaller screens. This creates discoverability issues. One of the first things retail sites should test is adding categories visibly on their home page.
- Value propositions for companies are usually cut for mobile. Instead of removing them, redesign them to make them work on mobile.
- People can be classified into four behavior types. Methodical people read completely and analyze before making decisions. Humanistic people react strongly to the opinions of others. Competitive people move quickly and expect things to work. Spontaneous people are emotional and fast-paced. You can design experiences that are appropriate for each of these behavior types.
- The companies that solve checkout on mobile are the ones that will win.
Meaningful Data - Simo Ahava
- It's quite simple to get a service like Google Analytics set up but how do we use these tools to really understand what we're doing. How can data become meaningful?
- Tactics (tool expertise) without a strategy (business expertise) are just party tricks and a strategy without tactics is just talk. What brings the two together is agility.
- Tools must be customized for your organization's needs. We are not trying to optimize metrics but our businesses. Default metrics and reports need to be adjusted to work with your specific needs.
Landing Pages - Anna Potanin
- Designers want to do their best and create unique interfaces but making things for the Web often requires understanding and using conventions. Only apply a unique visual design after you have followed best practices.
- 3 things all retail sites should have on their landing and home pages: call to action, value propositions, and visuals.
- The more prominent you make your search bar, the more searches you get. Why do you want to do this? Conversion rates are usually much higher for people who search