Bras N Things have been Australia’s Bra fitting specialists for over 25 years.
Inflow’s CRO team was asked to optimize BrasNthings.com during the launch of a new responsive site to help identify any post-launch issues.
As this was one of the earlier responsive eCommerce site redesigns done, there weren’t any real best practices to follow.
In addition to being responsive, the redesign was also introducing a new brand experience, as well as new core navigational elements.
Focus on your user
This is the most important piece of conversion optimization – if you don’t provide what your user wants, it doesn’t matter the quality of your product or how nice your site looks. There are a number of tools out there to help you better understand your users, including surveys, screen captures and website analytics. The tools just mentioned helped us optimize the homepage and category pages to help users more easily find their products and prevent “pogo-sticking” between areas of the site.
Remove distractions from your preferred path
This is most commonly applied in checkout flows, but for some sites, there may be other flows, such as custom builders that require similar treatment. You can see there are very few places to exit the checkout at this point.
Prioritize your work based on traffic
Analytics plays an important role in determining where to focus. If you see that your paid search drives a lot of traffic that doesn’t convert, it’s possible you aren’t doing enough to gain their trust. If the majority of your purchases come through email, maybe the focus should be building that list larger and streamlining the flow for those users.
Similarly, the device type is becoming a very important piece of the puzzle. For most retailers, mobile has overtaken desktop in traffic, however, conversion rates and revenue still trail significantly.
With Bras N Things, the second half of 2015 focused largely on improvements we could make to the mobile experience to help users find products, which seemed to be the largest drop off point.
Validate all changes
Regardless of how great an idea may seem, there’s always a chance it doesn’t pan out. Even CRO experts don’t get it right 100 percent of the time – our success rate is somewhere around 80 percent. To help mitigate this risk, we recommend validating all changes before making them permanent. The most reliable way is through split testing, however, there are many other methods you can use. Even if you have done split testing, we recommend reviewing the data again after the release.
This applies not only to changes you’d like to make to your own pages, but also to vendors you might use. Recently with Bras N Things, we determined that one of the checkout vendors was not driving purchases, but resulted in significantly more merchandise returns. Since it was not driving incremental revenue, it was clear that we needed to remove it.
To see for yourself how difficult it can be to figure out the best version, check out a few of the tests on WhichTestWon.com. This further highlights the importance of validating anything you do on your site.
The 90 percent conversion rate increase is significant because it takes into account how much each test actually impacted the bottom line with a conservative estimate. There are two factors at play here:
1) A test on the checkout page likely impacts 100 percent of your eCommerce transactions, but not every user who makes a purchase sees the homepage. As such when determining the impact of each test, we factor in the eCommerce Participation Rate for the test page. If only 50 percent of your traffic sees our homepage test, then the impact to the bottom line is cut in half from what we measured.
2) We always report conservative lift figures—not measured lift. What does this mean? When we test, we’re taking a sample of the total population, which means it may not be 100 percent representative of your users. Based on what’s measured, we determine a range within which we expect the conversion rate to fall—the 95 percent confidence interval—then report the low end of that range. This is a statistically sound method for determining the minimum impact we can expect to have.
Based on the above, we arrive at a 90 percent increase in real conversion rate across the 17 winning tests we ran. This does not factor in the handful of losing tests that, if implemented, would have dragged down the conversion rate rather than improving it.