There are hundreds of articles out there on how your startup can be awesome at web analytics. But if you really are a startup and following the 80-20 rule, which states that 80% of your results will be generated by 20% of your efforts, then you want your efforts to be judicious, specific and yield maximum results. I spoke with our search marketing director Craig Barrett, who named the three most important things startups need to do with their web analytics in order to really start cooking with gas.
1) Pay for user testing
Because SpareFoot relies heavily on a user’s interaction with our own website, we are constantly user testing to find ways we can enhance a customer’s time on the site.
“We run tests with usertesting.com anytime we make big updates,” said Craig. “We’ll run a type of specific problem, like how would a user search our website if they only knew the city they were moving to? Or what if they only knew the zip code or address that they were moving to? We’ll display different ranges to see how far they might engage with different properties.”
Rather than simply running basic user testing, Craig recommended giving specific cases and iterating based on the results.
“We’re also doing this aggressively on mobile. We’ll run tests every couple of weeks, so that we can see what it looks like on multiple devices and if there are any problems. For example, what if our site doesn’t work on Android devices and only on iPhones?”
While user testing won’t reveal anything specific about user behavior, it does provide anecdotal data that can be useful.
“User testing is also a great way of identifying opportunities,” said Craig. “At SpareFoot, for example, we found through user testing that most people wanted to look up zip codes rather than cross streets. We started sorting by zip codes to account for this. It was ultimately an opportunity to enhance the site and the services we offered. When you’re user testing, it’s easy to see how you can make a visitor’s experience better.”
Craig also suggested getting someone outside your startup to write the description before beginning your user tests.
“You can become so close to the product that your description might skew how the visitor would otherwise use the site.”
2) Figure out your bounce rate
“If you’re offering paid media, look at your bounce rate with google analytics and figure out where people are leaving your page most frequently. Then use a lightweight tool like unbounce.com to do a user landing page test to determine if you can modify the look, feel or messaging of your website’s products to decrease your bounce rate. Let that guide your design in the future.”
Craig also suggested using tools like Crazy Egg, which shows heat maps on your website, but not relying exclusively on them.
“Those can be instructive, but also incredibly destructive. What people look at and do with their mouse doesn’t really tie in with your conversion. This can be useful for seeing things like whether or not they can see your reservation button, for example, but don’t use them as a final word. If they hover over your picture, don’t go away thinking that you need a bigger picture. Ultimately, heat maps are useful, but not as a universal problem solver.”
3) Don’t fear the 40-year-old bikers
“Use your analytics to look at your referral sources and subsequent conversion rate, and try to understand which kinds of users are working best,” said Craig. “Then ask yourself, does it align with the user you thought would be the case? If not, how can you maximize the sources or people who are actually coming to your site?”
If you thought that your website or service would attract 18 to 20-year-olds, but actually get a lot of visits from 40-year-old bikers, don’t get tripped up. Look at your media and messaging to see how you can amplify them and better market yourself to those visitors who clearly have a need for your service.