Website testing and optimization is only as good as your senior executives think it is.
So convince them. Show them dollar values, keep records of your results and learn from your failed tests.
Keith Swiderski, head of e-commerce digital marketing for Avis Budget Group, offered some of these tips in the CMSWire-HP Optimost webinar, “5 Ways to Get Upper Management Excited About Web Testing.”
Test and Learn Culture
In the more than 18 months he’s worked in marketing with the car rental organization, Swiderski said his teams have created a test and learn culture. “It’s at the center of what we do,” he said.
So much so that senior executives themselves, when pitched ideas, push for iterative changes with testing at its core. “That,” Swiderski said, “is when you know you’re successful.”
It wasn’t always this way. About 10 years ago, art ruled when considering website changes. Organizations picked agencies with creative people, had them draw up storyboards like in the Mad Men days and did complete website overhauls.
Often, developers stifled marketers’ ideas because of technology limitations. “Every few years, you’d do a redesign,” Swiderski said.
About five years ago, the culture of A/B testing came into play more often. Companies went with smaller changes and got faster consumer feedback through testing surveys. Now, enterprises live within micro-environments led by science and statistics. Enterprises shoot for personalized websites, a platform agnostic user experience.
“Marketing leads now, with developer support,” Swiderski said.
Achieve KPIs Through Testing
The first step in a web-testing journey is to know your key performance indicators (KPIs) — the reason you get up every day and come to work.
“If you don’t know, then set them up,” Swiderski said. “Work with senior management to tell you what their KPIs are, and you can distill that down from there.”
Swiderski’s ultimate KPI is converting website visitors into car-rental purchasers. Those that give a name and email are not automatic conversions, however. Avis has put a lot of testing into its “pay now” and “save money” actions to ensure more conversions.
“The ‘pay now’ option became a KPI for my team,” Swiderski said. “We knew that was a driver of revenue. We spent a few months testing on all platforms making the ‘pay now’ call to action clearer and we deemphasized ‘pay later.’ And we made the process of checking out more attractive. We were able to double our ‘pay now’ customers.”
Other KPIs can include:
- Customer satisfaction, tied to things like Net Promoter Score
- Shares, likes, email signups, if you can tie that to revenue
- Omnichannel impact
Use Human Speak
Swiderski encourages marketers to pitch testing to executives in languages they speak. Most of them speak the money language, so that’s a good place to start.
“We were able to quantify the actual value of the customers switching from ‘pay later’ to ‘pay now,’” Swiderski said. “We showed the increased rate of customers showing up and tied it to a dollar amount. We annualized the fruits of our efforts in a real dollar amount. When you sit down with someone whose job it is to make money for the company at a higher level that’s important. They can understand what marketing does."
There are rights and wrong ways to bring in revenue examples to the testing equation:
Wrong: “This test achieved a 9 percent boost in adds to cart.”
Right: “Every 1 percent boost in adds to cart equals $10 million in annual revenue. This test annualized could bring in $90 million in revenue.”
Keep it Simple
Swiderski suggested that oftentimes the small tests are the most successful. Multivariate testing may take a while, but it can be easy, especially on a smartphone where you’ve got only three to four inches of real estate.
“Simple changes here can have a big effect,” said Swiderski, whose team uses the HP Optimost platform to power its web testing, targeting and personalization.
For instance, Avis began to see more call-center calls coming in. It assumed that was a bad thing because usually those who call had a bad customer experience online. But upon further testing, Avis officials realized they had placed the phone number more prominently on its website.
“What if we removed that number from there?” they thought as Swiderski recalled. “In three days we were able to increase conversions on our website as a result.”
Love Being Wrong
Failing is OK in testing. In fact, it’s good data for the next round of testing, Swiderski said.
“Science is most interesting when you’re proven wrong,” Swiderski said. “You’ll have information and data that something is wrong.”
For instance, Swiderski’s team assumed that red would be a bad “call to action” color for Avis. But it ended up working and proved to provide “excellent learning for us even though our hypothesis was wrong.”
Know Your Costs
Create a profit-and-loss function for all multivariate testing. Run it like a profit center and not a line item.
Know how much your testing has increased your revenue. Keep a log of all tests. Be conservative when making projections for revenue increases. Cut them in half — it will give you credibility later.
Swiderski offered some other testing tips:
- Always have a backlog of ideas
- Find a partner that understands your business
- Don’t forget about qualitative testing. “Don’t forget about qualitative testing. I like user testing.com,” Swiderski said. “Gets us tests results in an hour. We can get video of people giving us live feedback.”
- Don’t assume a winning test will just work everywhere. “Select my car” worked well in the US but “failed miserably” in Canada, where “view rates” worked better.
- Know where your users are coming from, and pay attention to how different segments interact
The article is a recap of a webinar sponsored by HP Optimost.
Title Image by Austin Schmid