The Interest in Testing Is Surging – Here’s Where You’ll Benefit Most
There’s a groundswell of interest in A/B testing, or “split” testing, when it comes to digital marketing efforts. It indicates a growing interest in new data – more segmentation and improved targeting allow digital marketers to better evaluate the indicators that lead to an organization’s success. The desire to uncover and analyze that data for its secrets is strong and getting stronger.
Chasing improvements isn’t the only thing that seems to be driving the desire to test. The surge in testing interest seems to be tied to the overwhelming amount of available data. For busy marketers or CEOs, it can lead to concerns over what they might be missing. Are ads being trounced by the competition? Are landing page barriers leaking customers? With many simultaneous digital efforts in play, it’s easy to be haunted by unknown unknowns. With testing underway, business owners can take comfort in knowing unknowns will rise to the surface.
At MCP, another concern drives our own testing: reams of data are often the result of averages. What is going on within the segments of those averages? Could a KPI be dragged down by a buried segment that would make it otherwise more attractive – a segment where directing funds would lead to value? Isolating and testing such a segment can make the difference between profitability and loss. And, when a 1% increase in conversion on your site can result in a significant increase in ROI, it’s easy to see why ongoing testing is a priority.
Should you test? Absolutely. Whether it’s A/B, split or multivariate, here’s a brief look at areas where testing can lead to crucial improvements.
At MCP, we regularly hear our clients say they want to know what ads perform best. They know regular ad testing can mean a dynamic campaign that is always improving. In Adwords, ads are optimized by default to serve the best performing ad. For more controlled testing, that setting can be changed to “Rotate Indefinitely” so ads are served evenly. Then, by changing a single element of an ad, such as the headline or a different call to action, your A/B testing has begun. Does stating prices work better than not stating a price? Do promotions drive more clicks than free demos? Check your KPIs. In most cases, it’s the click-through rate of your ad that will tell you if it’s grabbing users.
When to make decisions about your findings will depend on activity. If you have high click numbers, you’ll be able to make that decision in days. For others, it may take several weeks before data reaches a threshold of statistical confidence so you know what ads is performing better.
Alternatively, using Experiments in Adwords can be helpful to understand when results are statistically relevant so you’ll have more confidence in the changes you’re making. Colored arrows indicate whether confidence in the results is low, medium or high. Experiments involves creating a duplicate, or draft, campaign and making changes to that campaign while running both simultaneously. And, if you have less than profound confidence in your test campaign, Experiments allows you to run your test ads less so as not to disrupt your conversion rates. When the results are in, you can easily implement the changes into the original campaign.
While your first thought may be to test ads, don’t put off making improvements to your landing pages just because it takes more time. While click-through rates are important for ads, once users are on your site, you can lose them all over again. Leaking users because of a barrier on the landing page is one of those “unknown unknowns” you can make known by watching bounce rates, retention, or conversion rates. Testing for these metrics will drive improvements and plug those leaky holes.
Elements to test on a landing page may be a headline, a call to action, graphic and copy elements, or different of promotions. (For Adwords testing, we recommend using isolated pages not navigable from your site.) Plenty of third party companies offer out of the box A/B testing, but as a business owner in charge of your digital marketing, you can also test yourself using Content Experiments in Analytics. Content Experiments makes site testing easier for paid traffic and any other kind of traffic to your site. You’ll have to add a little code to your website. After that, it’s as easy as inputting A and B URLs (you can test up to 10 different pages) and launching the experiment. Analytics will handle serving the URLs to users who will see the variations of your page. Here’s some information about Content Experiments.
Soon to be the gold standard for DIY testing is Google Optimizer, currently in beta but available to most users. Those familiar with Google Tag Manager will recognize the interface, which uses “containers” to create experiments. Creating changes to headlines, call to action buttons, and a variety of layout options is at your fingertips with Google Optimizer. By setting the variant, stating your objectives, and inputting your hypothesis, you’ll be able to easily read the results, implement changes, and move on to the next experiment.
Testing can and should be a conversion optimization tool that provides data-driven insights to what leads to better performance and more return on all your digital marketing investments. Email marketing for example, can be tested by using different subject lines to compare open rates. Display ads provide many ways to test creative, so you can determine how changes like colors and buttons effect traffic. Dig into the “averages” mentioned above and test by breaking Google Search Partners away from Search, for instance, or evaluating positions, sources, geography and demographics, either with experiments, or simply by digging deeper into data to slice up your averages into more productive segments.
At MCP, we often find ourselves discussing marketing decisions with clients and wondering what move is best. Our answer is usually, “Let’s test it.”
Questions about testing your digital marketing efforts? Just ask.
Find out how the Obama campaign used testing to fund-raise in 2010.