Now that many of you’ve had a chance to digest Google’s recent Optimize sunset announcement, we are still hearing some concerns about how you can continue testing and what choices are available. Let’s take some time and look at arguably the hardest part of testing every company will face: long-term strategy.
For most of us, the first three to six months of testing are easy. We have plenty of testing ideas or hypotheses, we’re getting big results in our tests and maybe some people outside of your testing circle are interested in your work. However, most testing departments we’ve worked with inevitably hit a big wall at that three to six-month mark where testing often slows down considerably. Common causes for this are usually struggles with generating new testing ideas, issues with buy-in and implementing test results, and not planning far enough ahead. I’ve personally encountered these at some point in my career and am happy to share some proven tactics that can help you break through that wall or, ideally, avoid it altogether.
Generating New Testing Ideas
For most of us, the first tests we want to run are bottom-funnel focused; CTA buttons, lead forms and PDP pages are usually the first elements/pages we start testing. On those elements/pages, we’re usually testing colors, text, images, etc. These are all great places to start and can quickly lead to major improvements to the main KPIs of your website or app. But what comes next? What happens when you run through these initial low-funnel tests and start to run out of images, colors, and different text to test on these elements/pages? You hit that brick wall.
In a more mature testing environment, you don’t just test bottom-funnel KPIs; you test the entire funnel to drive more people to those bottom-funnel KPIs. This isn’t meant to be a KPI discussion but know that the entire funnel will have KPIs while each step in the funnel will have its own KPIs (I’ll refer to them as soft KPIs.) Some of these soft KPIs may be time on site/page, bounce/engagement rate, through rate for funnels, and even scroll depth.
So let’s say you are looking to test against these soft KPIs but are still struggling to think of ideas or hypotheses to test off of. Outside of looking at your site/app data with a platform like Google Analytics, here are some other places to look for potential tests. Remember: there are no bad ideas when it comes to testing.
Examples Of Where To Find Testing Ideas
- From Successful/Failed Tests Already Run
- From Your Competitor’s Websites or Apps
- From Market Research (yours or 3rd party)
- From Last Year’s Tests
- From Internal Channel Owners/Leaders
Use-Case Scenario: Using Tests Already-Run to Fuel Future Tests
Our Prior Test: We had run a simple homepage promo banner A/B/n test with the image design being the changed element. Clicks on this banner were the main objective of the test. The test ran and Variation 2 was the winner.
Observations: Pretty straightforward so far but as a practice, we would always look at the testing data within our analytics platform (GA360 in this case) to gather more info on the users who preferred the winning variation. Here, we found that most of the age demographics preferred the winning Variation but visitors over 55 preferred Variation 1.
Action: We set up the personalization tool to show visitors under 55 the Variation 2 banner and anyone 55+ saw the Variation 1 banner.
Results: We increased the conversion rate for that banner even higher than what was recorded in the test and went back to some of our old tests to generate new testing ideas for this age demographic.
Buy-In and Implementing Test Results
Every organization is different, but it’s not uncommon to have issues with buy-in and implementing the results of your test. Testing, in its very nature, strives to know the best experience for the end user (your customers) instead of assuming it. Sometimes these truths go against the perceived truth and can rustle some feathers. Call it growing pains. This can manifest in many ways, but at the end of the day, your testing department will be the one who suffers if not approached carefully.
Hopefully, you’ll never encounter these issues but here are some successful strategies I’ve implemented to work through these issues and come out the other side where everyone wants to be involved with your testing.
Involve People From Outside Your Department
In the last section, I mentioned you can find testing ideas from internal channel owners/leaders and that plays right into this strategy. You want to involve as many outside minds as makes sense when formulating your testing strategy for the year. Inviting internal owners/leaders from Media, Social, SEO/Search, creative, and UX are where I always start, but there have been times when it helped to bring in members from the strategy team, development team, or leadership went a long way to buy-in and getting new and creative testing ideas I would have never thought of.
Distribute Test Results In A Presentation
It’s always a good idea to share your testing results at regular intervals with the larger team, but where I see some companies fail at this, is not going far enough in reporting on those results. Instead of just sending a few lines of text in an email to your internal team with the results, dig a little deeper into the data (like the use case mentioned above) and put together a formal presentation to distribute. You’ll be surprised with the reception of a thorough deck vs. a small email with face-value results. Once again, every organization is different, but I’ve done monthly reports to the internal channel owners/leaders and larger quarterly presentations to our executive team. No matter how you approach this, take it that extra mile!
Run Outlandish Tests For Others
I can’t count the number of outlandish tests I’ve run for others that I knew, just knew, it wouldn’t win. Why did I do this? Two reasons: 1. I truly didn’t “know know” that the test wouldn’t work (remember, we’re building a knowledge base with testing), and 2. I wanted to be inclusive of their ideas so they would trust the tests, and results, from us which builds buy-in. Of course, you want to prioritize tests with high impact to the KPIs (more on that in the next section), but you might be surprised by the results from these ideas that seem outlandish.
Planning For The Year
Last but not least, you should be looking to plan a year out. 6 months out is not uncommon but I always aimed for a year out. You might wonder how you plan a full year of tests when the tests you’re running now only last a week and have issues always running some sort of tests. Here’s how we generally tackled the planning each year
Keep A Testing Roadmap With Results
Arguably the most important tool of testing, having a roadmap with all of your test ideas, hypothesis, priority, predicted results, description of where, when the test will run (and to who), and results will ensure you always know what test your running, when you’re running it, and what the priority of that test is. Most of the tools in the market today will keep all of your ended tests for historical purposes, but I always preferred to rely on our own record-keeping (just in case.)
Leave Room For New Ideas
In your testing roadmap, we leave plenty of room for new testing ideas that may come up from your brainstorming sessions.
Bucket Your Tests Based On KPIs
I mentioned prioritizing your tests in the roadmap but how do you prioritize and organize your testing ideas for the next six to twelve months? Bucket your tests each month/quarter into KPIs (main, secondary, etc.) By planning each month/quarter to run a certain amount of tests against the main KPI, another round of testing for soft KPIs and even more tests for ideas from your brainstorming session, you’ll be surprised by how quickly those buckets start filling up each month. And, if you’re using a good roadmap with outcome prediction formulas, you can see the runtime of each test and stagger as needed, so you always have some sort of test running. If not always, then most of the time. One last note, this can be tough for those just getting started testing so start using this strategy for a quarter at a time and work your way up to that six to twelve-month plan.
While this approach is not one-size-fits-all, these points can help you avoid issues typically found with testing and make you a center of truth for the company. Happy testing!
If you have any questions about further testing strategies, please contact us. We’re happy to help!