Everyone knows they should be A/B testing.
Except most actually shouldn’t.
Sure, making a few tiny changes that result in big conversion lifts sound nice. But those flash in the pan results typically stem from countless other tests that didn’t net a single improvement.
OR, the split tests you’re running are inherently flawed and the results will mean next to nothing (statistically speaking).
Here’s why A/B testing might not work for you, and 3 other conversion increasing tips to try instead.
Why You Shouldn’t Be A/B Testing
When people think of conversion optimization, they immediately jump to split testing.
Sure, landing page optimization techniques like this can be hugely beneficial. But only for certain companies, at certain sizes, with certain types of pages.
For the rest, running endless split tests of one tiny variable against another might end up as a big waste of precious resources.
Here are a few reasons why.
Problem #1. Inherent Bias
Conversion testing (especially split testing) can often contain internal biases that are hard to catch.
For example, if one test works with your traffic from Facebook then it’s easy to (incorrectly) assume it’s also going to work with your email subscribers. That over-generalization fails to take into account that your Facebook traffic lacks need awareness, while your email subscribers are already well versed in who you are and what you do.
Another is over-optimization. Focusing too much on the tiny details, without seeing the bigger picture, might lead you to make decisions that make one vanity metric go up (like Total Leads), while other more important ones come down as a result (like Sales Qualified Leads).
The most infamous bias of correlation vs. causation says that just because a green button out-converted an orange one, doesn’t mean that all buttons should now be changed to green.
These biases seem logical at the time, and it’s easy to fall victim to their siren song without the help of many well-versed experts involved in the process.
Problem #2. Opportunity Cost
Proper split testing should be carried out in a methodical, consistent process to systematically work your way through proving (or disproving) specific hypotheses.
It shouldn’t be done on a whim, because some blog post your boss read said that females respond to the color blue better.
Conversion testing is a time consuming, often fruitless task where only a tiny percentage of your tests pay off (which they do, handsomely). But otherwise, you’re running through a continuous cycle of:
- Examining Your Objectives: What would you like to fix or improve, and how are you going to measure everything?
- Determining Your Baseline: How are you going to define success based on where you’re at now (qualitatively or quantitatively).
- Creating Hypotheses: How (specifically) are you going to prove those hypotheses?
- Running Tests: Figuring out your timeline and what an appropriate sample size might look like
- Conducting Post-Mortum: What worked, what didn’t, and why? Based on this, let’s start again.
Simply put: this takes time.
It’s laborious, and many times you should be spending those hard-earned dollars elsewhere.
Problem #3. Statistical Significance
The most common A/B testing problem? Not enough peeps.
Conversion expert Peep (see what I did there?) recommends at least 1000 monthly transactions to make A/B testing worthwhile. And every individual test you run should have a minimum of 250 conversions before you put faith in the results.
Big eCommerce brand? No problem.
But even large B2B companies might have trouble with some of those numbers, where their volume tends to be much lower and transaction size per conversion much larger.
Depending on the size of the lift you’re looking at, those numbers might vary. For example, here are some basic rules of thumb to start with:
If you want to get even more granular, you can also try Visual Website Optimizer’s awesome Test Duration Calculator:
All of this means your split or A/B tests might not even result in statistically accurate results (which kinda defeats the whole data-driven, conversion optimization thing).
The good news, is that you can still improve conversions without running a single A/B test.
Here are three conversion increasing tips to try.
Tip #1. Customer Research
Continually learning more about your customers should be marketing priority #1.
Don’t take my word for it, but Conversion Rate Experts, who helped Moz rake in an extra $1 million.
Specifically, you should constantly be trying to uncover:
- What sold current customers?
- What potential pitfalls might be holding trial-ers back from signing up?
- Why caused ex-customers to become ex-customers?
The goal here is to identify underlying problems or motivations and re-position your offerings accordingly in light of this new evidence. Many times, you might be suffering from a customer development problem instead of a conversion one.
Customer personas can then be constructed, and evolve as you continually unearth new findings. This seminal piece from Mike King should be required reading.
In an ideal world, you could line up in-depth interviews with each of these different cohorts to go deep into context.
But if that’s not practical (or timely), start with simple surveys. Your own email database, social followings, or existing website traffic are ripe if the incentive is enticing enough.
Beyond your own audiences of people, you can also use tools Google Consumer Surveys to identify and target people who might fall in your ideal customer demographic.
You could also use the Survey Monkey Audience tool. One interesting idea might be to test the answers or results from your own audience against similar groups of people (who lack the brand awareness) to see how and where their answers differ.
Surveys and other forms of qualitative feedback are a great place to start. But more quantitative methods that highlight actual user behavior should then be conducted as well.
For example, affordable tools like Crazy Egg can indicate where you might be losing visitors on a page. (While other ones like ClickTale help you do this on steroids.)
Chances are, you already know all of this stuff. And you’re aware of all the tools. Now it’s time to implement.
Tip #2. Improve Popular Visitor Funnels
There are certain ‘visitor paths’ that people take to convert on your website (whether these were deliberately set-up or not).
The goal, is to analyze this funnel in its entirety before breaking it down into pieces.
For example, if you’re a B2B company with consultative sales, yours probably looks something like the following:
Basically, you’re laying out all of the individual steps someone might possibly take prior to converting (from Visitor to Loyal Customer).
You might have constructed this funnel internally, or you can take a look at what actual visitor data is telling. Customer journey’s today involve multiple channels at different times. And it usually takes somewhere between 7-13 ‘touches’ to deliver a sales-ready lead. That means your own visitor path should reflect how and where all of these are occurring (like thank you pages, automated email follow ups, etc.)
Once you have a workable prototype, it’s time to:
- Identify the bottlenecks
- Eliminate the roadblocks
The former can be addressed by zeroing-in on where big drop-offs are occurring.
Many times you can simply compare your own analytics data to see where this might be occurring. You can treat each little step as its own micro-conversion, and look at how conversions on one page line up with email Opens and Clicks on the next step. Or you can setup dedicated Goal funnels or Events in Google Analytics (or any number of more advanced tools).
Another way to uncover these problem areas is through usability testing, like UserTesting, which let you watch recorded visitor sessions. Getting the feedback from a new set of eyes is invaluable in defeating the curse of knowledge you might have from intimately knowing everything about the funnel already.
Eliminating roadblocks for segments of traffic can also help increase conversions with relatively little work.
For example, let’s say you have a large amount of site visitors that are already ‘brand aware’ and coming to your website directly. Sending those people to your TOFU offers is a waste of time (best case) or an incredibly frustrating turn-off (worst case) for them. Instead, brainstorm a few ways you can (a) identify them (b) appeal to them, and (b) get them further down into your MOFU and BOFU offers quicker.
Page redesigns or layout changes can then have a massive impact on success (i.e. short copy for brand aware vs. long copy for non-brand aware) if your creative decisions are well informed.
Tip #3. Demand Geneneration Low Hanging Fruit
More (quality) traffic almost always makes up for low conversion rates.
That’s because there’s often great opportunities for higher leverage that can increase traffic expontentially.
Instead of toiling away at trying 101 ways to bump tiny conversion rates, focus on potential ‘big win’s in customer acquisition and lead generation instead.
Where to start?
You could spend a few minutes looking at Google’s Customer Journey to Online Purchase tool, which will show you where and how individual channels influence your own customer’s journey.
Otherwise, search (both organic and paid) historically drives the most new customers. So that’s as good a place as any to start.
For example, had a new website redesign lately?
You’ve likely got some crawl errors then, like missing meta descriptions which might be impacting the amount of people clicking on your pages from Google Search Engine Result Pages (SERPs). Simple tools like Moz.com or Screaming Frog can help you uncover these quickly.
Another SEO-related technical problem (that just so happens to also affect conversions) is speed.
For example, eCommerce companies might lose half of their traffic if a page fails to load in under 3 seconds. While 75% of mobile traffic might bounce if it doesn’t load in 5 seconds.
Google’s PageSpeed Insights will immediately give you feedback on how your Desktop and Mobile experiences stack up, and how they can improve.
When it comes to paid search, obsessing over your relevancy scores (like Facebook’s Relevance Score or the Google AdWord’s Quality Score) can pay huge dividends because of how that metric can apply tremendous leverage to your results. For example, increasing your Facebook post engagement by just 1% can drop costs by 5%!
Same holds true for AdWords, where you might be sitting on a gold mine if you have highly searched keyphrases with embarrassingly low-Quality Scores.
Spin these keyphrases out into their own single keyword ad groups, create new relevant ads with better matching landing pages and results might skyrocket in a few days.
Conclusion
If you’re like most companies, there’s only so much time, energy, attention, and money to go around.
You need to prioritize. And look for the biggest areas of opportunity to focus your preciously limited resources.
A/B testing might be the answer if you’re doing thousands of transactions each month. But for most, they only promise an illusion of effectiveness by increasing your busyness.
Instead, start by getting to know your customers more intimately. Then see how you can make their lives better by helping them find what they’re looking for (faster, and easier). Last but not least, look for basic mistakes in customer acquisition or lead generation that might net you big gains in quality traffic (that will more than offset a low conversion rate).
Changing your CTA button color might lift conversion rates. But probably not.
And probably not in any statistically significant way that validates all of the hard work it takes to get that color changed in the first place.
Leave a Reply