We’re talking A/B testing.
We give a lot of tips and tricks for maximizing the performance of your marketing efforts.
It’s true, but you shouldn’t just take our word for it and rush into implementing all of our suggestions.
High-performing landing pages are not constructed overnight but rather through multiple rounds of testing. Knowing how and what to test on your page is critical for securing long-term success.
In this episode, we’ll reveal a step-by-step framework to test landing pages safely, identify your brand’s best practices, and achieve your desired results.
First, let’s talk about what A/B testing is, fundamentally.
A/B testing is a scientific method.
You’re performing a defined experiment.
When doing A/B testing you need to 1st decide the action you want to optimize
Example:
- I want more clicks on this button
- I want to sell more of this widget
To do that, then isolate a variable that you think will affect a change.
Example:
- You could say, I’m going to optimize for CTR.
- You decide to change the headline.
At the end of the test, you could decide the headline change wasn’t effective
Your work is not complete.
Now test another lever to affect CTR – like, adjust your CTA.
The key is to separate the desired result from the experiment.
Keys to Successful Testing:
Define your metric for success
Landing pages should have one clear goal.
- Collect emails
- Increase user signups
- or drive sales
Define your key performance indicator (KPI) from the start.
Otherwise, how can you measure the success of your test?
Test both small and big changes
Most people focus on testing only small changes
Example:
Headline copy
Image placement
The color of buttons or CTAIt’s easier (and less risky)
BUT typically leads to negligible improvements.
Sometimes a big drastic change is needed to make an impact.
Never run multiple tests at the same time
Running concurrent tests muddles results.
Hard to figure out which test actually led improvement
Run ONE test before moving on to the next one.
Ok, you’re ready to begin designing your own test.
Measure Landing Page Performance To Uncover Problematic Areas
Tests that provide exceptional results target big issues on your landing page.
In order to show big improvement, it has to be pretty broken to start with.
How do you find big problems?
Start by looking at landing page-specific data using Google Analytics.
Behavior > Landing Page section.
Data points to focus on:
- bounce rate
- time spent on page
- form abandonment rate.
Keep your end goal in mind when deciding what to focus on
Example:
Goal = collect lead’s contact info
Problem = high bounce rate
Look at the things that might be immediately causing a visitor to back out:
- Do you show an opt-in form before giving any value?
- Is your landing page out of sync with the ad or format of where traffic is originating from?
Next, gather data from visitors.
To figure out why your visitors are behaving in a certain way, you need to talk to them Understand their intent & user experience.
Option 1: Add an exit intent survey
Exit surveys “fly-in” when your visitors are about to leave the page.
Ask visitors why they’re leaving
This will help you focus on what to redesign and test.
Ask questions that reveal intent or points of friction.
Start with one easy & fast question.
The longer and more detailed the survey looks, the less likely your visitor is to participate
If they answer the first question, THEN populate additional questions or an option to add details.
Example:
First prompt says:
“Leaving so soon? How did we fail you?
Multiple button-click options:
- I don’t want to give up my contact info
- I don’t want what you’re offering
- This isn’t what I thought it would be
- It’s too expensive
- Other
Pick 2 + other, dependent on your goal
Selling a product – ask about purchase pain points
- Too expensive
- Want instant download, etc
Collecting contact info – ask about the value of what you’re offering in exchange
Once they choose one, you can populate with additional questions
They’re already started on the process – more likely to answer another 1-2 questions before getting fatigued.
Use a combination of multiple choice and open-ended questions.
Examples of open-ended questions:
“What was the purpose of your visit?”
“Were you able to find the information you were looking for?”
“What should we do differently?”
Multiple choice questions are easier – more likely to answer
Open-ended questions allow free-flowing answers — provide the most insightful data.
Create your own exit-intent surveys with free tools like Qualaroo and VWO.
Option 2: Feedback surveys
Send a feedback survey (via email) to your existing customers.
Goal = understand what motivated them to convert
or what prevented them from converting.
You can easily create a survey with SurveyMonkey or Google Forms.
Option 3: Perform usability testing
Surveys have limitations:
Not everyone will participate.
May not be accurate – no guarantee questions were answered honestly.
Usability testing reveals what visitors actually think of your page.
Record visitor’s interaction with your page.
See exactly where they look, click, and pause
Identify points of confusion that prevent them from completing your goal.
HotJar and UserTesting are a few tools you can use to run these tests.
Example of a heat map:
Shows potentially problematic areas on your landing page
Use that info to brainstorm the sequence of testing you’ll implement to improve performance.
Step # 2 – Prioritize What To Test
A few minutes of brainstorming and you’ll probably have a HUGE list of stuff you want to test
Some essential tests for landing page optimization include:
-
- CTA: Placement, Copy, Color, Size
- Copy: Headlines, Word Count, Value Proposition
- Forms: Length, Field Type (auto-selected vs. pre-filled vs. manual)
- Images: Placement, Relevance
- Pricing: Anchoring
Now you have to figure out where to start.
Again: Only test one thing at a time
It’s the only way to confirm success or failure of a change
Start off by creating a list of elements
Once you have a list of you want to test, you need to prioritize testing order.
Consider:
- How easy or difficult is it to deploy a certain test?
If a test will take months to design and execute, it’s probably not a great candidate for 1st round.
- What is the potential impact of your test?
Obviously, test that offer! Likelihood of big results should get higher priority
Assign a score to each item on your list.
- Low score to elements that will probably give a small boost
- High score to elements you think are causing a problem you’ve identified
There are times – if you know you have several major issues, you should overhaul the page in its entirety, test new vs. old, and then start testing elements on the new, better design.
Example:
Highrise did several rounds of small tests
Failed to achieve their desired result.
Did a major page overhaul:
Big tests let you innovate and try out drastic changes.
Small tests refine and keep your page performing well.
You can find ‘big wins’ hidden within customer feedback.
Find issues reported by the majority.
Step # 3 – Run Your Top Priority Test
This seems overwhelming.
Luckily, there are tools!
Unbounce or Optimizely let you create page variations through a visual ‘drag-and-drop’ approach.
Important:
Divide traffic between your control (original page) and the challenger (revised page) equally.
Example:
Testing between a red CTA and green CTA
Send 50% of visitors to page with red CTA
Send 50% of visitors to page with green CTA
This lets you compare conversion rate without additional variables
How long to run a test?
At least 7 days. Or a statistically significant amount of traffic (100/1,000/10,000)
Really, however long it takes to get enough page visitors to produce reliable data
Remember: anomalies can occur
On a certain day of the week
Seasonally
Around holidays
Extend or adjust your testing period accordingly
VWO offers a cool test length calculator.
When test has run for appropriate time, now interpret the results.
Results are always given with confidence intervals.
Confidence intervals represent the range of result reliability for each page variation.
Like when you say “plus or minus 5%”
Factor in the confidence interval to get a rage of conversion success
Example:
Original page produces a conversion rate of 2% with .3% confidence interval
That means it converted between 1.8 and 2.3%
Test page produces a conversion rate of 3% with .2% confidence interval
That means it converted between 2.8 and 3.2%
In the example, there is no overlap between these values
It’s clear that the variation is the winner here.
Smaller confidence intervals are always preferable.
Large intervals make it hard to trust results
- Are they due to your variation or a random sampling error?
Not confident in your ability?
You can outsource testing to a third-party
They’ll manage everything from strategy to implementation.
Step # 4 – Rinse and Repeat
Testing is an ongoing process.
You may think landing page has hit a ‘sweet spot’ and that you no longer need to test.
Remember:
Landing pages can always do better
You should continuously test for improvements.
After running a test, promote the winner to be a new control
Test the winning page against another (single) variation.
Your landing pages are often the first experience a new lead has with your brand. It can make or break the likelihood that a page visitor will act on your desired goal.
Making changes can be risky if you do it wrong.
But they can produce amazing results for your conversions – and your bottom line – when done methodically and correctly.
A/B testing gives you the best opportunity to get immediate results without risking the farm.