We’ve talked a lot over the last few articles about how to implement different digital marketing strategies. After you have had those up and running for a while, you will have a good baseline of metrics to figure out what’s working and what’s not in converting prospects. Only a fifth of businesses are satisfied with their conversion rates. That’s only 20%. This is why, for many businesses, conversion rate optimization (CRO) has become an integral part of their digital marketing strategy.
First, let’s be clear here. Conversion doesn’t just mean a sale. It can mean converting someone into a subscriber, for example. Not knowing what your definition of conversion means you can’t effectively improve your business. How are people going to know what action you want them to take if you are unclear yourself?
Methods of Well-Executed Conversion Rate Optimization
The main goal of optimization is to improve conversion rates of established traffic, with already established assets. You can’t optimize new assets, because there’s nothing to optimize yet. You’ll need a baseline of data to start off with in order to improve it.
Optimization is a repeatable process. Sure, you may be happy with the results after your first attempt at optimizing an aspect of your marketing, but that doesn’t mean it can’t still be improved. The process of CRO is cyclical; once you finish it, you start again. Thinking of this process like the scientific method can help you easily perform conversion rate optimization on any aspect of your marketing strategy.
Identify Your Goals
Before anything, you need to start with a clearly articulated goal. What are you trying to achieve? Without a goal, you can’t optimize anything. Getting specific can help you determine exactly what you want improved and can help you get the best end-results.
For example, optimizing your homepage is one of the most difficult to optimize because it performs so many tasks, and there are lots of things there that might need to be optimized. Your homepage is a welcome to new visitors, should tell why your business is relevant to them, and where they need to go next. For this page, you might set one or more of these three goals:
- Immediate goal, clicks or on-page form completion
- Campaign goal, leads generated or purchases
- Long-term goal, impacting long-term value, net revenue, average order value, lead quality, etc.
This is why you can’t optimize anything brand new. You need to set a baseline for your metrics so that you know where you can go. Log your current numbers, note what you’re aspiring for, and any relevant user data. Do this BEFORE making assumptions, as assumptions can lead to biased results. You can get this data from many places, but the most common are:
- Google analytics for site metrics
- Data provided by your email service (ConvertKit, Infusionsoft, Aweber, etc.) for customer email data
- Payment processor (Stripe, Paypal, etc.) for payment data
- User behavior charts from TruConversion show where people are clicking on the page
The key in analyzing data is to use relevant data you just gathered in order to develop meaningful optimization. Using irrelevant data can skew results and hurt you in the long run. When you are reviewing your baseline numbers, ask yourself the following questions:
- What is my conversion rate? Is it acceptable?
- What’s hurting my conversion rate?
- How/why is it hurting my conversion rate?
These questions can help you narrow down what can help you reach your goal, and from there you’ll develop your hypothesis.
Develop a Hypothesis
Not having a hypothesis means not having direction. After analyzing your baseline data, you hopefully know where things can be improved, and have ideas on what might improve them. There are three elements to a hypothesis:
- The change or approach you plan to test
- Who you’re targeting
- Outcome you predict
Your hypothesis will look something like I believe that  for  will cause  to happen. A hypothesis is about your intent, so make sure it’s something you can measure.
To save time in between tests, you may want to try and test a few things at once. While there’s no limit to the number of tests you can run, smaller sites should limit how many they run at once because it will take too long to get valid results, and some tests might interfere with others. A good rule is that if you don’t get a lot of traffic, you probably don’t want to run more than about 29 tests in a year. This gives each test time to breathe and get accurate results. Some examples of tests you might run are:
Test Example: Changing the Button Copy
This test is exceedingly simple because it only involves a few copy changes. This makes it quick to set up, but it’s not very meaningful or scaleable overall. This test also won’t impact other tests in a significant way, so it’s a good one to do while running other tests.
Test Example: Tripwire Control Page
This is a complicated test because it’s a major redesign. This test takes longer to set up and implement, but we’ll learn a lot more, your findings can be applied across multiple pages, and it’s scalable. For more complicated tests, like this one, build mockups first! This way you can tweak easily as you see fit if the initial results don’t go as you’d hoped.
Implement Testing Technology
Once your variants are made, implement any technology you need to make changes you’re testing. These may be a Visual Website Optimizer, Google Analytics, TruConversion, etc. These will help you keep track of your test results.
Run Your Test
Testing your hypothesis takes time — not just to create, but to run long enough to get accurate results. It’s important to know when not to run your test. Every test should run until it achieves statistical relevance, so you know that you can trust the results. Statistical Relevance is how you prove mathematically that the outcome of your test is reliable. Put simply, the higher number of variants you have, as well as the fewer conversions you have at the beginning, you should extend the length of your test. If you end a test too early, you might not have time to see any changes. This can be hard to get when you don’t get a lot of traffic, because no traffic means no data.
Remember, not all tests need to be completed. If your test shows a crash and burn on day one, stop the test because something went horribly wrong. Your goal is to minimize risk while testing new ideas, so continuing a test that goes bad from the start isn’t going to give you any helpful insight. If an organic traffic source is performing poorly, leave it be as it’s not costing you anything and come back to it later
Should I run the test?
Sometimes, you might not need to run a test. Certain things might objectively not be working, such as a broken link, and therefore just need a quick fix. To decide if something is worth testing, consider the following:
- Is this a functional issue with no ambiguity toward the solution? If it’s just a functional issue, don’t test it.
- Does this page directly impact long-term goals? If yes, run the test.
- Is there something else to test first that would have a bigger impact?
- Are the results going to be scalable to other parts of my site? If no, it might be a good idea to run a test you’d get more value from.
- Do I have the resources to both run the test and implement the results? If no, wait until you do.
Analyze the Data
You’ve run your test long enough to achieve statistical relevance, so now it’s time to analyze the data. Analyzing your results will give you the information, or the “why”, that will fuel your next campaign. During this stage, you should do these 5 things:
- Share any improvements or declines
- Figure out “why”
- Write a report:
- Include the name of the test, the timeline, metrics used, show the variants, and break down long-form numbers, explain what you learned from the results
- Archive your data.
- Share results with stakeholders.
During analysis, you should also ask these questions:
- Did you see any changes? Good? Bad?
- If your results contradict your hypothesis, why?
- Is it worth retesting? Were there unusual circumstances that you didn’t foresee that could have skewed results?
- How can this data be used in experiments in the future?
Rinse and Repeat
These findings will be used to start the process over again. Yes, after you finish a test it’s time to start over. Maybe this test triggered a dip somewhere else, so you need to focus on something else. Perhaps after the test you found another thing that could increase your conversion rate. Or maybe there’s been a market shift and you want to try something new. Testing and optimizing is one big loop; the end of one test leads to the beginning of another.
Still have questions about conversion rate optimization? We can help! Contact us here for more information.