UX + CRO = PROFIT: How to Use Your UX Skills to Improve Conversion Rates (Part One)
Step 1: Generate Ideas
The most effective way to generate ideas is to let research and data lead the way. Use the following approaches to get started:
Define your product’s drivers, barriers, and hooks.
Drivers are the intentions and motivations that push users to complete a transaction. They are determined by your value proposition, relevance, clarity, and urgency. You need to know:
- Who is the user?
- What is their mental model?
- What are their goals and needs?
- What is the job to be done by your product or service?
Use that information to brainstorm better ways to position the benefits and features of your product or service.
Barriers are the habits, anxieties, distractions, and uncertainties that prevent users from completing a transaction. You need to know:
Use that information to brainstorm ways to address those issues and overcome those barriers.
Hooks are the promise of a new idea or fix that entices a user to complete a transaction. You need to know:
Talk to your existing customers.
- What makes buyers want to buy?
- What persuasive techniques do we use?
- What other techniques could be used?
Analyze data to determine where you should focus your efforts.
Create a customer journey map.Journey maps are a helpful tool because they allow you to take a holistic look at the customer experience. They can expose issues and highlight opportunities for improvement.
Your existing customers are treasure troves of information. Use customer feedback forms, user interviews, and surveys to discover both issues and value propositions.
- Page value reports. These Google Analytics reports use event values to determine the impact of individual pages on conversion — a good indicator of the relative importance of pages.
- High bounce and high exit pages. Bounces can indicate that a page was unable to hold a user’s attention or didn’t clearly contain the content they needed.
- Funnels. Funnels are the pages, forms, and emails your website visitors go through to complete a transaction. Because all paying customers go through funnels, funnel analysis provides value. Funnels are also easy to test.
- Forms. Forms are the middleman between you and a new customer. Analyzing form performance can uncover opportunities. Look at data points like form errors and completion times.
- After completing some (or all) of these steps, you should have a list of potential ideas. Now it’s time to narrow down the options.
STEP 2: PRIORITIZE EXPERIMENTS
The mantra “Test Everything” is an unhealthy myth — especially when you’re a new brand or have limited traffic. Experiments need lots of traffic to generate data you can trust, and getting the required level of traffic needed to reach statistical significance means you can’t test everything at once; you need to prioritize.
There are several criteria that you can use to help you prioritize your experiences. The most important factors to consider are:
- the total visitors affected by a test
- how much we could reasonably expect to improve these visitors’ conversion rates
- the cost to implement
- the potential net revenue impact and ROI
Use these dimensions to prioritize your experiments and to determine which to run first.
It’s also important to perform a quick break-even analysis to set a threshold on the kinds of experiments you will consider. A break-even analysis will eliminate wasted time and energy on experiments that don’t have potential to make significant improvements to your site. In order to have a positive ROI, your estimated conversion rate and value improvements must be greater than the cost of development.
Estimated conversion rate lift * estimated conversion value lift >= cost of development
Estimating the conversion rate and conversion value lift can be challenging, but even order-of-magnitude estimates can reveal whether or not an experiment will be worth your time, and you’ll get a better sense for potential impact as you run more experiments.
Calculating cost of development is simply the cost of the time and materials it will take to implement an experiment variation, with totally new features typically requiring the most work, and modification of existing features requiring the least work.
The most valuable experiments should be conducted at the beginning the experiment cycle. If you are changing the layout of your site, it is best to experiment with large design changes first. This approach prevents you from experimenting with granular aspects of a design that may eventually become irrelevant. If you are validating a new feature, it is best to start with the minimal version. This approach allows you to validate assumptions as you go; these painted door experiments can be used to validate features without investing the resources in developing them fully.
STEP 3: CREATE AN EXPERIMENT PLAN
Once you’ve prioritized your experiment ideas, you can develop an experiment plan. Your experiment plan compiles all of the pertinent details in one place. It forces you to think through all aspects of an experiment and is a resource that the entire team can reference. Experiments should build on the results from previous experiments, so you should look for ways to conduct continuous experimentation.
Each experiment should be short, measurable and isolated:
- Short. Select experiments that can accumulate a statistically significant amount of data. A split testing calculator can help you estimate how long this will take given your traffic and conversion rates.
- Measurable. Define a specific measurement for each experiment so the results can be analyzed and appropriate recommendations can be made.
- Isolated. Focus on an isolated element in each experiment so it’s clear which factor causes a specific result.
Experiment plans are a critical part of the CRO process. The information documented in your plan will help you prioritize, run, and track all of your experiments.
Generating ideas, prioritizing experiments, and creating an experiment plan provide the foundation for the rest of the CRO framework. In part two, we’ll build on this foundation and talk about how to run and analyze experiments.