Optimization And Testing

Let’s take a look at the basics for testing and optimizing your funnels.

When a website visitor takes the desired action, it is called a conversion.

Common conversions on a web page are: filling out a lead form, adding a product to a shopping cart, or registering for a webinar.

Conversion Rate Optimization (CRO) is a process designed to intentionally increase the number of conversions on your website or on a specific web page.

Optimization and TestingA well-designed CRO program is comprised of an 8-step process:

  1. Step 1: Identify Goals – You need to identify metrics and goals first. Without a goal or KPI to measure, you cannot optimize.
  2. Step 2: Gather Data – Gather the data that is relevant to the goals and metrics you have identified.
  3. Step 3: Analyze Data – Analyze the data looking for anomalies, patterns or ?…….?
  4. Step 4: Create Hypothesis – Use your observation of the data to form a supposition for the competing variant.
  5. Step 5: Design Variants – Create the variations that will be tested. Heads up! This part of the process often involves outside help from web developers and others. As a result, CRO campaigns can get stalled in this step.
  6. Step 6: Implement Tech – Use a testing application (often called a “Testing Tech”) to run your test.
  7. Step 7: Test – This is the fun part. Run your experiment.
  8. Step 8: Analyze Results – Dig into the results and either implement the variation or leave the control in place.

The CRO process is cyclical. After results are analyzed in Step 8, the process begins again with identifying goals.

Perhaps the most critical skill in CRO is knowing whether a split should be conducted at all.

In some cases, this decision is driven by simple math. For example, to run a reliable test of two variants over a 7-day period, you’ll need 29 conversions per variant, per day to reach a reliable conclusion.

If you don’t have that number of conversions on that page, the math simply doesn’t work and the test should not be run.

Even in cases where the math points to a reliable test, there are times that a test should not be run.

For example, if something simply doesn’t function properly on a web page, it should be fixed rather than tested.

The good news is that split testing is just one way to optimize for conversions even if you have very little traffic.

There are other, often simpler and more lucrative, conversion rate optimization tactics such as:

  • User Surveys – Asking users to answer a question or series of questions that yield qualitative data for analysis.
  • Session Recordings – A recording of a website visitor interacting with your website.
  • Heatmaps – A graphical representation of the mouse clicks (and often scrolling patterns) on a web page.

Here are the important metrics to monitor:

  • Conversion Rate – The rate at which website visitors are taking the defined conversion action.
  • Lift Percentage  – The percent increase or decrease in conversions between the control and variation.
  • Confidence Rate – A measurement that shows the reliability of the test results.
  • Conversion Range – The range you should expect your actual conversion rate to land between if you set the variation live.

Here’s a breakdown of the terms I’m using (in case you’re not sure what these all mean).

  • Conversion – The activity you are hoping to improve with the campaign. (e.g. Registering for a webinar, adding a product to the shopping cart, etc.)
  • Control – The page in the experiment that does not receive the treatment. In conversion testing, the “control” is currently the best converting version of the page.
  • Variation – The page in the experiment that has received treatment. For example, the ‘variation page’ might have a shorter lead form than the ‘control page.
  • Quantitative Data – Data that can be measured numerically.
  • Qualitative Data – Descriptive data that is more difficult to analyze but often provides context to quantitative data.