Analytics, Testing and the Customer Journey
In any analytic system that involves a multi-step conversion funnel, the customer journey contains decision points. No one point is responsible for the conversion, but all points work together to create an experience that translates to success. Each point in the funnel qualifies users for the next step.
When driving the decision process, it is similar to a relay race. With micro-conversions, each area is responsible for three things:
- Qualification
- Recognition
- Delivery (hand-off)
Given an input from either the previous step or entering the process, you can choose an open path to maximize volume or you can qualify the event and increase the quality of the process. If you find the right balance between the two or you get a unique and defined understanding of your specific audience journey, you can increase conversion much higher than the traditional 1-3% success criteria.
Qualification
In addition to driving more users to a particular part of the funnel, you can also drive more quality users. This is also referred to as qualified lead generation. Qualification should always be derived from end of funnel conversion metrics. The largest subset of segmented users that either convert more frequently or have a higher value of conversion should be identified as your highest value segment. You also want processes that move other segments into the high value value segments, don't ignore those. Pre-qualifying "should" make all processes downstream easier and more flexible. A/B testing gets easier as well since you have larger conversion indicators when you have a typical segment known for conversion being tested. That being said, you should always be careful to use these results since it will only represent a small fraction of total users. I mostly use these results to define user experience.
Recognition
Previous decisions should be presented to the user in moderation. This acts as a reinforcement to push users through the process. These are typically done in a number of forms. Some of those are (but not limited to):
- email reminders
- Recently viewed callouts
- notification banners
- urgency messaging
This recognition model is referred to as "recognition over recall". Most people will be able to recognize past behavior before they can recall it on their own. Speed helps move people along through the process. Speed of returning to a previous decision point OR speed of not having to return to a previous decision they already made. Product and service managers who understand this principle can make huge leaps forward in short amounts of time.
Delivery
Delivering a user, or qualified user to the next stage of the customer journey in a seamless way is of great importance. The process has to look and feel smooth. The smoother you make that transition, you keep the user focused on the next stage of decision process. As with any process, user experience is key to success. All the transition elements have to be uniform, similar features, repetitive patterns, and presentation elements (colors, fonts, sizing). Users tend to focus on items that are out of place if they exist. You want people focused on the decision process not thinking, "that looks off" or "why can't I do...".
As a micro-conversion metric, consider that handoff as a measure of success. This, in addition to conversion will help you define if you are sending quality leads through the funnel. There are numerous ways to drive audiences through your process and increasing volume (SEO, etc) but if they aren't converting, what is the point. Find that right mix.
A/B Testing
If you are testing simply to state an increase conversion, you might need to rethink your strategy. Run both A/B testing and A/A testing. A/A testing is used to determine a baseline and measure the consistency of the tool you are using. If you run an A/A test (measure 2 identical pages) and your results are skewed, the platform needs to be recalibrated. Use shorter, more distinct changes with micro conversions, carefully monitoring macro conversion to make sure you dont affect the platform negatively. More distinct changes give you greater signal and help you reach statistical significance sooner.
Update: Thanks to my awesome friend Milos Curcic, I want to clarify that you should NOT use faster statistical significance as a reason for ending a test early. Let your tests run their course. It was only meant to reinforce the fact that the signal will be stronger. Thanks Milos!
Use the testing to continuously improve. Don't ever be satisfied with a result. Use it to inform your next test or set of tests that will help guide users through the decision process.
Source: Analytics & The Customer Journey