Building a data tracking plan in order to monitor and analyze your customer journey is at the root of customer analysis. However, setting up a tracking plan can seem daunting. If you’re just getting started with customer analytics, or are looking for advice on the best way to establish a tracking plan, this blog will provide a solid baseline introduction.
Customizing a data tracking plan can be broken down into these three key steps:
- Identify Key Use Cases
- Identify User Actions to Track
- Build a Data Tracking Model
Now, let’s dive into each step.
Identify Key Use Cases
The first step in building an effective data model is to define your business objectives and use cases. Think about your company’s overall goals as well as the metrics and KPIs you would like to analyze, track, and improve. Some examples of these could be:
- Increase new user conversion
- Increase subscriber retention and LTV
- Measure the success of a new app or website feature
- Identify your most effective marketing channels
- Identify points of friction or drop-off in your on-boarding flow
- Understand the flow of users through your conversion funnel
Next, identify which questions you need to answer to help reach your objectives. Let’s say your business objective is increasing usage of a new feature by optimizing your engagement funnel. Some examples are:
- Do new users who use X feature within their first session have higher engagement rates?
- Do paid plan customers use certain features more than those on a free plan?
- What screens do new users view first on the app?
Once these are defined, you can begin to identify which data points are necessary to find the answers you need. You can track hundreds or even thousands of different user actions, but most questions require less than a dozen to find the answer. Let’s take a look at which user actions you should be tracking.
Identify User Actions to Track
You’ll want to begin thinking about the different steps in your customer journey and how a customer may go through them. Throughout this article, the examples will be based on a fictitious company, KittyCarts, that sells cat products and has a site, blog, app, and camera feature. At Indicative, we use KittyCarts as our demo data. Using KittyCarts, let’s say we want to increase new feature engagement through our engagement funnel.
We determined the first user flow we are going to track is new user adoption of the KittyCam filter, we identified as a critical engagement metric. Continuing with our example, we have rolled out a new feature and want to analyze how users are moving between app installation and applying the KittyCam filter.
Mapping out this process is important because you are able to visualize what your customer must do to move from app installation to applying the KittyCam filter. Once this data is being tracked, you can also identify any points of friction. For example, if all new users open KittyCam, but never use the KittyCam feature, there may be something wrong with the filter.
Now that you’ve mapped out your first key funnel, you can create a clear definition of each step. Our example journey would look like this:
- App Installed
- A user installed the app on their device.
- Intro Landing Screen
- A user opened the app for the first time and landed on the intro screen.
- Sign Up Selected
- The user selected a signup method, using social media or email authentication.
- Personal Details Confirmed
- The user reviewed their name and phone number, then clicked “confirm.”
- Account Created
- The user clicked “Create Account.”
- Open KittyCam
- The user opened the KittyCam feature.
- Apply KittyCam Filter
- A user applied the KittyCam filter.
You can now begin to think about the context of each of these events and how they can give insight into user behavior. For example, is a user who completed “Sign Up Selected” using Facebook more likely to complete the flow than a user who signed up directly on the app? These contextualizing bits of data will become the event properties tied to each event, and they enable you to segment your users and compare behavior between groups.
Some event properties, such as date, time, device type, etc, can be applied to all of the events, and others, such as use KittyCam filter, will be specific to a single event. For example, let’s view a few event properties tied to the event “Apply KittyCam Filter”:
To ensure a clean data taxonomy, it is just as important to create definitions for event properties as it is to create them for events. Our example’s data taxonomy would look like this:
- Apply KittyCam Filter Type
- The type of filter chosen by the user.
- The device utilized by the user.
A well structured data taxonomy allows for easier analysis of the media dimensions that matter the most to your business and keeps your data consistently organized.
Build a Data Tracking Model
A data tracking plan lays out which events to track, where those events go in the code base, and why those events are important to your business. Having a tracking plan benefits all members of your team because it builds consistency.
According to Segment, a good tracking plan should:
- Summarize which events and properties need to be added
- Justify why they need to be tracked
- Detail where in the code base they need to be added
- Inform stakeholders of progress/completion
Data tracking plans work best when they are created with easy access in mind (remember, all employees involved should be able to access this file and make changes or additions when needed), so we created our data model in a Google Sheet. Based on a piece of KittyCart’s engagement funnel, here’s a sample tracking plan.
Our basic data model includes the use case event, key properties, source, table, mapping, unique identifier, timestamp, notes, and status. By keeping track of this information, anyone can look at this sheet and identify the information that they need.
By taking these three steps to build a tracking plan, you’ll be on your way to analyzing your customer data. If you need an analytics tool, you can try Indicative for free or contact sales for help setting up your data warehouse