There are a lot of threads tying a successful data program together. But whether those threads make up a beautiful tapestry of data insights or a chaotic tangle (like when you pull your headphones out of your pocket) comes down to a core, not-often-talked-about practice: project dependency mapping.
Project dependency mapping is the process of breaking down and understanding all the ways your organization’s tracking code and platforms connect to each other. As developers, for effective project dependency mapping in analytics, we must take some ownership in cross-team communication with our data counterparts. You’re not going to plug your nicely detangled headphones into your phone and listen to Dark Side of the Moon out of order — that would be outrageous.
Just like you need to listen to the album’s songs in order, you need to put in place, and follow, an order of operation for your project’s tasks.
Project dependency mapping helps us help data teams not break analytics with updates and new implementations. No one wants to be the developer tagged to fix data bugs and inaccuracies. When we can avoid breaking things, everyone wins.
Chances are, your product doesn’t exist just on one platform with two tools that connect to each other. Instead, your product produces multiple data streams that end up in different analytics and marketing tools. This makes creating and maintaining a healthy analytics ecosystem increasingly difficult as you scale. Whether you like it or not, there are probably several stakeholders with data ownership who will have their own input and expectations. A good analytics ecosystem keeps everything on track and easier to deal with later on.
Product managers and data experts alone can’t stay ahead of the evolution of your data ecosystem. Relying solely on them could cause an array of problems. With a limited group in the know, you run the risk of maintaining a hierarchy around your source of truth — a data silo ⛔️. It’s likely you’ll also end up with poor synchrony between teams, leading to inconsistency in data or, worse, bugs.
This, inevitably, will lead to unexpected requests for you to fix or rework the code later. The best project dependency mapping practices require that you communicate early and often with all teams to avoid these kinds of problems.
Downstream dependencies can be tricky to keep on track. You could be waiting for someone to finish a task to move on to the next, or someone could be waiting for you to finish a task to move on (people get impatient, especially when there’s a deadline). Someone could push ahead without realizing the consequences of their actions. By documenting downstream dependencies, it’s clear who is waiting on what, and what facet of the project could be affected. You don’t want to make a breaking change without some form of communication with the people responsible for that dependency.
Beyond a note system scattered throughout your suite of workflow tools, with a more comprehensive system in place (like Avo), you can utilize things such as metrics, categories, and tags to document important downstream dependencies — like marketing campaign events or KPI dashboards. Marketing campaigns can be costly for an organization to run, so you want to make sure they’re running properly at all times. KPI dashboards rely on good data for accurate reporting, so any breaking changes to them could lead to a misrepresentation of the data and a misrepresentation of the progress of organizational goals.
Having an exhaustive map of all downstream dependencies significantly helps those making changes to be aware of every part of the project they’re affecting. While metrics, categories, and tags are a boon to this process, successful project dependency mapping also depends on collaboration and communication in the overall planning process.
To keep all of your analytics threads woven together in harmony instead of in a tangle, you need to focus on three main things. First, you need to participate in pre-release analytics meetings. Second, you need to help maintain your company’s single source of data truth. Third, you need to check in with data stakeholders regularly to retire inactive campaigns and understand what downstream consumers will be impacted by changes.
Let’s break down what each of these actions looks like.
By connecting with the data stakeholders from other teams in your organization during the pre-release analytics meetings, you’ll get a high-level view of the campaigns and platforms connected to the data you’ll be tracking. Use this opportunity to understand which platforms and campaigns will rely on your data. You’ll also be able to uncover the existing dependencies of your tracking before you add more threads to the fold.
Refer to your single source of truth for tracking implementation instructions, and refer to notes on your project’s influence throughout your data stream. Have a dedicated column, field, or notes section in your tracking plan to document your analytics dependencies within the context of each event. This helps to maintain the single source of data truth and requires communicating with the owner of an event, like product or marketing teams. Remember to document events in each of the most important funnels in your tracking plan.
Make sure your data isn’t tied to obsolete dependencies. By doing this, you’ll understand what KPI dashboard or campaigns will be impacted by your changes before you make them. Again, we don’t want to break anything, and we don’t want data inconsistencies later on. You'll know a dependency is outdated when a marketing campaign has ended, or a KPI dashboard is no longer relevant. This will keep the history of your project dependency mapping visible and in context for all teams.
Regularly review your tracking plan with stakeholders to make sure your dependencies are up to date. By having this open communication with stakeholders for each dependency, you can avoid making breaking changes to an event that has something depending on it.
A solid tracking plan combined with well-documented downstream dependencies will make your life — and the lives of those you work with — easier. This comprehensive system might sound exhausting to set up, especially if there’s a known knot to unravel. But taking the time to untangle the mess will be beneficial in the end.
All analytics projects are the sum of their parts, and without a concerted effort to map those parts, things can get messy. As a developer, you’ll be able to manage expectations from those varying sources, cut rework, and improve data quality through solid project dependency mapping.
Find out how to implement Avo, and use metrics, categories, and tags to document dependencies and improve your data planning here.