You don’t need me to tell you that having a robust analytics platform is not just a must, it’s a hard requirement for product growth. Despite fulfilling this requirement, many companies often find themselves with a lot of data but little return on data investment. There are multiple reasons for that and, by the end of this article, you will understand how to solve one of the most challenging ones. So let’s discover how to set yourself up for success and set the foundation for analytics excellence!
More often than not, when I ask teams what their priorities are with tracking data, they tell me they want to implement everything. I usually consider that alarming because it suggests there isn’t a coherent strategy behind the analytics implementation, and the starting point is unclear. This is why a good implementation coach first recommends building a concise tracking plan that reflects the top KPIs or use cases. But it’s not always the best way forward, and tracking everything at once does have its benefits (especially if you’re leveraging auto-tracking features).
So what’s the correct way to implement and cut through the noise? Whether you’re implementing Amplitude, Mixpanel, Adobe, or another analytics platform, the method I recommend below is the same. And, as we’ll see, it’s more nuanced than choosing one of the above. But first, let’s have a quick look at the pros and cons of each to better understand them.
Method 1: Tracking All Data at Once
The “track everything” approach involves collecting all available data from the beginning. This method aims to capture a comprehensive dataset, which often means dozens or hundreds of events, ensuring no potential insight is overlooked.
Advantages
- Comprehensive Data: Provides a broad dataset that can be used for deep, multifaceted analysis.
- Unexpected Insights: Allows for the discovery of trends and correlations that might not have been initially apparent. Captures data you might need later.
- Flexibility: Future-proofing your analytics by having all data readily available for any type of analysis.
Challenges
- Data Overload: The sheer volume of data can be overwhelming, making it difficult to identify actionable insights. You may not have everyone ready to use the data at the early stage anyway.
- Higher Resource Cost: You will need to dedicate extra time and energy to this, which often translates to precious engineering (or data analyst) resources. You also want to double-check that you’re within your analytics provider contract limits (e.g., event volume, monthly tracked users in the case of Amplitude) as well as any other connected tools like Snowflake or Databricks (for the compute cost). Your analytics provider or agency may also charge you a higher implementation cost because they’ll need to dedicate more time to advise you.
- Data Noise: The presence of irrelevant or redundant data can obscure meaningful insights. Not all data is useful.
- Not conducive to change: You’re trying to do a lot at once, so be prepared to handle a lot of planning and change management.
Method 2: Tracking Only What’s Important
The “track what’s important” approach (or MVP analytics implementation) focuses on first collecting and analyzing only the most critical data. This method prioritizes key metrics that align closely with your main business goals.
Advantages
- Faster Insights: By focusing on essential data, you can quickly generate actionable insights.
- Cost Efficiency: Lower initial costs for resourcing (i.e. engineers). Data processing should be cheaper and more controllable if you’re implementing it through cloud storage solutions or a data warehouse.
- Manageability: Smaller datasets are easier to handle and analyze, reducing complexity and smoothing out your learning curve.
- Conducive to change: You will still go through organizational change, but it will be easier to manage, and you’ll build a healthier adoption for your chosen analytics platform.
Challenges
- Limited Scope: You may miss out on secondary insights that could be valuable in the long run. Remember, however, that most tools allow you to backfill historical data later.
- Selection Risk: Requires careful identification of key data points, which may not always capture the full picture.
- Scaling Needs: As business needs evolve, the scope of data collection will need to expand, requiring additional investments in data collection over time.
The Best Way to Implement an Analytics Platform
As you might’ve concluded, the decision often boils down to dedicating more resources upfront vs. being more strategic and allowing some initial data gaps. However, this choice is an illusion. That’s because, in reality, you always have to start somewhere and plan your implementation steps, whether these steps will be performed now or later. So we’re going to blend both methods into the ideal implementation method:
- If you’re already collecting data via CDP, data warehouse, etc. you can import all that data directly to your analytics platform of choice (provided an integration exists and you check you’re within the limits). Remember to hide/delete/block unnecessary data after starting the import and double-check any third-party costs (i.e. data warehouse compute).
- The same applies if the platform supports auto-tracking (or auto-capture) through its SDKs.
- Define a set of 4-5 KPIs and the necessary data that will support those KPIs. That means building a tracking plan with several events and associated parameters/properties. If your initial KPIs can already be constructed with the data from Step 1, you can skip to Step 4.
- Implement that tracking plan in your analytics platform with the help of your engineering team.
- Repeat steps 2 and 3 until all your KPIs or use cases are supported by your data. Continue iterating on your tracking plan to expand analytics literacy in your organization.
Ready to implement your analytics platform? Assess your data needs and choose the approach that fits your goals. If you need further guidance, contact me for a consultation or explore the other resources on the site.
