Thorough and early product instrumentation is a recommendation we make often for the founding technology teams we work with. In that vein, these scrappy teams usually want to know how much time and effort it takes to implement that instrumentation.
In my experience, a client asking this question is really after two different answers: First, how should I prioritize this among my other pressing development priorities?
Second, since as a marketing team we rely on a client’s engineers to implement a spec that we draft, they are also probably wondering, “I’m paying them, how much of this do they need me to do?”
Let’s take a step back and talk about the factors that affect how much time your developers will need to spend on instrumentation and what that means for your organization. A little lost? Check out our recap on what product instrumentation even means.
Solid Instrumentation: A 3-Step Process
We recommend a 5-step plan to roll out your instrumentation, starting with determining what instruments you will use.
1) Identify your technology stack
Start by choosing the tools that you will use to collect and process user data. Check out our recommendations on those choices of tools.
2) Write a spec
We recommend breaking down each type of API call and user data you will connect in a written document, so that your user data is deliberate and clean. Organize this document by the type of call, and list the data collected on each.
3) Implement, QA and Launch
After you have a solid plan on how to handle user data in your organization and what tools to use to process it, your engineering team can implement the data collection. From there, use a development environment in your product and data tools to test the instrumentation before launching into production.
Three Key Factors that Affect the Time to Implement
We’ve found there are three key factors that affect how big of a lift implementing robust product instrumentation can take:
1) Organization and documentation
No surprise, but without a clear and concise plan, instrumentation implementation can be a mess. If you wing it, not only will it be a hairy mess this first time, but you will likely incur a substantial amount of technical debt and need to redo it or re-document it at some point in the future so your product is built for scale.
2) Availability of an existing spec
The more complex or novel your product, the longer it will take to instrument. Our preferred Customer Data Platform, Segment, provides off the shelf specs for Ecommerce, Mobile, B2B SAAS and more common product types. These are more or less solved problems that are simple to instrument.
3) Number of user attributes and actions
If your product is extremely mature and there are hundreds or thousands of user attributes and actions to collect, instrumentation will take more time and effort than a minimum viable product with limited actions available.
The Bottom Line
In our experience, startup technology companies can instrument their product from a detailed spec in just about a day of development time, and QA and release in a single sprint. Of course, if you meet any of the criteria above, it might take longer. But I can promise, it will certainly take longer next quarter than it would if you started today.
You know what they say–the best day to instrument was pre-launch, and the next best day is today.