This tool is meant to help you organize and document an analysis project and increase the chances of a project’s success. Taking you through a project life-cycle from asking business questions to presenting insights, the tool covers the major stages of any analysis project from a high-level, repeatable, comprehensive process.
If this documentation tool meets your needs, feel free to use it in whole or part, or modify it to suit the needs of your own projects. And, if you have any questions on this process or comments from your own experience, don’t hesitate to reach out!
The data analysis life-cycle
Any data analysis project should flow through five stages:
- Business Understanding
- Data Understanding
- Data Preparation
- Data Analysis
- Data-Informed Decision Making
While the flow moves generally in a downward direction, it may eddy and swirl a bit from later to earlier stages during the lifetime of a project.
Whatever you do, do not skip this step. It is critical to get a solid grasp of the business requirements for your project before you dig in to the data or the analysis. Begin by collecting all the requirements for the project from the end user’s perspective. A framework I pull from again and again is Michael Quinn Patton’s Utilization-Focused Evaluation (U-FE). This approach can be summarized as “the intended use by the intended user.” Everything you do in your project should focus on how the end product is going to be used.
In this stage you will also determine and record the KPIs, metrics, outputs, or outcomes to be created through the analysis, as well as define terminology. Lastly, there is a section to record meeting dates and notes so that it is clear who was part of what conversations and the decisions that were made. This doesn’t mean that you cannot make changes, edits, or revisions to your main questions going forward, but it establishes a record of trust in the process and a reminder of the decisions made by key stakeholders.
To gain a solid understanding of your data you first need to know where it comes from, who the experts are and how data is collected by whom. This adds the necessary context to the data sets before you dig in and will tell you where to go with your questions.
If possible, set up meetings with the data experts and even a time to observe data being collected, if applicable. And if you need to do further research, record the sources you accessed.
If you are running reports or exports of data, be sure to record how this is done so it can be repeated accurately in the future by either yourself or others.
After providing general information, the data prep stage follows five steps: screen, diagnose, clean, verify, and report. If one doesn’t yet exist, create a data dictionary for your data sets.
The purpose of this section is to record the processes you followed so that they can be repeated in the future. There is also room here to include relevant background information as well as the user’s needs so that these remain at the forefront.
Data-Informed Decision Making
This last stage is critical to ensuring that analyses are used by the intended users. It concerns not just the final decisions that get made, but the efforts by the analyst to present this information in a way that is actionable.
Before the final draft, feedback should be baked into the project. Again, record meeting notes and the decisions made by stakeholders.
This stage provides a chance to think through how to make an informative (or persuasive) and engaging presentation.
Lastly, if available, record what decisions were made or actions taken with consideration toward the analysis.
All good analysis projects will incorporate these five stages in some capacity, whether a one-day, ad-hoc request, or a months-long project. If this documentation tool meets your needs, feel free to use it in whole or part, or modify it to suit the needs of your own projects. And, if you have any questions on this process or comments from your own experience, don’t hesitate to reach out!