Leading a Successful Data Analysis Project in 6 Steps

Leading a data analysis project has a unique set of challenges.  Here are some steps to ensure your analysis is a huge success - even if your hands don't touch the keyboard.

1.  Begin With the End in Mind.  What Questions Are You Trying To Answer?

Stephen Covey coined the phrase "begin with the end in mind" in his bestselling book The 7 Habits of Highly Effective People.  Simon Sinek wrote a whole book with a similar theme called Start With Why.  It can be tempting to just start going and figure it out as you go.  Don't.  There is a time to refine your goals and ask more questions as you get further in the process and are exposed to more information.  But you can't refine what you don't define. It doesn't have to be high-tech or formal.

2.  Guide and Focus the Data Discovery Process  

Even though you may not be gathering the data yourself, You should be an active participant in validating the work done by the technical teams as they locate possible data points. Be prepared to ask questions on what they find.  Some example questions: "Did you find information on how to split by different business units? What values are there? Why is the East Region not shown in your list of values? Did you use Salesforce data or ERP data for product name? How often is the data updated?" Keep in mind this is not meant to be a combative discussion, but rather collaborative and focused on the business questions you defined in Step #1.

picture of a keyboard with a discovery iconAs a leader, your job at this stage is to help focus the technical team.  Are they exploring a lot of data points that don't seem relevant? You should politely point that out and dig a little deeper to understand why that information is included. Is that one business question you dreamed up going to cost months of work to collect and process? Can that be moved to a later phase? Having an open discussion with the teams can help the project succeed. Technical resources can get lost on this step (I have been there!) and need your guidance to stick to the defined goals.  Just because you can pull other data points doesn't mean you should if they don't support the project goals.

Excel or Google Docs can be a great tool to communicate findings between the technical and non-technical teams for complex projects.  This will ensure everyone is on the same page, understands what each data point means and any limitations or "gotchas" in the data.  If you don't have time or resources to document all variables, do this for your top ones of critical importance.

3.  Evaluate and Prioritize the Data Standardization / Data Clean-up Effort

A co-worker of mine would regularly use the phrase "garbage in, garbage out!" when describing some of the poor results his technical solution was delivering.  Without addressing them, data quality issues will jeopardize your entire analysis and could lead the team to make incorrect decisions.  This step is one of the most challenging, time consuming and expensive steps of your analysis.  Here are some tips to get you through.

brush cleaning up data

It is inevitable that your team will come across some data points that don't make sense or are clearly wrong.  As a leader, your role is request examples and analysis on the data quality issues the team has surfaced.   Review the examples and quantity of occurrences. It may be impossible to eliminate all data quality issues.  Thus, adding prioritization and focus to the project will be critical.  As the technical team provides you this information, evaluate it using the following two criteria:

  1. importance of the metric to the analysis
  2. time and cost to resolve the quality issues

Based on these criteria, evaluate the technical team's solutions and decide how to proceed.  Maybe you let go the industry issue that only impacts three customers but you tackle the industries that impact hundreds of clients.  Be sure to suggest alternative solutions or resources that could be used to ensure the best end result is achieved.

4. Decide if Data Appends Are Needed

You may choose to explore options for supplementing the data points you have available.  This can be achieved by leveraging external (paid or free) resources to append industry codes, contact information or other data elements not native to your data source.

puzzle pieces coming together

You can also partner with your team to simply utilize the data points you have to derive new data points.  This allows you to better segment your data to answer your business questions.  An example could be to create a new data point called "US region" based on a zip code mapping of your data.

5. Ask Questions.  Does Everything Make Sense?

Your technical resources should be performing quality tests on the analysis to ensure everything makes sense and is explainable.  Have an open discussion with the team about anything you find that doesn't seem right.  Ask questions and hold your team accountable for digging in and getting to the bottom of anything out of line.

6. Review the Results and Ensure Your Questions Are Answered

At this time, you will be reviewing the team's findings and analysis.  Provide feedback on the visuals you find helpful and those that could be improved.  Make sure you understand them and ask questions if not.  Encourage the team to explore other variables or things they might not have presented.  Revisit the established goals from Step #1 and ensure the analysis provides the answers you were seeking. Learn from the project and provide feedback to the team.  Request feedback from the technical team on what they felt you and the leadership could have done better for next time.

Three data visuals about text messages