You’re in the middle of a budget pitch for your next analytics project when the CEO says, “I agree that we need to get better at using our data, but I want to be sure we’re clear about the value. What exactly are we going to get and how confident are you?
You’ve been working for months to get all the key executives aligned and ready for this discussion. Everyone seems to agree the project should move forward, but the CEO’s question has got you thinking…
Is your confidence about the project’s success
based on your gut or fact?
Obviously some projects are ‘no brainers’, but generally most analytic projects operate at the edges of your business. That’s usually where the greatest value opportunity is, but that’s where much of the risk is as well.
It’s unrealistic to expect that every project will be successful. So how can you know if the project will succeed without actually doing it? Seems like the old chicken and egg paradox.
While no one can predict the future, there are ways to improve your confidence in a decision to move forward with a new analytic project. To begin with there are some key questions you can ask that will help give you a more fact based assessment of a project’s chance for success. Generally the more effort you invest in answering these questions, the greater your level of confidence.
How clear is the business question?
Every successful analytics project begins with a clear business question. Example business questions include: What actions will retain more customers? Which pricing strategies will strike the right balance between growth and margin improvement? Getting the focus right will ensure the organization is in agreement that the project is solving the right business problem. This seems very basic, but it’s extremely common to see projects where the business problem can’t be expressed in a simple and clear fashion.
What is the business problem worth?
The key to successfully answering this question lies in defining the potential actions required to solve the problem and understanding what those actions could be worth. For example, solving the customer retention problem above means understanding how many customers are at risk, what it will cost to keep them, and what those customers are worth if you succeed in keeping them. Answering this question can be tricky because leaders often hesitant because they feel they’re being asked to make a commitment. However you have to question moving forward with a project if the organization is unable or unwilling to answer this particular question.
How complex is the analysis work?
Answering this question should be broken down into two parts; the data and the analysis. From a data perspective it’s important to understand how familiar your team will be with the required data sets. Will working with the data be exceptionally complex or relatively straight forward? From an analysis perspective you should consider if the work will be a ‘first’ for your team. For example, will it be the first time your team has done predictive analytics, spatial analysis, or visualization? If so you should evaluate if that type of analysis is truly necessary to achieve the project or can be developed at a later time.
What is the accuracy and completeness of the data set?
Most organizations have a general awareness of the data quality for the data used every day to manage their business. Unfortunately that’s often not the data needed for new analytics projects. This can mean using data that has a lower level of data accuracy and quality. If your project also requires external data you’ll be up against a whole new set of data challenges. It’s important to consider the level of confidence your organization will have in taking action from the answers derived from the data needed for the project.
How difficult will it be to achieve the business outcomes?
If a given project requires a change to the organization’s pricing strategy, that’s relatively straightforward. It’s an internal change that the organization has direct control over making happen. However if your project requires customers to behave differently (e.g. improving retention), then the outcome is significantly more difficult to achieve. As you evaluate this question consider elements like; how big of a change is required, the level of direct influence the organization has, how difficult it will be to measure success, and whether there is a strong executive leader.
After answering these 5 questions, you can simply score each on a scale of 1 to 10 to quantify your confidence in a given project. You can also use your answers to help craft a value story for the project.
How can you really know if an analytic project will be
successful without actually doing the project?
In situations where there’s not enough information to effectively answer the questions above, there are still ways to move a given project forward. Below are 3 ideas for making small investments in order to evaluate that project’s potential for success.
Do a proof of value project
A proof of value is a small project where you establish a controlled environment to experiment and see if the idea proves its value. The project could be a one-time analysis project to quantify the potential opportunity. Or it could be a project to prototype a specific solution or tool. The learnings from these types of efforts can be invaluable in evaluating whether to move forward with a project.
Conduct exploratory data analysis
Exploratory data analysis can be helpful in situations where you aren’t sure you have the business question properly framed and/or you’re unsure about the accuracy, completeness, or usability of the data. Often profiling, transforming, and cleansing data will highlight gaps and shed new light on the viability of a project. Many organizations will do exploratory data analysis as a standard part of their pipeline for developing new project ideas.
Look for others who have done something similar
Depending on the industry and project, you can often find others outside your organization who have experience with the given problem you are trying to solve. Sometimes there are vendors and/or 3rd party tools that can help accelerate your project. In those situation I’ll often advise clients to provide the vendor with a sample of their organization’s data in order to do a proof of value and prove out the project concept.
Every new analytic project entails a certain amount of risk but it is possible to manage through those challenges. By spending some time up front to properly define the business problem, the potential value outcomes, and the key risks areas, you can significantly improve your confidence in recommending your next analytic project.
What are your thoughts? What other ideas do you have to improve an organization’s confidence that a new analytic project will be successful?
Latest posts by Phil Kelly (see all)
- Five Ideas to Improve your Analytics Returns - October 24, 2018
- 5 Reasons to Take a Portfolio Approach to Analytics - September 27, 2018
- How Small Provider Networks Can Be Data-Driven without Breaking the Bank - May 11, 2018