In order for a company to have an effective data science strategies, that business first needs to design an effective roadmap of data science.
These are the steps in that roadmap:
- Recognize and note opportunities and challenges
- Make a pitch
- Make a plan
The steps to the roadmap/framework are crucial to making an effective and comprehensive portfolio of data science for business purposes.
1) Recognize and note opportunities and challenges
Step one is about searching for methods of implementing data science in the business. This should be based on input from varying shareholders and involve business problems that can be solved for your firm via machine learning.
A crucial part of step one is not producing the solution first. Solution-first thoughts could hinder the scope of your company’s efforts to brainstorm and stymy this process prior to even starting. Rather, concentrate on areas of problems or roadblocks which could be helped through data science.
The ideas you created need to be impact-driven, instead of data science-driven. These are places to potentially implement your solutions. There’s roadblocks in the way of your goals, and if these obstacles are dealt with, then things can progress.
2. Make a Pitch
Preliminary scoping is the initial piece when it comes to pitching. When somebody comes up with a possible idea, they need to:
- identify how the idea impacts the business—“If we determine which costumers will churn, we are then able to save XYZ dollars of revenue in advertising.” Make this clear to everyone ASAP to improve the impact.
- Create the business hypothesis
- Make note of the big risks/roadblocks
Part two of the process is the actual pitching. This occurs via a big, open meeting packed with a lot of people who have ideas which might be solved through data science. The goal is to obtain the buy-in of each member on the team, thus you need to stay optimistic at first. Every idea is an excellent idea so that you can create momentum. Hold of on any skepticism until the following part.
At the end, winnowing occurs. Winnowing is “the rose ceremony” for pitching, in which a consensus must be reached for the ideas kept. These ideas need to be applicable to a broad range, which also means “boring” or low hanging fruit ideas. The winning ideas need to receive massive buy-in from your team. These ideas could have a huge impact.
For more info on making a pitch, check out this resource.
For making a pitch to executives in particular, this resource is helpful.
Scope is often an overlooked portion. People might become excited once something sounds cool, but this leads to not realizing the hurdles which could instead be avoided. An ounce of scope equals a pound of fixing what was broke by moving too quickly.
Scope has 3 pieces to it:
- technical—technical members of the team assess data, infrastructure, and layout of possible paths to a solution, regardless of if they involve data science or not. Decide whether a solution is “good enough” and yields reasonably good results. You must define the solution and what’s “good enough” beforehand.
- non-technical—typically overlooked, but this piece is necessary. Non-technical experts, like stakeholders, managers, or experts in subject matter determine the effect, which includes both positive and negative effects, and identify what is considered a success. Effects must be analyzed that lead to success as well as failure.
- in-between—this part refers to the intersection of knowledge of the business and data science. It’s a collaboration between non-technical and technical people to determine the business impact fueled by reasonable performance of analytics. It also aids in determining possible alternative plans.
Once that occurs, documenting things could make iterations in the future easier. Proper documentation could streamline the process of scoping in the future and set a precedent.
4. Making a Plan
The next part of the process is making a plan. Plotting aids in determining possible paths and then choosing the most logical one. Plotting involves risk, benefit, and cost. To what degree is the unknownness? What’s the true estimate? What accuracy does it have? This includes determining the interdependencies between possible impact and projects.
This part is where your project portfolio really starts to take its form. You’re determining the possible roadmaps between the different projects and defining single projects that are their own portfolio. Paths have cumulative benefits and cumulative costs. Remember the dependencies between projects from the Scoping step? These dependencies may help reduce overall costs.
You need to balance the lowered cost of working on parallel projects with the possibility of five projects being at risk if a problem occurs with them. It also adds to the work necessary to monitor and maintain them. This is a crucial part of the planning process.
Additional documentation occurs at this step to help facilitate ease in future iterations. These roadmaps need to include areas to turn and descriptions to know how your progress.
The future will occur. Go through these steps to help you build in blueprints of how to handle things when that day arrives. We’re at the last piece of the puzzle for your portfolio in collaborating with shareholders and teams of data science professionals. Following these steps each quarter can aid you in building a well-defined strategy of data science to yield a practical impact.