Are you looking for advice on how to prioritize data science projects? Well then you are in the right place! In this article we tell you everything you need to know to prioritize your data science projects in your project backlog. We start out by discussing some of the main factors that need to be considered when assessing the priority of a project. After that, we provide two examples of frameworks for prioritizing your data science projects.
Factors to consider when prioritizing data science projects
What factors should you consider when it comes time to prioritize data science projects? Here are the main factors you should consider to help you prioritize data science projects.
- Value. The first factor you should consider when deciding how to prioritize a data science project is the value that the project brings to the company. In order to assess the value of a project, you should think about the main business metrics that the project will impact and size of the impact the project will have. Will the project bring in new customers or reduce customer churn rates to increase the number of customers the company has? Will the project increase conversion rates or introduce a new monetization strategy to increase the average revenue that is made per customer? Will the project automate a laborious task or increase the productivity of a given team to reduce the company’s operational costs? These are just a few examples of ways that a project can provide value to a company.
- Resources. The next factor you should consider when deciding how to prioritize a data science project is the amount of resources that the project will require. This can include anything from the amount of time that your team will spend on the project to computational expenses that will be required to complete the project. Make sure to also consider resources that will be required from other teams in addition to resources that will be required for your own team. Resources are not infinite and a resource that is required for one project will not be able to be used for another. That means that prioritizing one project that consumes a lot of resources may block you from being able to complete multiple other projects that consume fewer resources .
- Urgency. The final factor that you should consider when prioritizing the project is the urgency of the project. There are many different factors that can impact the urgency of a project and force one project to be pushed to the front of the queue. Are your deadlines internal or external? Is there a hard deadline that needs to be met for a product launch? Do you need to have a prototype done by a certain time in order to get on another team’s roadmap? These are just a few of the questions you should consider when determining the urgency of a project.
Assessing the value of a data science project
The first factor you should consider when deciding the priority of a data science project is the value that the project provides. Here are some points you should think about when you want to assess how much value the project brings to the company.
- What business metrics will be impacted. The first thing that you should consider when you assess the value that a project will deliver is which business metrics will be affected by the project. Not all business metrics are created equal and there are often cases where some metrics should be prioritized over others. The exact metrics that are prioritized will vary depending on the company, but metrics that are directly highlighted in the company’s strategic plan will generally be prioritized over auxiliary metrics.
- How large is the impact on business metrics. In addition to considering which business metrics will be impacted by the project, you should also consider how large the impact on the business metrics is expected to be. There are cases where a project that is expected to have a very large impact on an auxiliary metric may deliver more value than a project that is expected to have a very small impact on a key business metric.
- How long will the project be useful for. Another factor to consider when evaluating the value of a project is the long term viability of a project. A project that will continue to provide value indefinitely should generally be prioritized over a project that has an expiration date on it. One example of a case where a project might not have long term viability is if the project is centered around a product that will be retired within the next few years.
- Can the results be reused by other teams. Another very important factor to consider when you are trying to assess the value that a project brings is whether the results of the project can be reused by other teams. If other teams are able to reuse the results of your project, this will have multiplicative effects on the amount of value that can be derived from your project.
- Is there an easier way to achieve a similar impact. The final factor that you should consider when you are trying to determine the value of a project is whether there is a simpler way to achieve a similar impact. In some cases there might be another team or project that could capture a large amount of the impact on business metrics while spending fewer resources. If this is the case, it may make sense to think of the true value of your project as the incremental amount of impact over what the other project could deliver.
Assessing the resources required for a data science project
The next factor you should consider when deciding how to prioritize a project is the amount of resources required for the project. Here are some points you should consider when assessing how many resources a project will require.
- Time spent by your team. The first thing that you should consider when assessing how many resources a project requires is the amount of time that your team will need to spend on the project. If your team spends a lot of time completing one large project, they will not have as much time to spend on other projects that require fewer resources. This means that time is generally the most precious resource you should consider when prioritizing products.
- Time spent by other teams. In addition to considering the amount of time required by your team, you should also consider the amount of time required for other teams. If you are working on a project that will require a lot of time to be spent by other teams, you should also consult with those teams before deciding the priority of the project to assess whether they will be able to contribute.
- Expected time spent on maintenance. In addition to considering the amount of time that it will take to build the initial project out, you should also consider the amount of time that will be required to maintain the project over time. This is not often the case, but there are some cases where a project might be expected to require a lot of maintenance that will continue to consume resources indefinitely.
- Monetary expenses. In addition to considering the amount of time that will be spent on a project, you should also consider the amount of money that will be required to complete a project. For many data science projects, the time that is spent on the project will outweigh other monetary expenses. This is not always the case though. For example, if you are training a large machine learning model on a very large dataset, there may be computational expenses that you need to consider when prioritizing the project.
Assessing the urgency of a data science project
The final factor you should consider when you are trying to prioritize your data science projects is urgency. Here are some points you should think about when assessing the urgency of a data science project.
- Internal and external deadlines. The first thing you should consider when assessing the urgency of a project is whether any deadlines you have are internal or external. External deadlines tend to carry more weight than internal deadlines because if you miss an external deadline then you might risk losing a long term client.
- Hard and soft deadlines. The next factor you should consider when assessing the urgency of a project is whether the deadlines you have are hard deadlines or soft deadlines. Hard deadlines are strict, immobile deadlines that must be met in order for the company to stay on track. For example, if the results of a data science project must be delivered by a certain day in order for a product launch that has already been publicly announced to take place, this is a hard deadline. Soft deadlines are more flexible deadlines that can be pushed back if necessary.
- Getting time on another team’s roadmap. Another thing you should consider when assessing the urgency of a project is timelines for other teams you are working with. Some teams prioritize their roadmap 3 or 6 months ahead of time. If you need to get some work on their roadmap for a given year or quarter, you may be required to have a prototype of the project that they can build off of available by a certain date.
Frameworks for prioritizing data science projects
Now that we have talked about the factors that you should consider when you are prioritizing data science projects, we will talk about some frameworks you can use to prioritize your projects. We will talk about two different frameworks that you can use. The first is a loose framework that does not require concrete numeric estimates of quantities such as the amount of value delivered. The second is a strict framework that requires more concrete estimates.
A loose framework for prioritizing data science projects
- Create a roadmap based on hard deadlines. The first thing you should do is examine the projects that have hard, non-negotiable deadlines and absolutely need to be completed by a certain date. Estimate the amount of time it will take to complete these projects and work backwards to determine the date that you need to start working on these projects. Start to outline a roadmap containing only these projects with hard deadlines. Note that not all teams will have projects with hard deadlines. If your team does not have any projects with strict deadlines, simply move on to the next step.
- Classify the remaining projects as low or high value. After you pencil in the projects with non-negotiable deadlines, you can start to fill in the rest of your roadmap with projects that have more flexible deadlines. Take each of these projects and bucket them into low value or high value. It is okay if there are projects that will deliver varying levels of value in the same bucket, you can take the varying levels of value into consideration later when you actually decide what projects to prioritize. The main goal here is just to roughly group the projects based on the value they stand to deliver.
- Classify the remaining projects as low or high resource expenditure. After you assess whether your projects are low value or high value, the next step is to look at the amount of resources that the projects require. Again, bucket the projects into a low resources or a high resources category.
- Fill in the gaps one bucket at a time. After you have bucketed your projects based on value and resources, you will have four buckets that your projects can fall into – high value and low resources, high value and high resources, low value and low resources, and low value and high resources. Now it is time to fill in the gaps in your roadmap with these projects. You should do this by looking at one bucket of projects at a time and prioritizing candidates within that bucket as you see fit. You should not move onto the next bucket until you have allocated all of the projects in the first bucket into a spot in the roadmap. You should move through the buckets in the following order – high value and low resources, high value and high resources, low value and low resources, and low value and high resources.
A strict framework for prioritizing data science projects
- Create a roadmap based on hard deadlines. Just as you did in the other framework, you should start out by creating an outline of a roadmap and filling it in with only non-negotiable projects that need to be completed by a certain date. After that, you can fill in the gaps in the roadmap with projects that have more flexible deadlines.
- Calculate a numeric score for the value of each project. The first thing you should do is calculate a numeric score that represents the value of each project. You can calculate these scores in any way you see fit, but you need to make sure that the scores are all on the same scale so that they are comparable to one another. The fastest way to do this is just assign each project a score on a scale of 1 to 10 based on how you perceive the value. If you want a more precise method, you can try translating the value that each of the projects delivers into a revenue number to estimate the impact that each project will have on the company’s revenue.
- Calculate a numeric score for the resources required for each project. Once you have calculated a numeric score representing the value delivered by each project, it is time to calculate a score for the number of resources required for each project. The easiest way to do this is to estimate the total number of hours required to complete the project then use that as the numeric score.
- Fill in the gaps based on the ratio of value to resources. Finally, you should calculate the ratio of the value produced to the resources required and use this ratio to fill in the gaps in your project roadmap. Move down the list one by one and fill in the projects that have the highest ratio of value to resources first. One caveat to keep in mind here is that high value, high resource projects are often better options than low value, low resource projects. Pay mind to this and ensure that high value, high resource projects get the attention they need.
Check out our article on data science best practices for all of our best recommendations on how to increase the efficacy of data science teams.