3 Critical Aspects of Effective Stakeholder Management For Data Science Projects

Originally published here

Can’t forget that we still have to work with people

What does it take to do data science successfully? Data, of course. But what else? Great tools and software packages can get you only so far in the quest for analytical insights. In addition to having a great set of skills, you need an understanding of how to work with people — whether they are external stakeholders or your own team members. This blog post discusses three critical aspects that are crucial for effective stakeholder management in any data science project: expectations management, establishing clear communication channels, facilitating collaboration and feedback loops.

Why is it so easy to forget that it’s about the people?

For most of us data scientists, we enter into the industry for our love for data and creating products and services that provide value for our customers. The data, the models, and algorithms are all about people — who they are, what desires they have. We want to know why someone buys one product instead of another or when in their life cycle a certain behavior emerges.

In short, we’re not just talking about numbers here: human beings matter! And that always has been true for data science projects as well. It’s easy to forget this because it takes more time than simply analyzing tables and plots on a computer screen if you do things right from the start. But don’t be fooled by thinking that stakeholders can wait until later in your project before you work with them; most likely they’ll end up waiting around indefinitely while other organizations find success ahead of yours due to good stakeholder management.

Expectations management

Not all data science projects go the way we planned. In fact, data science projects are notoriously difficult for meeting deadlines and being on budget. This is not necessarily because of a lack of skill; rather it’s due to the complexity that arises from trying to understand human behavior, in all its nuances and variety.

Some data scientists think they can gloss over this by promising stakeholders — external or internal — what will be delivered at specific time points with certainty. But as we’ve seen, there is no way to control every variable when understanding how people behave around data-driven products and services you’re designing. And so often these promises end up unfulfilled (and sometimes resented) which means your organization has wasted precious resources through poor stakeholder management early in the process!

One approach for setting expectations is to have data scientists present a range of potential outcomes. They can then show stakeholders the data and models they used in their calculations as well so that there’s transparency and understanding about why certain expectations are likely more realistic than others.

There are many places where data science projects go wrong outside of simply not meeting deadlines or cost estimates: for example projects may take longer because data needs cleaning before it can be analyzed; predictions based on data might end up being inaccurate due to unforeseen external events like natural disasters disrupting supply chains (e.g., pandemic). It’s important for both sides — the stakeholder(s) and you as a data scientist — to maintain open communication throughout these difficult moments instead of letting them become time bombs waiting to explode.

In data science, no project is perfect — but stakeholders can still be happy with the outcome if expectations are managed properly from the start!

Establishing clear communication channels

Of course, it intuitively makes sense that data scientists need to communicate effectively with data stakeholders. But data science projects often have many different stakeholders, who don’t always work in the same department let alone varying requirements and expectations. For example, it’s likely that us data scientists need to collaborate with product managers and engineers on data-driven products or services they’re designing for internal and external customers.

The more people involved in data analytics efforts — both within your organization and outside of it — the more important communication becomes. Indeed, building strong channels throughout an enterprise may be difficult but crucial for success because data is shared among many people and data scientists need to be able to communicate what they’re doing, why they’re doing it, and how others can help.

This starts with a data scientist being clear about their goals for the project; these should be in writing so that stakeholders have an understanding of your expectations. It also includes setting up regular meetings (e.g., weekly) or other forms of communication channels like Slack or email where you are available for questions from data science team members as well as data stakeholders on all levels within your organization and even outside partners who may not use formalized communications methods regularly such as telephone conversations.

Facilitate collaboration and feedback loops

To ensure that your project adopts to data stakeholders’ needs, data scientists need to encourage collaboration with internal and external stakeholders. This means taking time out of data science projects — which are already data-driven themselves! — to invest in establishing clear feedback loops that make sure everyone is on the same page about what’s happening, and why it’s important for their team members or customers.

For example, data scientists might want to invite a stakeholder from one department (say product management) into the process as they’re exploring how data can be used within their project. Once data has been gathered, visualized, collected, analyzed etc., this may mean inviting another person who doesn’t work closely with analytics efforts such as an engineer who will use data insights from your research to inform development decisions but hasn’t been directly involved in data science.

This might sound like a lot of work, but it can be vital to an organization’s and the project’s success — and may even save you time by avoiding miscommunication or data errors due to differing interpretations from different teams across the enterprise.

It’s a soft skill that you can’t afford to ignore

The data scientist needs to make sure that stakeholders don’t have unrealistic expectations about what data scientists are able to do for them during this project. One way they can avoid unrealistic expectations is by communicating clearly with their stakeholders as well as setting up clear communication channels so there isn’t any confusion about what each stakeholder expects from this data-driven research endeavor!

Plus, data scientists should work to encourage collaboration and feedback loops with stakeholders on all levels within your organization as well as outside partners who may not use formalized communications methods. This is the soft skills that data science projects need in order to be effective!

Whether you are a data scientist or not, stakeholder management is critical to any project. We have seen too many projects fail because stakeholder management has not been done properly. If this sounds like something that interests you or if there are some questions about how to do these things in your organization, share them below!