Politecnico di Milano delivers actionable workforce sentiment insights with Microsoft Graph Data Connect

Generating unbiased employee sentiment data at scale

Politecnico di Milano is one of the largest science and technology universities in Italy. It focuses on delivering top-quality, innovative curricula in engineering, architecture, and design. Graduates are expected to drive innovation in the workplace, and they often begin that journey while completing their education. Such is the case for the research team led by Professor Matteo Matteucci in the Department of Electronics Information and Bioengineering at Politecnico di Milano. Matteucci, an expert in data analysis and AI, has tasked his research staff with using technology to better understand private-sector employee engagement and satisfaction.

Traditionally, in-person or digital surveys are the gold standard when companies want to understand their workforces. Politecnico di Milano Professor Leonardo Di Perna recently began looking for a way to use employee collaboration and communications data to reveal previously undetectable insights. “Every organization has unique pain points regarding employee satisfaction,” says Di Perna. “They get subjective replies through surveys, but we want to supplement that conversation with more objective data.”

During his search for a more objective source of sentiment data, Di Perna discovered Microsoft Graph Data Connect, which provides intelligent access to Microsoft 365 data at scale. This data includes insights into the ways workers communicate, collaborate, and manage their time across all the applications and services in Microsoft 365. It was precisely the type of information Di Perna and Matteucci were looking for. “Businesses need a workforce engagement sensor,” says Matteucci. “The one that we’re developing with Microsoft Graph Data Connect is objective and works in tandem with and as a proxy for the costly survey-based standards of today.”

Diverse ways of working demand diverse sentiment insights

The shift to remote and hybrid work in the last few years has changed how employees interact with one another. For businesses across many sectors, this transformation has come with new trials, including the need to better understand how shifts in work styles alter employee sentiment across a broad range of scenarios and settings. By developing and sharing easily digestible, objective key performance indicators (KPIs) for employee sentiment, businesses have more opportunities to assess and evaluate trends. “If you detect that employees are feeling more alone in their work, there are thousands of potential reasons for that sentiment,” says Di Perna. “By detecting and tracking digital traces linked to employee loneliness, you can uncover the factors actually linked to that feeling at your company.” 

Even in its early stages, the project has been working with diverse customer organizations, including a bank, an industrial manufacturer, and a leading company in the services sector. The bank wants to increase engagement across all employment levels, while the manufacturer wants to understand how blue-collar and white-collar employees might better interact within a single production plant. According to Matteucci, the goal is to use de-identified data to understand sentiment-related factors that are either left out of survey results because of underlying work dynamics or that respondents themselves haven’t fully realized. 

Currently, the team at Politecnico di Milano is using surveys to benchmark the results they’re revealing through Microsoft Graph Data Connect. “The goal of the project is to create models capable of relating the digital traces we’re uncovering to our most reliable survey answers,” says Matteucci. “When we can be sure that we’re consistently discovering results correlated to the survey responses, we can largely replace surveys with the digital tool.”

Despite being the current standard, surveys have their weaknesses, Matteucci explains. “People formulate their answers based on personal beliefs, status, and a multitude of factors,” he says. “It’s also an unfortunate reality that not everyone will respond to surveys, further skewing your data.” With Microsoft Graph Data Connect, Matteucci’s team can access data like conference calls, meetings, and emails in a way that preserves privacy and anonymity, giving a clearer picture than ever of what the workforce is really feeling. And because the solution is digital, an organization’s stakeholders can easily take more frequent readings of employee sentiment without disrupting workflows. Instead of relying on monthly surveys, businesses will be able to conduct weekly sentiment research and reduce their survey-taking to an annual or semiannual basis.

A purpose-built solution designed from the ground up

Di Perna began building the initial solution after watching online videos and immersing himself in Microsoft Graph Data Connect documentation from Microsoft Learn. He initially used Azure Data Factory to store the uncovered Microsoft 365 data but has since been exploring the potential of Azure Synapse Analytics. After creating his first proof of concept, he also reached out to the Microsoft team for advice. “Microsoft has been a great help, with easily tailored templates and advice about how to use Azure Synapse Analytics to filter out data that can compromise user privacy,” says Di Perna.

After the design and refinement of the prototype solution, it was introduced to the rigors of Politecnico di Milano’s experimental evaluation cycle. “We operate a standard hypothesis, evaluation, and validation cycle,” says Matteucci. “We’re experimenting with different techniques of data interpretation to refine our models and deliver the best solution to our customers.” Newer iterations of the solution incorporate Power BI dashboards that are customized for each customer’s needs. “We’ve shown the Power BI dashboards to a couple HR stakeholders, and they really liked what they saw,” Matteucci adds.

Deeper insights on the way

Even as the solution is fine-tuned for single-company insights, Matteucci envisions broader application for it in the near future. “Companies never want their data shared, but if they can compare their aggregate results against an anonymized group of competitors or peers for benchmarking, they would be thrilled,” he says. “First, you find what percentage of your workforce is truly happy, and then you see how that measures up against others, which creates clarity.” In this example, if a company were to find that its employees are happier than the competition, it would know that its workplace satisfaction efforts are paying off. Even oft-uncertain findings like the ROI of employee engagement programs might become easier to define.

This federated view, Matteucci notes, will help organizations understand their strengths and shortcomings in relation to their peers without sharing KPIs or the outcomes of individual surveys. “Customers will better understand their human capital,” he says. “From there, they can reduce stressors or react to other factors that prompt people to seek other employment—that’s our ultimate goal.”

Di Perna agrees, noting that when employees leave the business that trained them, the losses aren’t only economic but also intellectual and social. “We have an opportunity to increase retention rates through enhanced data granularity revealed at scale,” he says. “With these insights, organizations can improve the quality of their working environments and encourage positive engagement.”

Find out more about Politecnico di Milano on Twitter, Facebook, YouTube, and LinkedIn.