Beyond Surveys: Using AI to Decode Employee Sentiment

60% of communicators are frustrated with their ability to track employee sentiment, and a third don't have a documented listening strategy, according to Gallagher's 2025 State of the Sector report. Michael DesRochers, founder of PoliteMail, says it doesn't have to be that way. In this sponsored Q&A, I talk to Michael about how new tools can help communicators move beyond surveys to understand what employees are really thinking and improve business outcomes.

Can you please describe what “Voice of the Employee” is?

The “employee voice” is the collective opinions, ideas and concerns of employees regarding the organization. Whether positive or negative, constructive ideas for improvement or dissatisfaction with current conditions, these opinions may vary by group or topic and can be utilized as an opportunity to improve employee engagement. The objective is to instill a culture of listening and acting upon the findings to improve work relationships, business processes, and the whole organization.

What is the value of employee listening?

While the value of an organization depends upon its business strategy and performance driven by executives, actual execution and production occur at the manager and employee level. The value of listening is in elevating these ground-level experiences up to those who control the strategic and tactical levers, enabling insights to improve decision making.

Listening may lead to important strategic and tactical adjustments, process improvements, and obstacle clearing to achieve better business results while increasing employee satisfaction and longevity. In fact, employees whose companies financially outperform others in their industry feel heard, according to a study by Forbes.

On which channels can communicators find different sentiment analytics?

The traditional listening approach in small groups is simply management observation and asking questions. Larger groups rely on compiling surveys.

However, in today’s environment, where employee groups communicate via digital channels such as Teams, it is more efficient to leverage AI sentiment analysis and opinion mining tools to simply analyze the data stream to discover how employees are thinking and feeling about different topics or events. Sentiment isn’t static, and measuring it over time gives communicators the power to spot patterns, intervene early, and strengthen culture proactively.

According to the annual PoliteMail/Ragan communicators survey, online meetings is now ranked #2 behind email, overtaking intranets, in terms of employee communications effectiveness – why do you think this is the case?

The rapid adoption of online meetings and internal social chat channels came at the start of the pandemic period and, like email, is now fully incorporated into routine business communication processes.

Personal communication is essential, and even when digitally intermediated, it is still the preferred method of communication for most people. It's more effective because it's interactive, versus reading static posts on an intranet page. Meetings allow for immediate clarification, tone interpretation, and back-and-forth dialogue. Employees can ask questions, get answers, and read body language or vocal tone—all of which reduce misunderstandings.

It’s true that Teams and Viva Engage have become very popular tools for internal communicators, but there are barriers in terms of breaking into existing flows or knowing how to change behaviors. How can internal communicators utilize these tools for corporate communications?

Employees have readily adopted these tools, which mimic the functions available in consumer social channels such as Facebook, Instagram, X and WhatsApp, but it has been a bit wild west, mostly without a clear governance structure. Multiple groups and channels overlap and spontaneously form, yet never go away, resulting in a high noise-to-signal ratio. Communicators don’t necessarily have a good way to broadcast into appropriate and well-utilized channels, and employees often don’t know where to look for what. Teams and Engage are great collaboration tools, but remain a challenge for both corporate communicators and technicians to effectively manage.

The opportunity is to leverage modern analytics tools to identify the channels that have reach, readership and engagement and which are stagnant or underutilized and may be removed.

How can employee sentiment actually be measured?

There are certain words you use to describe when you are feeling good or bad, so you know words can be utilized to determine emotional states. Professional communicators know word choice matters. They craft sentences and stories and write in a specific voice for a reason. Social media tools enable the use of emoji icons for this purpose as well, so a thumbs up or smiley face signals a positive feeling, while a thumbs down or sad face is negative.

Today, machines and specifically AI programs are trained to interpret these words, to perform natural language processing for a linguistic analysis. This starts with a classified lexicon, which are words placed on a rating scale from positive to neutral to negative. This analysis progresses using machine learning algorithms to classify traditional and emerging sentence fragment patterns on that same scale.

The result of that processing is typically a score, from 0 to 100, which indicates the sentiment of the text in question. This score could be overall, for a specific time period, within a specific group, or about a specific topic.

Today, there are programs and apps, such as Commlytics, being developed to give internal communicators the insights and analysis of employee sentiment to help tailor messaging and improve business outcomes.

What are some ways internal communicators can use data from measuring employee sentiment to align with organizational goals?

Understanding employee opinions and feelings about certain subjects helps improve business outcomes by increasing situational awareness and enabling actions. Issues and opinions raised can be addressed before they become bigger problems. Programs can be designed in real time to improve employee understanding, change processes, and address concerns, resulting in more engaged and satisfied employees who perform better.

Often, companies institute change management programs, but don’t know until well after the program is completed whether it had the desired effect. With sentiment analysis tools, organizations can measure before, during and after the change management program runs, and make real-time changes to address issues, opinions and concerns to achieve the results they want. The before/after analysis can be utilized as an objective measure of whether the change campaign worked, well ahead of production results, because you have to change minds before you change outcomes.

Acquisitions and integrations are ripe with challenges, most of them being either psychological or process-oriented. Here again, measuring sentiment and mining opinions, at the group and top level, helps companies address these head-on and up-front, before they fester and drain value from the deal.

How should communicators think about sentiment analysis in the context of existing employee listening tools (e.g., annual engagement surveys, pulse surveys, etc.)?

What people say in answer to a survey question may be quite different from what they are willing to post between colleagues. Surveys take time to execute, analyze and report.

While engagement surveys and pulse surveys provide structured feedback, sentiment analysis adds depth, emotion, and immediacy, helping communicators tell a more complete story and respond with greater precision. Mining employee opinions directly from communication data streams is very valuable as issues can be raised and addressed, often leading to higher employee satisfaction and engagement. AI sentiment analysis tooling is an essential component of any modern employee listening program.