Multi-signal Intelligence™ looks across the data you already hold in Culture Amp, such as survey responses and employee data, to find the groups in your organization that may need attention. It shows you three types of insight:

The two turnover risk insights are predictions about what may happen. A survey hotspot describes how a group answered your survey.

Each insight is an important clue. It draws on patterns across your data to show a group that you should pay attention to. AI Coach then helps explore why that group may need attention by looking across your survey results and Culture Amp's research and, if needed, asking you for more context.

Coach will:

Multi-signal Intelligence™ is in early access, so some of Coach's behavior may change as we refine it. For setup, see Multi-signal Intelligence™ (MSI) (Early Access). For how AI Coach works more broadly, see The Science Behind AI Coach.

Three principles shape how Coach reads an insight, explains it, and recommends what to do.

When a group is small, a few responses can have a large impact on the group's scores. This means that its results are often less reliable on their own. Coach is designed to:

When the data alone doesn't explain a pattern, Coach may instead suggest actions such as a stay conversation or a check-in with the group's managers. These conversations can confirm what the data suggests or surface what it can't.

Coach frames risk in terms of conditions you can change, such as growth opportunities, workload, role clarity, and recognition, rather than attributing it to the people in a group or their manager.

This approach draws on research into why people stay and leave, including Job Embeddedness and the Job Demands-Resources model (Mitchell et al., 2001; Demerouti et al., 2001). These frameworks help Coach focus suggestions on underlying conditions, such as people's connections at work, career growth, and access to support.

For example, Coach may read a recent manager change as a disruption to trust, and as an event that can prompt people to reconsider whether to stay (Graen & Uhl-Bien, 1995; Lee & Mitchell, 1994). It may suggest resetting expectations, clarifying role paths, and holding regular 1-on-1s.

An insight only sees what's available in your data. You may hold additional important context needed to bring color to that insight, for example a restructure or another recent event. Coach may ask what you're seeing and use your answer to confirm its explanation, or look in a different direction.

For example, Coach may point to lower survey scores on workload, compared with your organization overall, as one likely reason behind a group's risk. You might explain that the group was covering two open roles when the survey ran, and both have since been filled. Coach can then weigh whether workload is still a concern, and consider what else the evidence points to.

You'll see these principles at each step, starting with how Coach reads an insight.

Insights use the latest data in your account, and are updated each time you start a new Coach conversation. Turnover risk insights come from a predictive model, and survey hotspots from a statistical comparison of your survey results.

For the method behind each one, see MSI Insight - Turnover Risk, MSI Insight - High-Performer Turnover Risk and MSI Insight - Survey Hotspot.

When you open an insight tile, or ask Coach about a group, it starts with what the data says before offering any explanation. It names the group, says what kind of signal it is, and points out any limiting factors, such as a small group size. You can ask Coach follow-up questions or look more closely at one part of the group as you go.

A turnover risk insight also comes with predictors: the signals Multi-signal Intelligence™ found that most set the group apart from the rest of your organization. For example, these might include recent manager changes, colleagues on the same team leaving, a drop in intent to stay since the last survey, or a lower score on a particular survey question.

Next, Coach looks for supporting evidence:

Importantly, an insight and your survey results won't always describe exactly the same people. Insights use your current employee data, while survey results reflect your demographics at the time the survey ran. If a team has been renamed, merged or restructured since then, Coach may not find survey results for it, and it relies more on research and on the context you provide.

When you're ready to act, Coach helps you get started, even if the picture isn't complete. Where the evidence is still thin, it may suggest a first step to learn more, such as a stay conversation or a check-in with the group's managers. As the evidence gets clearer, its suggestions get more specific. They draw on what you've looked at together and on Culture Amp's research into engagement and retention.

Coach is designed to make its suggestions:

We believe that actions that ask more than someone can take on are unlikely to succeed. This means that Coach suggests actions that you, or the people you work with, are able to carry out.

Coach points you to a few steps that are most likely to make a difference.

Coach can look at which survey questions are most closely tied to intent to stay or engagement for the group, and suggest actions in those areas.

For turnover risk, our research suggests that actions that improve people's day-to-day work experience may be more effective than one-off financial fixes. This may include career conversations, clearer role paths, and stronger manager support.

We recommend treating fair pay as foundational. Beyond that, additional money is less likely to reduce turnover risk when other work conditions are driving it. A retention bonus may help when compensation is part of the problem, but on its own it may not address the reasons someone is considering leaving.

When the insight highlights a focus area that can have a wide impact (for example having defined career paths), Coach may suggest fixing it more widely, not only for the group identified in the insight.

A fix limited to one group can look like favoritism to other teams, and their morale and intent to stay can drop.

Lastly, when you've chosen where you'd like to take action, Coach can turn its suggestions into a plan or a brief for the people who'll act on them, or add a talking point to one of your existing 1-on-1s. Coach drafts the wording, and nothing is added until you approve it.

Example: An HR admin opens a survey hotspot for a group with lower engagement. For this group, engagement is most closely tied to whether leaders communicate a motivating vision. The admin asks: "Give me three actions to recommend to this group's leader. We have a limited budget." Coach suggests three actions focused on how the leader shares direction with the group, chosen to fit that budget, and drafts a short brief for the admin's conversation with the leader.

Coach helps you plan action for a group, but some decisions sit outside its role. It won't:

Coach's suggestions are designed for groups, like the insights they come from. If you ask about a specific person, such as whether someone is likely to leave, Coach is designed not to speculate.

Coach is designed to explain a group's results through what its members experience at work, not who they are. You can ask about any group you have access to. But if Coach is asked a question grounded in a stereotype or generalization about a group, such as "Isn't this just because younger employees don't stay long?", it's designed to redirect the conversation to what the group's results show, such as lower scores on career growth or recognition.

Coach can explain what an insight means and which predictors sit behind it. If you ask how turnover risk is calculated, it points you to MSI Insight - Turnover Risk rather than describing the model itself.

What you know is part of the evidence. The data can't show events like an upcoming restructure or something specific to the team, and your answers help Coach confirm or adjust its explanation before it suggests what to do.

When the evidence doesn't fully explain an insight, a conversation is often the most useful next step. Stay conversations, or check-ins with the group's managers, can surface what's behind the risk, and what you learn shapes which action makes sense. When the evidence is stronger, Coach suggests more targeted actions.

No. An insight describes a group, and its risk is an average across that group, so it can't tell you about any one person. Coach helps you understand the group's conditions and plan steps that apply across it. If you're concerned about a particular person, a direct conversation with them is the place to start.

Coach explains what the insight means and shows the predictors behind it, but it doesn't go into the details of the model. For how turnover risk is calculated, see MSI Insight - Turnover Risk.

Coach can add a talking point to an existing 1-on-1 you're part of, once you've approved the wording. It can also help you draft a plan or a brief for managers. It doesn't create action plans or goals in the platform for you.

💬 Need help? Open a Support Conversation to speak with our Product Support team.