Research Drop: Designing AI Feedback Systems by Industry

Research Drop in Brief:

• Employees want to meaningfully shape AI transformation, yet 34% still lack opportunities to give feedback on their AI experience.
• High frequency AI users are more likely than low frequency AI users to report regular requests for input, suggesting that listening and adoption positively reinforce each other.
• AI feedback is social: Informal conversation and team meetings are the most utilized feedback channels across industries, making them critical early-signal channels for AI rollout success.

Big transformations don’t stick by default; they stick by active listening and iteration. Organizations are networks of distinct communities and workflows, so a one‑size‑fits‑all change rollout rarely lands. The fastest way to de‑risk change is to give employees real ways to shape it: invite feedback early, often, and in the flow of work. Research shows that employee voice and involvement are significant drivers of change readiness and commitment to change – and their resistance and cynicism around the change is reduced¹.

The opportunity for listening in AI transformation is vast. In our research, 66% of employees feel they have adequate opportunities to share feedback on their AI experience and specifically among individual contributors (ICs), that drops to 52%. Increasing feedback opportunities matter as high frequency AI users (use AI daily) are more likely than low frequency AI users (use AI monthly) to report regular requests for input (e.g., company‑wide surveys, in‑product prompts), signaling that listening and adoption positively reinforce each other.

Integrating employee feedback can be a multiplier on change ROI. When organizations listen across channels and act on what they hear, they can accelerate AI from pilot programs to organizational value at scale. This level of iteration creates an agile change process that can adapt to the consistent technology and culture changes associated with AI transformation.

This month’s Research Drop dives into building an AI feedback system for transformation and outlines various feedback channels to explore. We’ll also map industry differences in preferred feedback channels to learn how different industries are meeting employees where they are in the flow of work. Because integrating AI tools is fundamentally a behavioral change², not just a technological one, context matters: the right channel in one industry may be the wrong one in another.

High frequency AI users have a feedback advantage

When you involve employees in change, you signal their experience matters and earn buy‑in as the transformation unfolds. The question is: how well are organizations actually doing this?

The answer depends on who you ask. Seventy-five percent of executives say their company is employee‑centric, yet only 23% of individual contributors (ICs) agree³.  In our 2025 Agentic Teaming & Trust Research Report, this leader-IC gap shows up in enablement too: 78% of leaders report feeling supported to integrate AI agents into their work, compared to 50% of ICs. Closing this gap is critical, as employee‑centric organizations were found to be 7x more likely to succeed with AI³.

A practical way to narrow that gap is to build a positive feedback loop. The more avenues employees have to share their experience, the more involved they feel, and the faster teams can iterate toward fit. In our data, high frequency AI users are markedly more likely than low frequency users to report receiving:

• Quarterly company‑wide surveys on AI (93% vs. 71%)
• Monthly pulse surveys on AI (83% vs. 50%)
• Weekly in‑product feedback requests (53% vs. 23%)

This is how the flywheel turns: listen → act → adoption → deeper listening. By asking people to reflect on what’s working (and what isn’t), organizations uncover friction early, improve the rollout, and strengthen the employee experience, especially for ICs who often feel least supported in agentic AI integration.

Listening where work happens: Industry patterns in AI feedback

Workflow redesign – and the value AI ultimately delivers – will look different across workplaces and industries. Roles, risk profiles, and communication norms vary widely, so the feedback system that informs change should vary, too. Think desk‑based flows and scheduled meetings in one sector versus shift‑based, patient‑ or customer‑facing work in another.

Depending on existing workflows, different industries lean on different ways of gathering employee input. Some rely more on traditional online surveys, while others surface feedback organically through standups, team huddles, or informal conversations. The point: there’s no one ‘right’ channel – what matters is building consistent, reliable ways for employees to signal what’s working and what isn’t. To understand how this plays out across industries in our sample, we examined industry-level patterns in feedback channel usage. In our data, most industries cluster around two‑thirds of employees who feel they have adequate opportunities to give feedback on their AI experience. By contrast, healthcare sits closer to half, underscoring the need for innovative channels tailored to shift‑based routines, clinical workflows, and higher privacy constraints.

Differences by industry likely reflect frontline mix, regulatory burden, and maturity of AI transformation as some sectors simply started experimenting earlier and have more established listening practices.

So, what changes by industry? It’s not just the cadence; it’s the mix of channels that meet employees where work actually happens.

We found differences in top‑utilized AI feedback channels across industries. The headline: AI feedback is social. Informal conversations and team meetings rank #1 and #2 nearly everywhere – teams process change together, and quick back‑and‑forths create the fastest path to signal and iteration.

Other interesting patterns:

• Internal social forums are popular in construction, healthcare, and transportation/travel/hospitality. These are frontline‑heavy sectors where mobile‑friendly, non‑desk access makes forums easy to reach between tasks or at shift hand‑offs.
• Pulse surveys show up strongly in manufacturing and transportation/travel/hospitality. A potential explanation is that short, targeted check‑ins fit shift schedules, can be team specific, and support rapid action loops without burden.
• In‑product feedback tools are common in construction, financial & professional services, retail/food/beverage, and technology. Their workflows are often anchored in specific applications that make embedded prompts at the point of experience both relevant and efficient.

Different work, different ways to listen. Build a channel mix employees will actually use, one that is fast, familiar, and in the flow. This meets people where they work so feedback follows. Keep it easy, visible, and routine; these small, habitual touchpoints beat big, occasional asks.

Make employee feedback the engine of AI change

Change succeeds when we understand employees and their experience. Providing listening opportunities also enables employees to lean into the change by feeling involved and actively participating in the change. This process also helps change leaders learn in real time where AI initiatives are landing versus where they’re struggling.

Our data shows that feedback is social – so it’s time to build systems around social feedback. Treat hallway conversations, team huddles, and quick chats as consistent signal:

• Capture it in the moment with lightweight prompts (e.g., “what’s one friction point you’ve had with this tool today?”) or notes tied to the task/tool.
• Route it automatically to the owners of AI transformation (product, IT, HR) with the context needed to act.
• Close the loop visibly so employees see their input become decisions.

Change moves faster when listening is woven into how work gets done. Surveys and pulses provide the backbone of organizational memory, while social feedback in meetings and everyday conversations adds the nuance that keeps AI transformation grounded in reality. Keep those signals connected in the flow of work, and the transformation becomes something employees actively shape, not something that just happens to them.

Stay tuned for our February Research Drop to keep up with what the Microsoft People Science team is learning! 

This month’s Research Drop analyzed 1,800 global employees from the Microsoft People Science Agentic Teaming & Trust Survey from July 2025. Leaders, managers, and individual contributors were represented across the industries included in this blog: Construction, Financial & Professional Services, Healthcare, Manufacturing, Retail, Food, & Beverage, Technology, and Transportation, Travel, & Hospitality.

¹Potnuru, R. K. G., Sharma, R., & Sahoo, C. K. (2021). Employee voice, employee involvement, and organizational change readiness: Mediating role of commitment-to-change and moderating role of transformational leadership. Business Perspectives and Research, 11(3), 355-371.

²De Cremer, D., Schweitzer, S., McGuire, J. J., & Narayanan, D. (November 19, 2025). How behavioral science can improve the return on AI investments. Harvard Business Review.

³Lovich, D., Meier, S., & Taylor, C. (November 26, 2025). Leaders assume employees are excited about AI: They’re wrong. Harvard Business Review.