Think about the last employee survey you filled out. How long did you spend on it? Did you read every question carefully, or did you start clicking through on autopilot somewhere around question 15? Did you write what you actually felt, or what felt safe to write in a box that your manager might eventually see?
Now think about the last real conversation you had with a colleague about how work was going. Not a status update. An actual conversation. The kind where someone leans back and says, "Honestly? I'm struggling with this project." The useful question is what helped you explain the situation. A conversation can offer room for follow-up; writing can offer time to think. Neither format guarantees an honest or complete account.
What question are you trying to answer?
For an HR team helping managers understand working preferences, start with the task. A standardized survey can collect answers to the same questions across a group. A conversation or written reflection can explore a particular example: what context someone needs before a meeting, or how they prefer to discuss feedback.
Don't confuse the format with the quality of the question. Both spoken and written answers can leave important context unstated. Follow up with “Can you give an example?” and “What would help next time?” rather than assuming you already understand.
Lattice's survey guide describes both rating-scale and open-ended questions. Surveys aren't limited to numerical ratings, and a voice interview needn't replace them. Lattice's survey guide
Ask what a pause means rather than interpreting it
A pause, a change in pitch or a faster answer is not enough to establish someone's feelings or intent. If someone hesitates before saying their relationship with a manager is good, ask whether they want to add context. Don't turn the hesitation into a hidden negative rating.
The same applies to AI-generated interpretations. A plausible explanation of someone's tone is not evidence that the explanation is correct. This article does not establish that AlignWithMe infers emotion, detects concealed meaning or evaluates employees from vocal delivery.
A conversation about one working preference
Choose one situation rather than trying to produce a complete personality description. For example, a manager could ask:
- When you need to make a decision, what information helps?
- Can you describe a recent request that was easy to respond to?
- What should a colleague do differently next time?
Offer the same prompts in writing when that suits the person better. The aim is to gather a concrete example and a preference they recognize, not to claim that a fixed number of minutes replaces a survey.
A Personal User Guide can document the preference. Treat the wording as a draft until the person confirms it. This is a recommended practice, not a claim that every product automatically provides review, editing or sharing controls.
The Trust Equation
One legitimate concern with voice interviews is trust. If an AI is analyzing my voice, what's it doing with that data? Who sees it? Can it be used against me?
These are valid questions, and any company deploying voice AI needs to answer them directly. The critical factors are transparency, consent, and control. People should know exactly what's being analyzed and what isn't. They should actively consent to the process, not have it buried in a terms-of-service document nobody reads. And they should have control over what gets shared and with whom.
Ask the person which format they prefer, explain the purpose, and provide a way to correct the resulting summary. Don't promise anonymity, access restrictions or deletion controls without checking the system and process you actually use.
Review a summary before using it
If your workflow uses AI to draft a summary, compare it with the person's actual answers. Remove inferred motives, unsupported labels and details irrelevant to the stated purpose. Then ask the person whether the proposed wording is accurate and appropriate to share.
For example, “I appreciate the question and deadline in writing before a call” is a practical preference if the person confirms it. “This person avoids conflict” is an interpretation that the same answer does not establish.
This workflow makes no claim of department-level burnout detection, real-time team-health monitoring or superiority to a human analyst.
Choose the format that fits the task
Use standardized questions when consistent measurement is the purpose. Use a spoken or written follow-up when you need to understand a particular work situation. Some people will want time to compose a response; others will want to discuss it.
For a manager conversation, offer that choice where feasible. Ask the same practical questions, preserve the context of the answers, and agree on what will happen next.
Finish with an agreed next step
If someone says they need preparation time before feedback, ask what that would look like in practice. Agree on a format, try it, and revisit it together.
That is a useful outcome for either a conversation or a written reflection: a working preference the person recognizes and a next action both people understand.
References
- Lattice: Employee Engagement Surveys: What to Ask and How to Run One. Supports the distinction between rating-scale and open-ended survey questions; not a claim of voice-analysis efficacy.
Editorial correction, September 7, 2026: removed unsupported vocal-inference, format-superiority and product-outcome claims. The original publication date and URL are preserved. Conversation prompts are editorial suggestions.