Short answer
An employee interview study is not anonymous if someone reads the individual answers, and participants should know that before they start. Decide in advance:
- who will read the answers, and who will see only findings;
- what participants are told;
- what an AI tool processes, and where;
- how long answers are kept;
- how someone can stop or ask for deletion.
Write those decisions into the invitation in plain language. Confidential is not the same as anonymous. A study that promises anonymity it cannot deliver damages trust in the next one.
Confidential, anonymous and identifiable
| Term | What it means in practice | Typical use |
|---|---|---|
| Anonymous | Nobody, including the organiser, can link an answer to a person | Large surveys that suppress small groups |
| Confidential | The organiser can see who said what but limits who else can | Interviews read by a researcher or consultant |
| Identifiable | Answers are shared with names attached | Rarely appropriate for employee research |
Interviews are usually confidential, not anonymous. The person reading the answers knows, or can work out, who said what, especially in a small team. The anonymity threshold guide explains why even aggregated results can identify people. If a study does need to be anonymous, start with the ICO's introduction to anonymisation.
Seven decisions to make before inviting anyone
| Decision | Write down |
|---|---|
| Who reads individual answers | A named role, such as the HR lead or the consulting team |
| Who sees findings only | Managers, the steering committee, the sponsor |
| What participants are told | That their answers are read, by whom, and for what purpose |
| What the AI processes | Answers and shared documents; whether contact details are included |
| Where processing happens | The region where the AI runs, and any transfer outside it |
| How long answers are kept | A period, and what happens at the end of it |
| How to stop or withdraw | How participants can stop, and how to ask for deletion |
The GOV.UK guidance on managing participant privacy recommends telling participants as early as possible what data you will collect and how you will use it. It also recommends a review by a data protection adviser, which might include a data protection impact assessment.
Tell participants plainly
A good privacy note for an interview study fits in a few sentences. It says:
- who is asking, and why;
- who will read their answers, and who will see only a summary;
- that an AI tool asks the questions and summarises the answers, and where it runs;
- how long answers are kept;
- that they can stop at any time and ask for their answers to be deleted.
Avoid "your answers are anonymous" unless they are. Avoid "only aggregate results will be shared" if quotes will appear in the findings. Say instead that quotes may be used, attributed to a role rather than a name.
Employment adds its own considerations
Research inside an organisation happens within an employment relationship, so people may worry about what their manager will learn. The ICO's guidance on monitoring workers is about monitoring rather than research. Its emphasis on transparency still applies: tell people what is collected, and use it only for the stated purpose. The ICO's guidance on AI and data protection covers using AI tools with personal data. If answers will be processed outside your country, read the ICO's international transfer guidance.
Three rules help in practice:
- Do not use research answers for decisions about individuals. If an interview reveals a conduct issue, handle it through the proper process, with the person's knowledge, not by quoting the study.
- Keep managers out of individual answers, unless participants were told otherwise.
- Ask about work, not about named colleagues. Topics should not invite people to report on each other.
A fictional example: a careful invitation
A fictional logistics company plans a study on why new hires leave in their first month. The HR lead writes the invitation:
“We're trying to understand what the first weeks here are really like. You'll have a short conversation with an AI interviewer, on any device. I (Priya, HR) will read your answers; your manager won't. We'll share what we learn with the operations team as themes and short quotes, attributed to a role, never a name. Your answers are kept for 90 days and then deleted. You can stop at any time, and you can ask us to delete your answers.”
Every sentence describes something the company can do. Nothing is promised that it cannot keep.
Where Lontra fits
Lontra is not anonymous, and it says so: participants are told before they start that the organiser reads their answers. Other participants never see each other's answers, and email addresses are never sent to the AI. You choose for each study whether the AI runs in Europe or the United States, with no fallback to another region or provider.
Studies are kept for 90 days on the Free plan and up to two years on a paid plan, then deleted for good, shared documents included. Participants can stop at any time, stop follow-ups from any email, and ask for their answers to be deleted. Every support access by the Lontra team is logged. SOC 2 and ISO 27001 audits are under way. Trust has the detail.
What you tell participants about who else sees the findings, and what you do with them, is your decision and your responsibility.
Decide, then write it down
The privacy decisions take half an hour before a study and save a great deal of trust afterwards. Write them into the invitation and keep to them, and the next study will be easier to run.
Frequently asked questions
- Are AI employee interviews anonymous?
- Not usually. If an organiser can read individual answers, the study is confidential rather than anonymous, and participants should be told so before they start.
- Should managers see their team's interview answers?
- Generally not, unless participants were told so in advance. Managers can receive findings and role-level quotes, and act on the work conditions they describe.
- How long should interview answers be kept?
- For a stated period that matches the purpose of the study, then deleted. Tell participants the period in the invitation, and include shared documents in the deletion.