Hey Juno

How can interviews improve community consultation research?

By Hey Juno · Published · Updated

Community consultation research works best when it combines the formal route (submissions, surveys, public meetings) with conversations that let people explain themselves in their own words and in their own time. Conversations reach people who will not write a submission, they show why a view is held, and they can be reported faithfully when every claim traces back to what someone said. They do not replace formal consultation, and they do not make a sample representative on their own.

This guide is for councils, government bodies, non-profits and member organisations running public engagement research. It covers what submissions and surveys miss, how to reach under-represented groups, how to avoid consultation fatigue, what representativeness can and cannot mean, and how to report back and use findings in policy.

Why conversations reach people that submissions and surveys miss

A written submission asks a lot. The person has to know the consultation exists, understand the proposal, find time, and write something worth reading. The people who do this are often the ones with the most time, the strongest views or the most experience of the process. A survey lowers the effort, but it asks people to pick from answers someone else chose.

Privacy matters too. G+S Communications used Juno, the AI researcher from Hey Juno, to run 200 interviews for Climate Sentiment in Australia 2025, and reports that participants shared things in private that they would never say in a public forum. That is one study, not a rule, but it suggests private conversation is a different channel from a town hall, not a weaker copy. See the case study.

A conversation asks less of the person and more of the interviewer. Someone can talk through the last time they tried to use a service, say what got in the way, and be asked a natural follow-up. Interviews done by voice or text, at a time of the person's choosing, remove the need to attend a meeting or to write well.

Conversations also surface the reason behind a view. A survey can tell you that most respondents are unhappy with something. It cannot tell you whether that is about cost, trust, habit or a past bad experience. For policy work, the reason usually decides what you do next. For more on the difference, see AI interviews versus conversational surveys.

Reaching under-represented groups

Under-representation in consultation usually has practical causes. The form is not in the person's language. The meeting is at a time they work. The process assumes digital access, confidence with writing or trust that anything will change. These are barriers you can often name and plan around.

Some practical steps:

  • Go to where people already are. Share an invite link through the organisations, community leaders, services or member lists that people already trust, rather than expecting them to find your consultation page.

  • Interview in the participant's language, by voice or text. Nobody has to write or speak in a second language to be heard, and participants join from a browser with nothing to download. Confirm the languages and modes you need while you brief the study, because support varies.

  • Use per-person links when you hold a list. If you already have a membership or service list, individual invite links can carry context about each person, so the interview does not ask for what you already know.

Hey Juno's participant standards cover consent, incentives and safeguarding, and the safety standard sets out how wellbeing is handled during an interview. If your topic is sensitive or your participants may be vulnerable, review both before launch and check whether this method fits. See also participant wellbeing in AI research.

Consultation fatigue

People get tired of being asked. Fatigue tends to set in when people see no result, are asked the same thing by several bodies, or are given a long form for a small decision. Conversation research can help, but only if it is used with care.

  • Ask less often, and say what for. Tell people the decision the research feeds, and when it will be made.

  • Close the loop. The most effective answer to fatigue is a visible result. Tell participants what you heard and what changed, or what did not and why.

The International Association for Public Participation (IAP2) describes a spectrum of public participation that runs from informing the public, through consulting, involving and collaborating, to empowering it (IAP2 Spectrum; IAP2 is currently reviewing it). Be clear which level you are offering. Asking people for views on a decision that has already been made is a common source of fatigue and distrust.

Representativeness, and its limits

Be honest about what a set of interviews can show. A well-designed consultation can tell you what people said, which views recur, where they conflict, and which groups described different barriers. It does not, by itself, tell you what proportion of a whole population holds a view.

Three points help:

  1. Who took part depends on how you invited them. An invite link shared with a member list reaches members who open it. A panel recruit reaches a different group. Say which you used and report it alongside the findings.

  2. Volume is not the same as coverage. Hundreds of interviews can still be thin on one ward, age group or language. Juno states where evidence is strong, limited or uncertain, and it is worth reporting the gaps as plainly as the themes.

  3. Qualitative and representative methods answer different questions. The OECD's Guidelines for Citizen Participation Processes (2022) list public consultation as one method and representative deliberative processes as another. In the latter, randomly selected people form a microcosm of the community. If your decision needs that kind of legitimacy, use a method designed for it, and use interviews to understand the why.

A sound approach is to treat conversational research as one input among several. It can bring quieter voices into the evidence without claiming that one study replaces every other form of engagement.

Reporting back what people said, faithfully

The reporting stage is where public trust is won or lost. Four habits protect it.

  • Keep minority views visible. A view held by few people can still matter for the decision. Report it, with the number of people behind it.

  • Tie every claim to evidence. In Juno, each finding links to the participant, the quote and the transcript moment, and to the audio where available, so a reviewer or oversight body can inspect how a conclusion was reached. See checking AI research findings.

  • Separate what was said from what you infer. Say plainly which is which.

  • State the limits. Say who took part, who did not, and how firm each finding is. Provisional and honest beats confident and wrong.

Juno does not redact transcripts today. Automatic redaction of hard identifiers such as names, emails and phone numbers is planned but not yet switched on, so treat transcripts and exports as containing personal information. Access is limited to people in your organisation who can see the study, and participants agree to recording before a voice interview starts. Collect less, and share findings rather than raw transcripts. See PII in customer research. Confirm the controls with your own privacy team before collecting sensitive material.

Using findings in policy

Findings are most useful when they are attached to a decision. The Arthritis Movement ran a national Juno study with 239 participants to ground its advocacy in what people living with arthritis described. Juno distilled the themes into five key problem statements that now guide its advocacy. The details are in the case study.

Some ways to make findings usable:

  • Brief on the decision, not the topic. "What will make people use this service?" gives a sharper study than "views on the service".

  • Keep the evidence available. Decision-makers, reviewers and the public may ask how you know. A trail from claim to quote answers that.

Hey Juno's institutions page lists the public sector uses it is built for, and the methodology standard explains how studies are designed and checked for leading wording.

Where this doesn't apply

  • Statutory consultation. If the law requires a specific process, that process comes first. Check with your own legal team.

  • Decisions that need a statistically representative sample. Use a method designed for it.

  • Very sensitive topics and vulnerable groups. Requires a scope review. Another method or a mixed approach may suit better.

  • When the decision is already made. Consultation without a real choice damages trust.

Frequently asked questions

Does conversation research replace public consultation?

No. It adds a channel. Submissions, meetings and surveys each reach different people for different reasons, and conversations help you understand the reasons behind what you hear.

Can interviews be representative of a community?

Only to the extent that your invitation and sampling are. Interviews are strong on depth and on showing where views differ. Report who took part and who did not, and use a representative method when your decision needs one.

How do we keep findings credible with the public?

Keep every claim traceable to a participant, a quote and a transcript moment. State the limits, show minority views, and tell participants what happened as a result.

Can people take part in their own language?

Yes. Juno interviews people in their own language, by voice or text. Confirm the languages and accommodations you need while you brief the study.