Human-centred design starts with better questions. But it only works if people trust what happens next

Human-centred design is often talked about as if the magic happens in the interview.

Ask better questions. Listen properly. Follow the thread. Get beneath the obvious answer.

All true.

But the real test comes afterwards.

Because once you have the transcript, the quotes, the patterns, the emotional moments and the uncomfortable truths, you have something powerful in your hands. You have people’s lived experience of work. You have their frustrations, ambitions, blockers, motivations and fears.

That is not just “data”.

It is trust.

And what you do with that trust determines whether human-centred design becomes a force for better work, or just another extractive process dressed up in empathy.

At Freeformers, we believe organisations need to move beyond transactional HR towards something more relational. That means treating insight as a shared-value system, not a research asset to be mined. Done well, discovery helps employers design better experiences, employees feel heard and protected, and customers ultimately benefit from stronger cultures and better performance.

That balance matters.

Because if people do not believe they are safe to tell the truth, they will not tell you the truth.

And if you do not protect what they share, you probably did not deserve it in the first place.

Anonymous is not always anonymous

One of the most honest parts of the conversation was the distinction Emilie made between anonymity, anonymisation and de-identification.

They are not the same thing.

A survey might claim to be anonymous, but if someone is logged in with a work email, asked to identify their team, and chased when they have not completed it, the experience may not feel anonymous at all.

People notice that.

They notice when a system says “anonymous” while behaving like it knows exactly who they are.

That does not always mean someone is misusing the data. Many organisations and platforms have genuine safeguards in place. But perception matters. Trust matters. And in smaller or medium-sized organisations, even a few demographic details can make someone easy to identify.

So when we run interviews, we do not casually promise perfect anonymity.

We talk about de-identification. We explain what will happen to the data. We remove names, teams, manager references and colleague references. We consider whether product names, locations, roles or protected characteristics could identify someone. We ask: what can be removed without stripping the meaning from what was said?

That last part is crucial.

Because ethical insight is not about sanding everything down until it says nothing.

If someone says, “I feel like an outsider because I’m the only woman in a male-dominated environment,” removing gender may protect identity, but it may also destroy the meaning. The work is to preserve the truth while protecting the person.

That takes judgement.

It also takes care.

The point is not just compliance. It is how people feel

Yes, there are legal responsibilities. Consent matters. Data retention matters. GDPR matters. People need to know what is being recorded, why it is being recorded, how it will be used and who will see it.

But this is not just a legal conversation.

It is a human one.

Before an interview begins, the participant should understand the purpose of the conversation. They should be asked for permission to record. If they say no, that answer should be respected. No awkward pressure. No “it would be easier if…” No bot quietly joining the meeting without proper consent.

Because the goal is not just to collect cleaner data.

The goal is to create enough psychological safety for someone to speak honestly.

That does not mean pretending an interview is therapy. It does not mean over-validating every answer or steering the conversation towards what you want to hear. But it does mean remembering there is a person in front of you.

A person who may be taking a risk by telling the truth.

Less extraction. More stewardship.

That is the shift.

AI can help analysis. It cannot replace judgement

Once the interviews are complete, the tempting move is obvious.

Drop the transcripts into ChatGPT and ask it to find the themes.

You will get an answer.

It may even be broadly valid.

But it may also flatten the nuance. It may overstate a theme mentioned by two people and give it the same weight as something said by twenty. It may miss gaps. It may tidy messy human experience into categories that feel neat but are not quite true.

That is where human judgement matters.

Emilie described using a thematic analysis approach, supported by AI but not surrendered to it. That means starting with research questions, creating a codebook, testing and refining themes, extracting meaning units, challenging the outputs, and using a “red team” process to find holes in the analysis.

The point is not AI versus humans.

It is AI with human accountability.

Used well, AI can speed up the work. It can help process large volumes of qualitative data. It can challenge patterns, identify inconsistencies and support synthesis.

But it cannot understand organisational context the way a skilled human can. It cannot feel when something important is being said between the lines. It cannot know which insight is commercially useful, emotionally loaded, politically risky or strategically urgent without guidance.

Human-centred design still needs humans at the centre.

Funny, that.

Analysis tells you what happened. Synthesis tells you what to do about it

There is a moment in every discovery process where the question shifts.

You move from “what did people say?” to “so what?”

That is synthesis.

Analysis might tell you that early-career consultants ignore internal emails unless they come from someone they trust.

Synthesis asks what that means for design.

In that example, it means any learning communication sent after onboarding needs to come through a trusted person or channel. Otherwise, it will be ignored. Not because people are disengaged. Because they are busy, overwhelmed and filtering for relevance.

That is the value of qualitative insight.

It does not just tell you that something is happening. It helps you understand why it is happening, and what design choices could make the experience better.

From there, you can create outputs that actually help the organisation act: insight statements, design principles, personas, journey maps, stakeholder summaries, product requirements, user stories or value proposition canvases.

But the output should never be theatre.

A beautiful deck that changes nothing is still a failure.

The question is: what will help this organisation make a better decision?

Surveys have a role. They are just not the whole story

Surveys are not the enemy.

Used well, they are useful. They can help validate whether an insight found in interviews is widespread. They can give scale to a pain point. They can help prioritise where to act first.

But survey data alone rarely gives you the full picture.

It can tell you what people selected.

It cannot always tell you what they meant.

It cannot always reveal the contradiction, hesitation, emotion or context behind the answer. It cannot always show you that someone ignored a learning email not because they did not care, but because they receive too much organisational noise and have learned to filter ruthlessly.

That is why the best discovery blends methods.

Use interviews to understand meaning. Use surveys to test scale. Use both to design with confidence.

Evidence over optics.

Always.

Insight should become part of the operating model

Perhaps the biggest opportunity is this: human-centred design does not have to be a one-off project.

It can become part of how the organisation works.

When interviews are de-identified properly, stored responsibly and retained only for as long as they remain useful, they can form part of a living insight bank. Not a dusty archive. Not a surveillance tool. A responsible, timely source of understanding.

That allows organisations to return to existing insight and analyse it through new questions.

What are people saying about AI?

What is getting in the way of learning transfer?

Where are new starters losing confidence?

What do managers misunderstand about employee experience?

This is where the work becomes more than research.

It becomes a new operating system for people, culture and performance.

One where decisions are shaped by real experience, not assumptions. One where HR is not designing from the centre out, but from the human reality in. One where culture is not a poster on the wall, but a system of evidence, care and action.

The excuse is getting weaker

For years, human-centred design in employee experience was treated as expensive, slow or luxurious. Something big organisations could afford. Something customer-facing teams did. Something HR admired from a distance but rarely had the time, budget or confidence to adopt.

That excuse is getting weaker.

The tools are faster. The methods are more accessible. The need is more urgent.

The bigger blocker now is often belief.

Some organisations have never thought about working this way. Some assume they already know their people. Some worry they do not have time. Some fear they will uncover things they cannot fix.

But “we already know our users” may be the most dangerous sentence in design.

You are not your user.

You are not your employee.

And if you have not asked properly, listened carefully and protected what you heard, you probably do not know as much as you think.

Better questions are only the beginning

Human-centred design does start with better questions.

But it does not end there.

It depends on what happens after the question is answered. How the data is protected. How the insight is interpreted. How the organisation acts. Whether people see change, or whether they simply remember being asked.

Because asking people to tell the truth creates a responsibility.

Not to fix everything overnight.

Not to promise perfection.

But to treat their experience with respect, turn it into useful action and design systems where humanity and performance strengthen each other.

That is the work.

Less performative listening.

More responsible action.

Less data extraction.

More mutual value.

Less “we know our people”.

More “we are willing to learn”.