Recently, I read a Content Design London blog post about mental models that used restaurant experiences as a metaphor: users arrive with expectations (like how to order food), and we either reflect those expectations or “rebut” them by explaining a new way.

While this analogy is helpful for illustrating basic concepts, it dramatically oversimplifies what happens when people interact with complex, high-stakes services, especially when they’re vulnerable.

Have you ever been to a McDonald’s late at night?

This isn’t about simply ordering in different ways at McDonald’s or Nando’s. McDonald’s has design teams that understand that customers interact differently throughout the day – from families at lunchtime to people coming out of clubs and bars late at night – and design the experience around that.

They ensure that menus are simpler at night, with fewer options and no confusing add-ons, for example. Even then, those pesky printers that never work when they’re supposed to give you your order number cause absolute cognitive overload no design team clearly ever imagined.

At Vapiano, the pizza and pasta chain, guests were handed a card on arrival and told to order at various counters, with everything tracked on the card. Despite clear instructions, many people avoided Vapiano because the process felt stressful and confusing.

The issue wasn’t a lack of explanation; it was the feeling of being out of control, something no sign or script could fix.

If ordering dinner can be overwhelming, imagine the stakes when someone is applying for benefits, seeking crisis support or trying to report a crime.

When the usual rules don’t apply

When I worked on the national drink spiking advice service for Police.uk, the users weren’t just browsing. They were often in distress, possibly under the influence, anxious, and afraid.

In these circumstances, traditional user research methods and “rebut and re-educate” approaches simply don’t work. We can’t ethically recreate these states for testing, and there’s no room for error. The information has to be “first-time right” because people’s safety depends on it.

Years ago, during a drama school audition, I had to play someone drunk and lost at a carnival. It taught me that, in crisis, people use all their focus just to process simple instructions. Every extra cognitive burden can mean the difference between getting help or giving up.

When language and culture collide

At Universal Credit, we tried to avoid jargon and unnecessary definitions. Yet, when we used the word “partner” in forms, user testing revealed that some thought we were asking about sexual orientation, not marital status.

Their personal and cultural mental models clashed with our assumptions, leading to confusion that could impact their eligibility for critical benefits.

This isn’t just a minor misunderstanding, but a fundamental disconnect that can have serious consequences for people already under stress.

Reading levels and real-world barriers

While redesigning Ofsted’s childcare provider registration, I discovered the guidance manual required postgraduate-level reading skills. Yet, most users were young women without university degrees, simply wanting to work with children. The result? Frustration, tears and a helpline overwhelmed with calls.

The solution wasn’t to explain the process better, but to completely reimagine it: combining five forms and a dense manual into an intuitive online journey, with help and examples exactly where users needed them.

The hidden bias in user research

Too often, user research involves people who are available, comfortable with institutions and in a stable state of mind. This systematically excludes those in crisis, those with trauma histories, people with limited digital skills, or those who distrust official systems.

For example, my elderly mother gets lost in cookie notices that block her screen. Her confusion is a daily reminder that many services are not built with her mental model in mind.

Over time, my personal rule has shifted: if my mother can’t understand it, it needs to be simpler. Vulnerability isn’t just about literacy, but about context, life experience and cognitive state.

Beyond “reflect or rebut”

Most of the services I’ve worked on – Universal Credit, Covid-19 grants, drink spiking information – are used by people who are, at that moment, cognitively impaired by stress, anxiety, uncertainty or other factors.

We need a more sophisticated approach that considers:

• How stress, trauma or substances affect information processing
• Cultural and linguistic differences that shape interpretation
• Limited digital skills and complex interfaces
• The emotional state of users in crisis
• How trust (or mistrust) affects engagement  

Mental models aren’t static. They shift with context, emotion, and experience.

As the CEO of Safeline said of my drink spiking work: “Spot on, with exactly the right tone of voice throughout and putting victims in charge of every next step, as it should be. We wouldn’t change a word”.

That wasn’t achieved by sharing experiences alone, but by understanding how vulnerability shapes decision-making and designing content to meet those needs.

A new framework for mental models

Instead of “reflect or rebut,” let’s ask:

• How might stress or trauma affect how users process this?
• What background knowledge am I assuming?
• How would someone with limited digital skills or cognitive impairments manage this journey?
• What if users are in an altered state (from stress, substances, or medical conditions)?
• How does trust (or lack of it) shape their interpretation? 

This vulnerability-aware approach doesn’t just make services more accessible; it makes them better for everyone. After all, we all experience vulnerability at some point in our lives.

From restaurants to real life

Mental models in content design aren’t like restaurant experiences. They’re complex, shifting frameworks influenced by emotion, cognition, culture and circumstance.

When we design for vulnerable users – in other words, for real people in their moments of greatest need – we must go beyond simplistic frameworks. We need to create content that works for people as they actually are: sometimes confused, sometimes stressed, sometimes cognitively impaired and always deserving of services that meet them where they are.

That’s not something you learn from restaurant analogies. It’s something you learn by recognising the full humanity of your users, including their vulnerability.

*If you’re a content designer, I encourage you: spend time with real users, especially those at their most vulnerable. That’s where the real learning begins.