Mental models in web design: examples and how to research them
- Published
- Reading time
- 8 min
- Published by
- Alex

Users arrive with expectations
Every visitor brings a picture of how websites work, built on all the other sites they use. Design that matches that picture feels obvious; design that breaks it feels broken – even when it works.
A mental model is what a user believes about how a website or app works, based on their earlier experience. Those beliefs let people predict what will happen when they click, scroll or type – so when a design matches them, the site feels intuitive, and when it does not, people make mistakes or leave. This article explains the concept, the closely related Jakob’s Law, seven common examples from web design and how to find out what your own users expect.
What is a mental model?
Nielsen Norman Group defines a mental model as what the user believes about the system at hand – a website, an application or another product. Mental models help users predict how a system will work, and therefore influence how they interact with the interface (NN/g, Megan Chan, 2024, updated from Jakob Nielsen’s 2010 article).
Three properties matter in practice:
- It is based on belief, not fact. A user who thinks the logo always leads to the homepage will click it, whatever your design intended.
- It is individual. People build their models from their own experience, so a first-time visitor and a long-term B2B customer may understand the same site differently.
- It changes slowly. NN/g notes the strong inertia in users’ mental models: familiar patterns are sticky, and new interaction styles need to earn their place.
There is also a gap between designers and users. The people who build a site know its structure in detail, so everything seems logical to them. Users have a simpler, less complete model – which is why features that seem obvious internally are often missed by visitors.
Jakob’s Law: why conventions work
Jakob’s Law was formulated by Jakob Nielsen in his article “The End of Web Design” (July 2000). Its core is simple: users spend most of their time on other sites, so they prefer your site to work the same way as all the other sites they already know (NN/g, 2000).
This explains why so many websites look similar. The logo top left, the cart top right, the search box with a magnifying glass, links in the footer – these are not a lack of imagination but shared conventions that let visitors spend their attention on the task, not on learning your interface.
For Swedish companies, local conventions are part of the picture too. Swedish users expect to identify themselves with BankID in services that require login, and in e-commerce they expect Swish and “buy now, pay later”: according to the Riksbank, these were the single most common payment methods for online purchases in its September 2025 survey. A checkout without them breaks a mental model many Swedish buyers have.
7 examples of mental models in web design
1. The shopping cart

The online cart borrows directly from the physical trolley: you collect products, can see what is inside, add or remove items and pay at the end. Users expect the cart icon in the top-right corner, a count that updates when they add something, and the freedom to change their mind before paying.
Breaking this model is costly. A cart that only shows a total, hides the remove button or forces account creation before people can review their order creates doubt at the very moment they are about to buy. In B2B shops, the same model extends to features such as saved carts, quick order by article number and requests for quotes – but the basic logic should remain the same.
2. The hamburger menu and navigation

The three-line icon is now widely understood as “menu”, especially on mobile. But recognising the icon is not the same as using it. In a study of hidden navigation, NN/g found that menus hidden behind a hamburger were less discoverable than visible navigation, and that the penalty was larger on desktop than on mobile (NN/g, Kara Pernice and Raluca Budiu, 2016).
In practice: on desktop, show the main sections of the site as visible links. On mobile, a hamburger is acceptable, but consider keeping the most important one or two destinations – for example the product range or “request a quote” – visible outside it.
3. Forms

Forms are often the main point of contact between a company and its customers, and users bring strong expectations: a label for each field, one column, a clear send button and helpful feedback if something is wrong.
A common way to break this model is to use placeholder text inside the field instead of a label. NN/g lists several problems with it: the hint disappears as soon as people type, they cannot check their answers afterwards, error correction becomes harder, and light grey placeholder text often has poor contrast. Its recommendation is to keep labels visible outside the field (NN/g, Katie Sherwin, 2014).
4. Phone numbers and chunking

A long string of digits is hard to read, remember and check. Chunking – splitting information into small, distinct units – is a well-documented way of making it easier to process, and NN/g uses phone numbers as an example of it (NN/g, Kate Moran, 2016).
Users also expect numbers in their local format. A Swedish visitor recognises a number written as 08-123 456 78 more easily than an unbroken sequence of ten digits. Format phone numbers, organisation numbers and order numbers the way your users are used to seeing them – and let input fields accept spaces and hyphens instead of rejecting them.
5. Typographic hierarchy

Readers expect larger, heavier text to be more important, and they scan pages by jumping between headings. A clear hierarchy – one main heading, descriptive subheadings, short paragraphs and lists – lets people find the part they need without reading everything. When styles are used inconsistently, for example bold body text that looks like a heading, the page becomes harder to scan.
6. The search field

A rectangular field with a magnifying glass is one of the most stable conventions on the web. NN/g recommends against an icon-only search on desktop websites: keep an open text field next to the icon so that search is easy to spot, and make sure clicking the icon submits the query (NN/g, Katie Sherwin, 2014). On mobile, the field can be hidden behind the icon to save space.
Users’ model of search also comes from the large search engines, so they expect it to tolerate spelling mistakes, suggest results while they type and find products by article number as well as by name. For B2B shops with large product ranges, search is often the most important navigation of all.
7. Pricing tables
Pricing tables follow a familiar structure that users read without instruction: plans side by side, the price prominent at the top of each column, features listed in the same order and a clear call to action under each plan. Users expect to compare like with like. Hiding prices behind a contact form, or listing features in a different order in each column, makes comparison harder. If you sell to companies, also state whether prices include VAT – Swedish B2B buyers usually expect prices excluding VAT, consumers including it.
Mental model or design convention?
The two terms are often mixed up. The table shows the difference.
| Mental model | Design convention | |
|---|---|---|
| What it is | The user’s belief about how something works | A pattern that many websites use in the same way |
| Where it lives | In the user’s head | In the interface |
| Example | “Clicking the logo takes me to the homepage” | A linked logo in the top-left corner |
| Changes | Slowly, through experience | When enough sites change together |
| Your role | Understand it through research | Use it, or break it only with good reason |
Conventions exist because they match common mental models. When you follow them, you benefit from everything users have learned elsewhere.
How to research your users’ mental models
You cannot see a mental model directly, but you can see its effects. These methods are well established and work for most company websites:
- Open card sorting. Participants group cards with your content into categories that make sense to them and name the groups. It shows how users would organise your information. NN/g recommends at least 15 participants for a qualitative card sort (NN/g, Samhita Tankala and Katie Sherwin, 2024).
- Tree testing. Users try to find specific items in a text-only version of your menu structure. It tests whether a proposed navigation works before anything is designed.
- Think-aloud user tests. Users complete real tasks while saying what they expect to happen. The moments of surprise reveal mismatches between your design and their model.
- Search logs and customer service questions. The words people type into your search box and the questions they ask support show which terms and categories they use.
- Interviews with sales staff. In B2B, the people who talk to buyers every day often know exactly where customers get lost on the website.
The balance between originality and usability
Following conventions does not mean every website has to look the same. Brand, imagery, typography, tone of voice and storytelling leave a great deal of room for originality – and that is where the emotional side of a website lives, as we describe in our article on emotional design and the customer experience. Interaction patterns, navigation and checkout are where predictability pays off.
When you do break a convention, do it deliberately: because research shows that your users have a different model, or because the new solution is clearly better. Then test it with real users, and make the new behaviour easy to learn.
Planning a new website or a redesign? Our website redesign guide covers how to keep what works, and you can read more about how we build B2B websites and e-commerce.
Key insights
- 01
A mental model is a belief
It is what users think a system does, not how it really works – and it decides what they try to do next.
- 02
Jakob's Law
Users spend most of their time on other sites, so they expect yours to work the same way. Conventions are an asset, not a lack of creativity.
- 03
Designers are the worst judges
You know your own site in detail. Your users do not, so their model is simpler and their mistakes are predictable.
- 04
Research beats guessing
Card sorting, tree testing and think-aloud tests show how real users group, look for and understand your content.















