Product #91

Product #91

image

Information Architecture · Louis Rosenfold, Peter Morville and Jorge Arango · 2015

In today's information-rich world, Information Architecture (IA) has become an essential discipline. At its core, IA is the art and science of shaping products to support usability, findability, and understanding.

Understanding the Ecosystem

Central to effective information design is an appreciation for what some call information ecology. This approach examines the dynamic intersection of users, context, and content.

There are four primary types of information seeking to consider:

  • Known-item seeking: When users know exactly what they are looking for—say, "What's the population of London?"
  • Exploratory seeking: When users have a general idea but need several results to hone in on a goal, such as finding the best restaurants in Kings Cross.
  • Exhaustive research: When users want to see every piece of relevant information.
  • Re-finding: When users are looking for information they have seen before—similar to "pin to top" or using favourites.

Often, users blend these approaches—searching, browsing, and occasionally asking questions. They might even employ what is known as berry picking behaviour, iterating their approach as new information is revealed, or engage in pearl growing, where a single resonant piece of content leads them to seek out similar information.

One of the key contributions of Information Architecture is that it creates consistency and coherence. Even as content evolves and channels change—with each medium having its own capabilities and limitations—users grow accustomed to certain semantic structures. Sudden changes in these structures can leave users feeling disorientated.

Components of Information Architecture

Information Architecture comprises several key components:

  • Organisational Systems: These dictate how information is arranged (by subject, chronology, etc.).
  • Navigation Systems: These help users move through content, often via a clickable hierarchy.
  • Search Systems: These allow users to find content by querying an index.
  • Labelling Systems: These define how information is represented through titles, abstracts, links, and more.

IA can be approached from the top down—shaping structures based on user journeys—or from the bottom up—organising according to the inherent structure of the content. It's important to remember that a single user action, like executing a search, touches on numerous IA decisions, from query building to display choices. This interconnectedness means that even small changes in IA can have significant impacts on user experience. Yet, designing effective IA is not without its challenges: language is ambiguous, information is often diverse and seemingly unrelated, perceptions differ from one person to the next, and company politics can complicate decision-making.

Organisational Systems

When we talk about organising content, there are several strategies available. Organisation schemes may be either exact—such as alphabetical, geographical, or chronological ordering—or ambiguous, based on topics, tasks, audience specifics, or even metaphorical groupings. Supermarkets, for example, often use hybrids of topic and task categorisations. While exact schemes tend to work best for known-item searching and are relatively easy to maintain and automate, ambiguous schemes are more challenging to design. However, they offer greater flexibility for exploratory searching, serendipitous discovery, and associated learning. Libraries provide a good example: they offer author, title, and subject schemes, with subject-based arrangements often being the most popular since users might not know precisely what they need.

In a top-down approach, hierarchies are used to create mutually exclusive groups with clear parent-child relationships, giving users an overall view of the information space and helping them understand where they are within it. This method requires careful balancing. On one hand, there is the tradeoff between exclusivity and inclusivity—ambiguous schemes rarely lend themselves to mutually exclusive groupings, sometimes resulting in polyhierarchical structures (where items appear in more than one place). On the other hand, designers must manage the balance between breadth (the number of options per level) and depth (the number of levels in the hierarchy). Starting with a broad and shallow structure is often advisable, adding more levels only as content grows.

A bottom-up approach leverages databases and metadata. Even without a predefined hierarchy, well-managed metadata (for instance, through tags) can support powerful search, browsing, filtering, and dynamic linking capabilities. Meanwhile, hypertext links provide flexible, creative pathways between content. Although they can unlock new relationships, relying solely on links can leave users without a sense of place. In the same vein, social classification—where users generate content and tag it, creating so-called "folksonomies"—can enhance discovery by breaking conventional groupings, though this approach only works well with a large, engaged user base.

Navigation Systems

Navigation is where many of the everyday decisions in IA come together. Labels, for instance, are powerful representations of concepts. A simple "Contact us" label can evoke a well-understood set of expectations without the need for explanation. Unlike spoken language, which naturally carries context and allows for immediate feedback, product interfaces must be deliberate and clear in their labelling. Consistency is key: applying a uniform style, syntax, and vocabulary across the product helps prevent confusion.

Creating effective labels is challenging because of the inherent ambiguity of language—synonyms, homonyms, and context all play a role. To build better labels, it is important to audit them regularly, benchmark against competitors, and consider using industry-standard taxonomies or controlled vocabularies. Observing your content, analysing search terms, consulting authors and user advocates, and employing testing methods like card sorting or free listing can all help refine your labelling system. Additionally, links and headings must be handled with care. While links require clear context to build trust and indicate destination, headings establish a visual and semantic hierarchy that guides the user through the content.

Navigation systems generally fall into several categories:

  • Global Navigation: Visible on every page, these systems help users understand where they are and provide quick access to major tasks.
  • Local Navigation: These provide context about what's nearby in the hierarchy, complementing the global view.
  • Contextual Navigation: This approach suggests related content or pathways based on what's currently being viewed.
  • Supplemental Navigation: Often used in large environments, these include sitemaps, indexes, and guides that serve as backups or additional aids to the main navigation.

For more complex interactions, configurators help users navigate decision trees by showing the impact of each choice—whether that means updating product images, prices, or shipping dates. Search remains a vital navigation tool, although it brings its own set of challenges due to the ambiguity of language. Features like personalisation (tailoring the experience based on user behaviour) and customisation (allowing users to tailor the experience themselves) must be balanced carefully. Typically, only power users return frequently enough to benefit from customisation, and even they may not always choose the best settings for themselves.

Social navigation is another emerging field. By leveraging the actions of related individuals—often through dynamically generated feeds—users can discover content in new ways. However, this approach also carries risks, such as creating echo chambers, so designers must proceed with caution.

Innovative navigation should always respect established norms and platform conventions. It's important to ensure that users immediately recognise where they are through consistent branding, understand the underlying hierarchy, and are aware of what actions are available. Testing tools like navigation stress tests can help designers balance flexibility with clarity, ensuring that global, local, and contextual navigation work harmoniously together.

Search Systems

Search functionality is a critical but resource-intensive component of any large information system. Before investing in complex search systems, consider whether the underlying navigation is solid—often, enhancing navigation will reduce the need for search. That said, search systems excel when content is abundant, dynamic, or fragmented. They provide valuable insight through search logs, revealing what users are truly looking for.

When implementing search, it is important to index most of the content to generate richer results, but equally important to exclude documents or content elements that might confuse users. For example, reviews that mention competitors could dilute the relevance of search outcomes. The concept of search zones—segmented areas of search results based on criteria like content type, audience, subject, or chronology—can be useful, but only when users expect or require a segmented view. Overcomplicating the search interface can lead many users to ignore these additional filters altogether.

A key challenge in search is balancing precision (the proportion of useful records returned) with recall (the proportion of all useful records that are returned). Enhancing one often comes at the expense of the other, so understanding whether users are engaged in known-item searches, exploratory searches, or exhaustive research is critical.

To enhance query effectiveness, designers can incorporate several tools:

  • Spell checkers correct errors before the query is processed.
  • Phonetic tools (like soundex) help capture variations in spelling (for example, "Smith" versus "Smyth").
  • Stemming tools ensure that words with the same root are recognised as related.
  • Natural language processing (NLP) can interpret queries such as "how to" or "who is" to narrow results.
  • Controlled vocabularies and thesauri can expand queries by including synonyms, thereby improving recall.

In presenting search results, it is important to consider user needs. Known-item searches may benefit from leaner results with less content per entry, while exploratory searches might require richer information like abstracts. Providing users with options to adjust result density or switch display types (such as a list versus a map) can enhance usability. Although pagination is often necessary, designers should note that most users rarely venture beyond the first page of results. When documents are flat and lack headings, showing contextual snippets surrounding search terms can help clarify results. Displaying the search query and the total number of results also supports iterative refinement. In some cases, incorporating key calls-to-action, enabling users to save searches or pin results, and offering explanations when the system modifies a query can further improve the experience.

Relevance is the default metric for ranking search results, typically determined by factors such as the frequency and proximity of query terms within a document, their placement (e.g., title versus body), and even the document's overall popularity. Yet there are situations where alternative sorting—alphabetical for names or chronological for time-sensitive content—can be more appropriate. Additional sorting options (by price, editorial choice, or user reviews) may also be valuable. For audiences that use search frequently, supporting an advanced query language interface might be worthwhile, though simplicity (like maintaining a single search box) is often preferable. Features such as AutoComplete and AutoSuggest can further guide users by offering suggestions based on partial input, providing hints about system structure, and aiding in query refinement.

Thesauri, Controlled Vocabularies, and Metadata

At the foundation of many powerful IA systems lies metadata—a descriptive tool that provides context about content. Metadata includes details such as how a piece of content was created, its purpose, the time and date of creation, the author, location, and standards used. Tags, a common form of metadata, help describe everything from documents and web pages to images, software, videos, and audio files. For example, HTML's meta tag allows authors to include keywords that assist search engines in indexing content.

In large, metadata-driven environments, controlled vocabularies are the glue that holds the system together. There are several types of controlled vocabularies:

  • Synonym Rings: These connect sets of words that are defined as equivalent for retrieval purposes.
  • Authority Files: These are lists of preferred terms, functioning as synonym rings where one term is selected as the standard.
  • Classification Schemes: These extend authority files by introducing hierarchical relationships. Libraries use the Dewey Decimal Classification (DDC) as one example—a system that starts with a limited number of master categories and expands into more detailed subcategories. Similarly, platforms like Netflix use a mix of macro genres (e.g., drama, comedy), micro genres (based on real-life nuances), and even micro tags (such as "#happyEnding") to refine categorisation.
  • Thesauri: This approach goes beyond simple synonym rings or hierarchies. In a thesauri system, each preferred term becomes a central hub, managing synonyms (equivalence), establishing classifications (hierarchical relationships), and linking associated terms that may not fit neatly into a hierarchy.

When implementing controlled vocabularies, designers should be cautious. Using these vocabularies directly in search results without sufficient explanation can confuse users, and there is always a tradeoff between enhancing recall (potentially by 20–80%) and maintaining precision. A balanced approach might be to use synonym expansion by default but to prioritise exact matches in the ranking.

Effective Information Architecture is about much more than organising content—it is about understanding users, context, and the intrinsic relationships within the content itself. Whether through careful organisational schemes, intuitive navigation systems, sophisticated search functionalities, or robust metadata management, each component plays a critical role in creating products that are both functional and delightful to use.

Full Book Summary · Amazon

Subscribe Button 

Quick Links

UX Research Canvas · Article

Why ‘Work Smarter, Not Harder’ Is Wrong for Beginners · Article

PMing with 01 Pro · Video

The No Surprises Rule for Influencing Upwards · Article

The AI PM Playbook · Article

image

The Computer as a Communication Device

J.C.R. Licklider, Robert W. Taylor. 1968. (View Paper → )

We believe that we are entering a technological age in which we will be able to interact with the richness of living information—not merely in the passive way that we have become accustomed to using books and libraries, but as active participants in an ongoing process, bringing something to it through our interaction with it, and not simply receiving something from it by our connection to it.

Licklider captured people’s imaginations with his predictions for the future. He was also working hard to make them happen, by building and orchestrating the research roadmap that would get us there.

He saw the power of online communication before most people, he knew how transformative it would be, and he wanted everyone to have access to it.

He predicted online communities of interests would happen, and that would make us happier than only having access to our local communities.

image

Book Highlights

While a job describes the overall task the customer is trying to execute, an outcome is a metric the customer uses to measure success and value while executing a job. Anthony W. Ulwick · Jobs to Be Done
The companies that grow the fastest are the ones that learn the fastest. The more experiments you run, the more you learn. It’s really that simple. The high volume is ideal, because most experiments fail to produce the results you’re hoping for. Others produce some indication of success but are inconclusive, not producing results significant enough to support making the change tested. Some produce small but not earthshattering wins. Only very few tests produce dramatic gains. Finding wins, both big and small, is, in other words, a numbers game. Morgan Brown & Sean Ellis · Hacking Growth
Thefacebook had captured the nation’s campuses, one by one, displaying almost military efficiency. Multiple colleges had asked for the service; to get to the front of the line, they were required to provide student emails, information about sports teams and clubs, class lists, and other information. That way, Thefacebook could sign up a high fraction of the students on each campus as soon as it launched, achieving critical mass immediately. Moreover, as it added more students to its community, Thefacebook was experiencing the opposite of the Friendster conundrum. Most college kids had high school buddies at other universities, so when those universities joined, the early loyalists grew even more engrossed in the platform. Thefacebook confronted no trade-off between expanding user numbers and diminishing user engagement. Sebastian Mallaby · The Power Law
image

Quotes & Tweets

I hate how well asking myself ‘if I had 10x the agency I have what would I do. @nickcammarata
Nearly everything awesome takes longer than you think. Get started and don’t worry about the clock. James Clear