Theory & Method Foundations

This page exists so that an experienced information architect can check our plan against established theory. For each part of the plan, it names the concept we used, where the concept comes from, and how well established it is.

We are honest here about three different things:

• Where we used a proven concept, we name the concept and its source.
• Where we used our own way of presenting an idea, we say so, and we name the proven principle underneath it.
• Where a popular idea is actually not proven, we say that too, and we explain what we used instead.

The reason for this honesty is simple. An expert reviewer will trust the plan more if we are clear about what is a law, what is a method, what is a rule of thumb, and what is only a presentation choice.

How to read the status labels

Every concept below has a status label. Here is what each one means.

• Established law — a research finding that holds reliably across many studies.
• Established method or practice — a widely used and accepted way of working in the field.
• Heuristic — a useful rule of thumb, not a law. It is usually right, but not always.
• Practitioner framework / practitioner research — widely used in industry and useful in practice, but with thin peer-reviewed validation. We do not put it at the same level as an established law.
• Our presentation choice — a way we chose to communicate an idea. The idea underneath is proven, but the specific format is ours.
• Myth we avoided — a popular idea that research has shown to be wrong, which we did not rely on.

The discipline itself

The whole approach of defining structure before visual design comes from the field of information architecture.

• The four systems of information architecture. We organise the work into content, labels, navigation, and search. This comes from Rosenfeld, Morville, and Arango, Information Architecture for the World Wide Web (often called "the polar bear book"). They describe four systems: organization, labeling, navigation, and search. Our deliverables map directly onto these four systems. Status: established method.
• Findability. Our test that a user must be able to find a job matches Peter Morville's concept of findability, from Ambient Findability (2005). Status: established concept.

Concept behind each deliverable

The table below maps each deliverable to the main concept or concepts it uses, the source, and the status.

Deliverable	Concept used	Source	Status
Content Inventory	Content inventory and audit	Content strategy practice; Halvorson & Rach, Content Strategy for the Web	Established method
Content Inventory	Prioritise by frequency of use	Long-standing interaction-design and IA practice	Established practice
Content Inventory	The "Tier 0–4" numbering	Our own labels	Our presentation choice
Content Inventory	"Job to be done" per item	Jobs To Be Done; Tony Ulwick, popularised by Clayton Christensen	Practitioner framework — useful, not proven theory (see below)
Vocabulary Lexicon	The vocabulary problem	Furnas, Landauer, Gomez & Dumais (1987), The vocabulary problem in human-system communication	Established law
Vocabulary Lexicon	Speak the user's language	Nielsen's 10 usability heuristics, heuristic 2 (match between the system and the real world)	Established heuristic
Vocabulary Lexicon	Use search logs to find real words	Search analytics for IA; Rosenfeld, Search Analytics for Your Site	Established method
Navigation Hierarchy	Broad and shallow beats deep	Breadth-versus-depth research (for example Larson & Czerwinski, 1998) and Nielsen Norman Group guidance	Established guidance
Navigation Hierarchy	Information scent	Pirolli & Card, Information Foraging Theory (1999)	Established theory
Navigation Hierarchy	Keep the choice set small	Breadth-versus-depth research (Larson & Czerwinski, 1998); Hick's Law is often cited here, but it does not cleanly apply	Established guidance (see the honest note below)
Navigation Hierarchy	Job type as a facet, not a separate site	Faceted classification; Ranganathan (1933); modern faceted search, Hearst, Search User Interfaces (2009)	Established method
Navigation Hierarchy	Mobile bottom bar within thumb reach	Practitioner touch research (Steven Hoober); Fitts's Law (1954) is related, but it does not prove the thumb-zone map	Practitioner research (see the honest note below)
Taxonomy & Filters	Faceted navigation (filters as facets)	Ranganathan (1933); Hearst (2009)	Established method
Taxonomy & Filters	"Under 3 clicks"	The 3-click rule	Myth we avoided (see below)
Taxonomy & Filters	Each click must feel like progress	Information scent (Pirolli & Card)	Established theory
Taxonomy & Filters	Build categories from user mental models	Card sorting; Spencer, Card Sorting: Designing Usable Categories	Established method
Primary Flows	User flows and task analysis	Standard interaction-design practice	Established method
Primary Flows	Conversion funnel and drop-off	Web analytics and conversion-rate-optimisation practice	Established practice
Primary Flows	Record the application before the profile	Deferred (lazy) registration and progressive disclosure; Nielsen; Wroblewski on sign-up friction	Established pattern
Primary Flows	Friction lowers the ability to act	Fogg Behavior Model, B=MAP (BJ Fogg, Stanford)	Practitioner framework — useful, not proven theory (see below)
Screen Map	Sitemap and screen inventory	Standard IA deliverable	Established method
Screen Map	Keep, redesign, or remove each screen	ROT analysis (Redundant, Outdated, Trivial) from content audits	Established method
Friction Points	Funnel drop-off analysis	Conversion-rate-optimisation practice	Established practice
Friction Points	The "why" behind each drop-off	Heuristic evaluation; Nielsen & Molich (1990)	Established method
Prioritization Roadmap	Impact-versus-effort ranking	Action Priority Matrix; value-versus-effort and RICE prioritisation	Established practice

Where we were honest and did not claim a proven law

These are the places where a popular idea is weaker than it sounds. We call them out so the expert can see we did not dress them up.

The "Tier 0–4" grouping is our presentation choice

• What is proven: grouping and prioritising content by how often users need it. This is standard practice.
• What is ours: the specific "Tier 0", "Tier 1", and so on, and the phrase "core loop", which we borrowed from product and game design.
• Why it is still safe: the numbers are only a way to communicate priority. They do not claim to be a formal model.

The "under 3 clicks" rule is a myth

• What is popular: the "3-click rule", which says a user should reach any page in three clicks or fewer.
• What the research shows: Joshua Porter, at User Interface Engineering (2003), tested this and found no link between the number of clicks and either task success or user satisfaction. Users do not count clicks. They keep going as long as they feel they are getting closer to their goal.
• What we used instead: information scent. Each step must clearly point toward the job the user wants.
• How this fits the taxonomy page: the Taxonomy & Filters page still uses a 3-click budget as a design constraint, because shorter paths reduce user effort — but we do not treat it as a law, and task success matters more than the click count.

Miller's "7±2" does not apply to menus

• What is popular: using Miller's "magic number seven, plus or minus two" to limit how many items a menu may have.
• What the research really says: Miller's 1956 paper is about short-term memory span, not about how many links a navigation menu should hold.
• What we used instead: breadth-versus-depth hierarchy research, which is the correct basis for deciding navigation breadth. We do not lean on Hick's Law for this either — see the next note.

Hick's Law does not cleanly apply to menus

• What is popular: citing Hick's Law to argue that menus must stay small.
• What the research really says: Hick's Law describes fast choices among a small set of options that the person already knows and has practiced — for example, pressing the right button in a reaction-time task. Using a menu is a different activity: the user visually scans and reads labels they may see for the first time. Reading and scanning do not follow Hick's time formula.
• The logic, step by step: the fact is that menu use is visual search, not practiced reaction. Because of that, Hick's Law does not describe menu use. So Hick's Law cannot be the basis for our small menus.
• What we used instead: the breadth-versus-depth research (Larson & Czerwinski, 1998; Nielsen Norman Group guidance), which studied navigation structures directly and supports small, well-labelled menus. The conclusion — keep the choice set small — stands. Only its basis changes.

The thumb-zone map is practitioner research, not a Fitts's Law result

• What is popular: saying that Fitts's Law proves the bottom of a phone screen is the easiest area for the thumb.
• What the research really says: Fitts's Law (1954) is a real law, but it says something narrower: bigger and closer targets are faster to hit. The specific map of "easy" and "hard" thumb areas comes from practitioner touch studies — mainly Steven Hoober's observations of how people actually hold phones.
• Why we keep the decision: Hoober's studies use real observation data and match wide industry experience. The bottom-bar decision stands. We only label its basis honestly: practitioner research, not an established law.

Fogg Behavior Model and Jobs To Be Done are practitioner frameworks

• What is popular: treating B=MAP (Fogg) and "jobs to be done" (Ulwick, Christensen) as proven theory.
• What the research really says: both are widely used industry frameworks with thin peer-reviewed validation. They give useful language — "friction lowers the ability to act", "what job is the user hiring this page for" — but they are not validated the way, for example, Furnas's 1987 vocabulary-problem study is.
• Why we keep them: they are thinking tools, and they helped us structure the flows and the inventory. We label them "practitioner framework — useful, not proven theory", and we do not rest any load-bearing claim on them alone.

What we have not yet validated, and how to

Good information-architecture practice does not stop at a plan. It tests the plan with real users before and during the build. We have not yet run these tests, and an expert will expect to see them planned. Here are the standard methods, and what each one would check for us.

• Card sorting. Give users the job categories and ask them to group and name them. This checks whether our 8-category taxonomy matches how users actually group work. Source: Donna Spencer, Card Sorting: Designing Usable Categories.
• Tree testing. Give users the navigation structure with no visual design and ask them to find a job. This checks whether the labels and grouping in the Navigation Hierarchy actually lead users to the right place. Source: tree testing, described by Spencer and by Nielsen Norman Group.
• First-click testing. Show a screen and measure where users click first. Research (Bob Bailey and others) shows that when the first click is correct, task success is much more likely. This checks the job card, the filters, and the apply button placement.
• Usability testing of the apply flow. Watch real users try to apply on a phone. This checks the redesigned apply flow and the friction points directly.

We recommend running card sorting and tree testing before the build starts, and first-click and usability testing during the build.

Small-team reality — every round needs one named owner

We have no researcher on the team. A research plan without a named owner does not happen. So each round below is scoped to be cheap — about half a day per round, with 5–8 participants recruited from existing users (for example, recent applicants) — and one engineer owns each round end to end.

• Card sorting: use the free tier of an online card-sorting tool (for example OptimalSort's free plan, or the open-source kardSort). 5–8 participants, half a day.
• Tree testing: use the free tier of an online tree-testing tool (for example Treejack's free plan). Same recruiting, half a day.
• First-click testing: the cheapest version needs no tool at all — share a screenshot on a call and ask "where would you tap first?". A free click-test tool (for example Chalkmark's free plan) also works. Same recruiting, half a day.

Before a round starts, write the owner's name next to it in the roadmap. An unowned round is a round that will not run.

References

• Rosenfeld, L., Morville, P., & Arango, J. Information Architecture for the World Wide Web (4th ed.).
• Morville, P. Ambient Findability (2005).
• Furnas, G. W., Landauer, T. K., Gomez, L. M., & Dumais, S. T. (1987). "The vocabulary problem in human-system communication." Communications of the ACM, 30(11).
• Nielsen, J. "10 Usability Heuristics for User Interface Design" (1994, Nielsen Norman Group).
• Nielsen, J., & Molich, R. (1990). "Heuristic evaluation of user interfaces."
• Pirolli, P., & Card, S. (1999). "Information Foraging." Psychological Review.
• Hick, W. E. (1952); Hyman, R. (1953). Foundational studies of choice reaction time (Hick's Law).
• Fitts, P. M. (1954). "The information capacity of the human motor system in controlling the amplitude of movement."
• Larson, K., & Czerwinski, M. (1998). "Web page design: implications of memory, structure and scent for information retrieval." CHI '98.
• Hoober, S. (2013). "How Do Users Really Hold Mobile Devices?" UXmatters, and follow-up touch studies.
• Ranganathan, S. R. (1933). Colon Classification (faceted classification).
• Hearst, M. (2009). Search User Interfaces.
• Ulwick, T. Jobs To Be Done: Theory to Practice; Christensen, C. (Jobs To Be Done).
• Halvorson, K., & Rach, M. Content Strategy for the Web.
• Spencer, D. Card Sorting: Designing Usable Categories.
• Fogg, B. J. Fogg Behavior Model (behaviormodel.org).
• Porter, J. (2003). "Testing the Three-Click Rule." User Interface Engineering.
• Wroblewski, L. Writing on sign-up and form friction.