Method: the five planes of user experience
This review uses the model from Jesse James Garrett, The Elements of User Experience (2002, second edition 2011; the original diagram is at jjg.net/elements). It is one of the most established frameworks in the field — old, but not outdated, because it describes dependencies that do not change with visual trends.
The model in plain English
Garrett describes every digital product as five planes, stacked bottom to top. Each plane answers one question, and every plane depends on the decisions of the plane below it:
| Plane | The question it answers | On our two screens |
|---|---|---|
| 5. Surface | What does it look like? | Colours, type, spacing, the brand orange |
| 4. Skeleton | Where do things go on the screen? | Field layout, button placement, error messages, list formatting |
| 3. Structure | How does the product behave and how is content organised? | Section order, the category picker, page hierarchy, where Apply lives |
| 2. Scope | What features and content are included? | Which fields the form collects, which content the detail page shows |
| 1. Strategy | What do users need, and what does the business need? | Employers want applicants fast; seekers want to decide fast; Jod wants quality structured listings |
Two rules from the model matter for how we wrote this review:
- Problems flow upward. A wrong decision on a lower plane cannot be fixed on a higher one. No amount of visual polish (surface) fixes a missing feature (scope). This is why each page of this review works bottom-up: strategy first, pixels last.
- The duality. Garrett splits every plane into a functionality side (the product as software — tasks, forms, actions) and an information side (the product as content — reading, scanning, deciding). The job form is mostly a functionality product; the job detail page is mostly an information product. They fail in different ways, and the review treats them differently.
Why this model for this review
- It matches the situation: the trigger for this review was a structure-plane change (the category picker, PR jodapp-web#1311) and the question "what else on these screens deserves the same treatment?". The model gives us a way to sort findings by depth instead of by how visible they are.
- It keeps us from redesigning for taste. A finding must name its plane and its evidence (a screenshot, a code path, or a ticketed bug).
- It scales into prototyping: surface and skeleton findings can be tried directly in Claude Design with the Jod Design System; structure findings need a prototype plus a discussion; scope and strategy findings need product decisions before any design work.
Honest limits of this review
Like the IA theory foundations page, we label what this review is and is not:
- It is expert review, not user research. Two screens were walked through by one reviewer against an established model, with code and shipped bugs as evidence. It finds violations of known principles. It cannot find what only watching real employers and seekers would find.
- Screenshots are from 2026-07-10; the category picker shipped after the form screenshot was taken, so the form review covers both the screenshot state and the current picker state, and says which is which.
- Analytics were not re-pulled for this review. Where a claim needs numbers (for example, how many live jobs have a one-sided pay range), the roadmap lists it as a data task instead of guessing.