PMaker home
Start from the information; page count is pushed out by content volumeStart from pagesYou fill in content by copying interfaces you've seen, patching gaps as you go—the page count keeps shiftingHalfway through a page you find a missing field and rework the layoutStart from informationList everything first, then split pages by frequency and length—the result falls out on its ownThe inventory doubles as a data-model draft; the table structure is mostly decidedHigh-frequency and short goes in a list; low-frequency or long goes in a detail view; rarely seen isn't shown at all

Pages aren't imagined into being—they grow out of the information inventory. List it all first and the pagination step becomes busywork.

Content Inventory

List everything you need to show before deciding how many pages.

List out every piece of information to display, then decide which page each one goes on. Reverse the order and the page count keeps shifting. A content inventory is the first step in turning page design from "by feel" into "derived," and a standard opening move for web and product information architecture. content-inventory-wiki

What you'll run into:

  • You're halfway through a page when a missing field forces a full layout rework
  • The same information appears on three pages with different levels of detail
  • You ask the AI for a detail page and it invents a pile of fields you don't need

Start from the page and you'll unconsciously fill in content by copying interfaces you've seen, plugging gaps as you go. Start from the information and the page count and layout become a result driven by content volume—not a guess.

A minimal example: a "project" page needs at least five pieces of information—name, status, owner, due date, and notes. Listing them first and then paginating is far more reliable than drawing an interface that "looks like a project page," which tends to get the detail levels of notes versus history backwards.

The inventory has a side effect: it's basically a draft of your data model. Once you've spelled out which fields the user fills in and which the system computes, the table structure is mostly set.

Which columns the inventory needs

Column What to fill in What it decides
Information One specific piece, not a group What actually shows on the page
Source User-entered, system-computed, or external Whether you need an input UI, and whether to handle empty
Who views it The user, the customer, or an admin Permission design, and whether multiple views are needed
Frequency High, occasional, rarely List page or detail page, or fold it away entirely
Length range Shortest and longest possible Whether the layout survives extremes, and whether to truncate

The last column is the one most often skipped, and the cost is direct: one user names their project with forty characters and your list layout collapses.

There's another dimension that's easy to miss: dependencies between pieces of information. If "owner" is empty, the "remind" button shouldn't appear. Flagging these dependencies at the inventory stage is far cheaper than discovering them in code.

Once listed, the pagination rule is clear: high-frequency and short content goes in a list; low-frequency or long content goes in a detail view; rarely-viewed content isn't shown and is fetched only when needed.

Note for the AI

Having the AI produce the inventory first is far more stable than having it draw pages directly—and much easier to revise. Keep the loop two-way: every time the inventory changes, have the AI redraw the affected page instead of patching the old one, so the inventory stays the single source of truth.

PROMPT · Content inventory

I'm building: [one-sentence description of the feature]
Use case: [who, when, what they need to accomplish]

Don't draw pages or write code yet. First produce a content
inventory, in table form, with these columns:

Information | Source (user-entered / system-computed / external) | Who views it | View frequency | Length range

Requirements:
1. Only list what this scenario truly needs. Anything you think
   "most products have" but this scenario doesn't need goes into
   a separate "suggested to skip" block, with a reason—not into
   the main table.
2. For each item, note what the interface should show if it's empty.
3. After listing, propose a pagination plan and explain why each
   page holds these items.

Wait for my confirmation before drawing any pages.

The first requirement is the key. Without it, the AI fills in fields based on similar products it has seen, and you get a generic template instead of your product.

How to check the inventory is complete. Cross-check against the final pages in reverse: every field on every screen maps to an inventory entry, and every inventory entry has a home on some page. Both directions must hold for the inventory to be closed-loop—if either side doesn't match, that spot was patched in by guesswork.

References

  1. Content inventory — Wikipedia