Importing a Documentation Site Into Figma for an Information Architecture Review
Import a sample of documentation pages into Figma to review structure side by side. Inventory first, then capture, import report, and page layout.
You can review a documentation site's structure in Figma by importing a small, deliberately chosen sample of its pages and laying them out next to each other. Do the inventory first, pick one page per template, capture each page, read the import report, and arrange the frames in Figma pages that mirror the site's sections. The result shows organization and labeling at a glance. It won't show whether people can find things. This guide marks where that line sits.
Prerequisites: a Figma file you can edit, a list of the documentation URLs you care about, and the ability to save a page as HTML or capture it with a script.
Start with an inventory, not a canvas
A content inventory is the process of cataloging everything on a website, and the resulting catalog. Wikipedia's entry on content inventory describes it as quantitative: it answers "What is there?" A content audit is the qualitative counterpart and asks whether each item is any good. Both are planning aids for information architecture, and you want the inventory before you touch Figma, because it tells you which pages deserve a place on the canvas.
What the inventory records
The usual fields are the URL, file format, meta title, meta description, content owner, date last updated and audit notes. Teams often add a status (keep, revise or delete), the topic, and which pages link to and from each page. A spreadsheet is the normal tool. If you already have a site map, which MDN defines as a list of the pages of a website, use it as the first column and fill in the rest.
Pick pages worth importing
You don't need every page. You need one example of every template, plus any page your inventory flags as odd. For a typical documentation site that means:
- The landing page.
- A section index page.
- A short how-to article.
- A long reference page.
- A search results page, if it renders without a query service.
- A 404 page.
Add the pages with the oldest "last updated" date and any page with no inbound links. Ten to twenty pages is a reviewable set. Write the inventory row number in each page's name later, so the canvas and the spreadsheet stay joined.
Capture the pages you picked
UnHTML, the Figma plugin used here, accepts three kinds of input, as listed on its own page (checked 2026-11-18): a self-contained .html file dropped on the plugin's drop zone, a Safari .webarchive, or a scene.json captured with the open-source CLI or the bookmarklet. The scene.json route is the one meant for live sites, remote assets and pages built by JavaScript, which describes many documentation sites.
Choose the route per page
- Static documentation generated to plain HTML: save a self-contained file. See exporting self-contained HTML from Chrome.
- Pages you only have in Safari: save a page as a Safari web archive.
- Documentation behind a login, such as an internal handbook: read capturing a page behind a login before you start.
- Client-rendered documentation: use the
scene.jsoncapture.
Viewport and long pages
The plugin offers Desktop (1440 px), Mobile (390 px) or a custom width. For an architecture review, Desktop is usually enough, because you are looking at navigation and page structure. Add a Mobile import of the landing page and one article if the sidebar collapses into a menu, since that changes how the structure is presented.
Long reference pages are where imports run into limits. One full-page capture, layer caps, and lazy images covers them in detail.
A note on privacy
With default settings the import makes zero network requests, and fetching remote assets is an opt-in toggle. For internal documentation that matters: leave the toggle off unless a page cannot be captured without it.
Read the import report as an audit aid
After each run the plugin lists font substitutions, frames reverted to absolute positioning, and skipped images. Treat that list as a small audit of the page, not just a to-do list for the import.
| Import report entry | What it can mean for the review |
|---|---|
| Font substitution | The page's font is not installed in Figma, so a substitute was used. Usually a setup matter, not a documentation defect. |
| Frame reverted to absolute positioning | The captured geometry could not be reproduced with auto layout. Often marks a one-off layout, such as a hand-built landing block. |
| Skipped image | An image the import could not place. Check diagrams and screenshots, because documentation leans on them. |
A pattern across pages is more useful than a single entry. If every article page falls back in the same region, that region is probably a shared template, and fixing it once fixes the set. If only one page reverts, it was likely built differently from its neighbours, which is itself an inconsistency worth noting in the inventory. For context on how often these entries appear, see layer counts and fallback rates.
Mind the layer cap
As checked on 2026-11-18, the free tier allows up to 500 layers per import, while Pro and Studio remove the cap. The cap is per import, so a short how-to page fits and a very long reference page may not. A page that hits the cap is still reviewable: you see its top portion, which is where navigation and headings live.
The import is a faithful, editable set of layers, not a pixel-exact clone. Frames that cannot reproduce the captured geometry fall back to absolute positioning, and the report says which ones.
Lay the pages out in Figma
Figma's help page on creating and managing pages sets the practical limits. Starter team files are capped at three pages. Professional, Education and Organization teams have no cap. If you are on Starter, put everything on one page and use the section tool, which designates and labels a region of the canvas, to separate the documentation areas.
A layout that mirrors the site
On a plan without the cap, use one Figma page per top-level documentation section: Getting started, Guides, Reference, and so on. Then:
- Place each imported frame in the same order as your site map, left to right, with the landing page first.
- Start a page name with a hyphen to make a divider page between areas. Figma only allows this on a page with no content.
- Reorder pages by dragging them in the page list.
- Name each frame with its inventory row number and URL path.
If you imported a page onto the wrong Figma page, move the layers. They keep their coordinates on the new page, so nothing needs repositioning.
Mark what you see
Add a frame or section labelled "Orphan candidates" for pages with no inbound links in the inventory. Use comments for observations such as "two labels for the same thing" or "the sidebar lists this page twice". Keep each comment tied to a frame so it survives when someone rearranges the canvas.
What an import can and cannot show you
Information architecture combines organization, labeling, search and navigation systems. A Figma review is good at the first two and partly at the third, as rendered: you can compare how pages are labelled, how deep a sidebar goes, and whether sibling pages look like siblings.
It can't show search quality, because a captured page is static. Nor can it show whether readers find a page, which depends on their tasks and vocabulary. For that, the same Wikipedia article points to tree testing, a method for evaluating how findable topics are in a hierarchy. Use the Figma review to decide what to test, then test it with real readers.
Summary
Start from the inventory, import one page per template, read the import report as a small audit, and arrange the frames in Figma pages that mirror the site map. The canvas gives you a fast comparison of structure and labels. Findability still needs tree testing. If you want to try the import on your own pages, UnHTML runs inside Figma with the free tier's 500-layer cap per import.
Frequently Asked Questions
Can I review documentation structure in Figma without a spreadsheet?
Not completely. A spreadsheet inventory records owners, dates and status for every URL, which a canvas does not hold well. Figma shows how pages look and relate side by side. Use both, with the inventory row number in each frame name.
How many documentation pages can I import on the free tier?
Each import is capped at 500 layers on the free tier, as checked on 2026-11-18. That limit applies per import, not per file, so a typical page fits and a very long reference page may not.
Does the import tell me whether people can find a page?
No. It shows layout, labels and visible navigation as rendered. Findability depends on readers' tasks, and tree testing or task-based tests with real readers is the way to measure it.
How many pages can a Figma file hold?
Starter team files are capped at three pages. Professional, Education and Organization teams have no cap. On Starter, keep everything on one page and use sections to separate documentation areas.
Questions
- Can I review documentation structure in Figma without a spreadsheet?
- Not completely. A spreadsheet inventory records owners, dates and status for every URL, which a canvas does not hold well. Figma shows how pages look and relate side by side. Use both, with the inventory row number in each frame name.
- How many documentation pages can I import on the free tier?
- Each import is capped at 500 layers on the free tier, as checked on 2026-11-18. That limit applies per import, not per file, so a typical page fits and a very long reference page may not.
- Does the import tell me whether people can find a page?
- No. It shows layout, labels and visible navigation as rendered. Findability depends on readers' tasks, and tree testing or task-based tests with real readers is the way to measure it.
- How many pages can a Figma file hold?
- Starter team files are capped at three pages. Professional, Education and Organization teams have no cap. On Starter, keep everything on one page and use sections to separate documentation areas.