Web Annotation and Read-Later Apps

The original Awesome Highlighter sat awkwardly between these families: it marked the live page like an annotation tool, then saved a copy like a read-later app so the link would keep working. Modern tools have specialised. Knowing which family you are in explains almost every difference in how highlights behave, where they are stored and what happens when the page changes.

Annotation layers: highlights on the live page

An annotation tool (Hypothesis is the best-known open example; Diigo and Weava are commercial ones) adds a layer over the real web page. You highlight on the page as published, your highlight is stored on the tool's server with enough surrounding text to find it again, and when you return the layer re-applies it. Because the page is the live page, annotations can be shared with others looking at the same page, discussed in threads, and made public. This is what makes annotation layers the tool of choice for classes, reading groups, peer review and fact-checking.

The cost is fragility and dependence. If the page changes, the tool has to guess where your highlight belongs (good tools anchor by text with context and usually succeed; if the passage is deleted the highlight becomes an orphan). If the page disappears, so does the context for your highlights, though the quoted text survives in your account. And the tool needs to be present, as an extension or bookmarklet, to show anything.

Read-later apps: highlights on your own copy

A read-later app (Pocket was the classic; Instapaper, Readwise Reader, Matter and Omnivore-style tools are current examples) saves a cleaned copy of the article's text into your account. You read and highlight that copy in the app's own interface, on any device, offline if you like. Highlights never break because your copy never changes, and the reading view is calm and consistent across every site. Most of these apps export highlights automatically to note-taking tools, which is the main reason writers and researchers use them.

The trade is separation from the web. Your highlights are not on the page, so nobody else visiting the page sees them, and you cannot annotate together. Pages that resist extraction (paywalls, heavy scripts, tables, comment threads) save badly. And the saved copy is yours only as long as the service exists and you keep exporting.

Getting highlights into your notes

Whichever family you choose, the highlights are only useful if they end up where you think. Look for export as plain text or markdown that includes the quote, the source address and the date, then set up a routine: weekly for annotation layers (most need a manual export or a sync tool), automatic for read-later apps that connect to note apps. Our guide on turning highlights into notes covers what to do with them once they arrive, which is the part that actually produces learning or writing.

Worked example: what a read later app shutdown looks like

Pocket, the read-later app most people had heard of, was closed by its owner Mozilla on 8 July 2025. The published timeline is worth reading because it is typical of how a service ends.

The window between the announcement and the end of export was under six months, and anyone who had not exported by 12 November lost every highlight. Readers who had a routine export, or an automatic connection to a notes app, lost nothing but a habit. The pattern, an announcement, a short export-only period, then deletion, is how most services close, and it is why this site treats exporting as part of using any highlighting tool rather than as an emergency measure. What an export needs to contain to count as a record of work lists the fields.

Expert tips

• Set up the export before you make the highlights. Connect the app to your notes, or put a monthly export in your calendar, on the day you sign up.
• Export to a format you can open without the app: plain text, markdown or CSV, never only the app's own backup format.
• Check one exported highlight against its source once. If the export lacks the page address or the date, it is a list of quotes, not a record.
• Keep the article's address with every highlight, and where the page is likely to move, keep a highlight link as well.
• Let the export do the bibliography. An export with quote, address and date is most of the work; turning a read-later export into an annotated bibliography shows the remaining steps.

Choosing, by situation

• A class reading one text together: annotation layer with groups (Hypothesis). Students see each other's highlights and questions in the margin.
• A researcher reading fifty articles a week for a project: read-later app with automatic export to notes.
• A student studying from web pages and PDFs: a tool that handles both (Weava, or a read-later app that accepts PDFs) with colour codes.
• A team fact-checking or reviewing a published page: annotation layer in a private group, so comments sit on the exact claim.
• Someone who only wants to send a colleague to a paragraph: neither. A highlight link is faster and stores nothing.

A note on privacy

Both families hold your reading history in an account. Read-later apps know every article you saved; annotation extensions can see every page you visit while enabled. Read the policy, prefer tools with a paid tier or nonprofit funding, keep exports, and turn extensions off on sites where you handle money or health records. For quick, private highlighting with nothing stored, the bookmarklet remains the simplest tool on this site.

Annotation apps overlap with two other pages in the tools and extensions section: the comparison of browser highlighters covers the extensions that feed these apps, and how highlight links work explains the lighter alternative when you only need to point at a passage. Annotation layers are also well suited to verifying the claims in an article with a class. Shared annotation is also useful for reviewing writing fairly; why a detector score needs a human reading explains the reason.