Choose stories for a curated newsletter

If you want to know how to choose stories for a curated newsletter, use a simple rule: include only stories that are relevant to your readers, supported by the original source, and useful once you add your own context. A curated issue earns attention when your selection saves time and sharpens judgment, not when it simply proves you read a lot.

A useful issue usually has three layers:

• the story itself
• why it matters to this audience
• what to do, watch or ignore next

That is the difference between curation and aggregation. Aggregation says, "here are ten links." Curation says, "these four developments changed something for you this week, and here is why."

Start with a selection rule, not a reading pile

Before you collect stories, decide what qualifies. Otherwise every interesting link feels eligible.

Use three tests.

1. Relevance: is this for your specific reader?

Ask:

• Who exactly is this for?
• What job are they trying to do?
• Does this story affect that job now?

A story can be big news and still be wrong for your issue. If you write for startup operators, a research paper may matter only if it changes tooling, pricing, hiring, regulation or product choices. If you write for security teams, the same paper may matter for risk and policy instead.

A quick relevance filter:

Write your audience sentence in one line before you read:

Illustrative template:

"This newsletter helps [specific people] understand [topic] so they can [decision or action]."

If a story does not help that sentence, it probably does not belong.

2. Evidence: can you verify the claim from the original source?

A curated issue should not pass along second-hand claims without checking them. Many weak newsletters link to commentary about a story instead of the source that made the claim.

Prefer:

• company posts over reactions to company posts
• the original article over a social summary of it
• the video transcript over a paraphrase of the video
• the paper, release note or repository update over hearsay

This matters because your subscribers borrow your judgment. If the claim is overstated, missing a condition or contradicted later in the piece, you carry the error into the issue.

3. Usefulness: will the reader know what to do with this?

A story becomes newsletter-worthy when it gives the reader one of these:

• a decision to consider
• a change to watch
• a shortcut to understand a trend
• a practical example to copy
• a warning about noise, hype or weak evidence

A useful item often answers one of these questions:

• What changed?
• Who is affected?
• How soon does it matter?
• What is still unclear?
• What should the reader do next?

Use a simple scoring method

When several stories compete for space, score each one quickly instead of choosing by instinct alone.

Use 0 to 2 points for each criterion:

• Relevance: 0 = tangential, 1 = somewhat relevant, 2 = directly useful
• Evidence: 0 = unclear or second-hand, 1 = partly checked, 2 = verified in the original source
• Usefulness: 0 = interesting only, 1 = informative, 2 = leads to action or a clearer decision
• Original context you can add: 0 = nothing to add, 1 = brief note, 2 = strong angle or interpretation

A story scoring 7 or 8 is usually strong. A story scoring 4 may still fit if it supports a bigger theme. A story scoring 2 or 3 is often just clutter.

Illustrative example:

• New model release everyone is discussing

  • Relevance: 2
  • Evidence: 2
  • Usefulness: 1
  • Context: 2
  • Total: 7

• Viral opinion thread about the release

  • Relevance: 1
  • Evidence: 0
  • Usefulness: 0
  • Context: 1
  • Total: 2

The second item may be interesting. It still should not take a slot from something your reader can use.

Add context so the issue is worth reading

The most common failure in curated newsletters is stopping at the link. Your subscribers can collect links themselves. They open your issue for compression and judgment.

For each chosen story, add one or more of these:

Explain why it matters to this audience

Do not write, "Interesting read." Write, "This matters if you run a small data team because it lowers the work needed to evaluate outputs across providers."

State the strongest claim carefully

Summarize the main point in plain words without overstating it. If the source says "in our tests" or "with captions available," keep that condition.

Pull out the useful detail

A good item often turns on one detail:

• rollout date
• supported feature
• limitation
• cost implication
• workflow change
• new evidence

Add your decision lens

Your original contribution can be small but clear:

• "Worth testing this week"
• "Useful, but not enough evidence yet"
• "Important only for larger teams"
• "A sign of where tooling is moving"

A short, honest judgment is more valuable than a long generic summary.

Build each issue around a small set of story roles

Most curated newsletters improve when each link has a job.

Try a simple mix:

• 1 lead story: the main change of the week
• 2 to 4 core items: developments readers should know
• 1 practical resource: tutorial, data, transcript, release notes or repository
• 1 contrarian or clarifying item: what is being overstated, misunderstood or missed

This stops the issue becoming five versions of the same story.

Do it with Repurai

The manual version of this process means too many tabs, missed updates and weak recall when you try to write. Repurai helps by keeping your sources in one place, refreshing them about every hour, helping you triage quickly, and saving the pieces you want to use later.

1. Group sources by newsletter promise

Create a Space for the issue or topic you write about. Add the websites, newsletters, YouTube channels, Substacks, Medium publications, subreddits, GitHub repositories or other public sources you actually trust for that niche.

You can add a source by:

• pasting a link to the source
• searching a topic and previewing likely sources
• bulk-importing links or your list of sources from another reader

This is better than one giant reading pile because selection starts with a narrower source set.

2. Check what changed this week, not everything ever published

Open your Space and use Today view to see what your sources published today in your timezone. For weekly issues, scan daily through the week and save strong candidates as you go, instead of doing all curation at the last minute.

Switch layouts depending on the source:

• Card or Magazine for scanning visually
• List for speed
• Article when you want to focus on one piece

If a source produces too much noise, add feed filters to hide recurring items by keyword, author, link or length. That keeps routine announcements or low-value posts out of the way without deleting the source.

3. Triage with summaries, then verify in the original post

Open a post in the side panel or drawer and read the AI summary first to decide whether it deserves attention. For YouTube, Repurai can summarize from the transcript when captions are available.

Use the summary to triage, not to publish from memory. If the story looks promising, read or watch enough of the original source to verify the claim and extract the detail you want to cite.

A practical rhythm is:

• summary first
• original source second
• save if it passes your three tests

You can also use full-text search across everything your sources published to check whether a topic has appeared before, which helps when you want to show a pattern rather than treat every post as isolated news.

4. Save candidates and ask focused questions

Use Saved to Read for items that may make the issue. By the end of the week, your saved list becomes a short shortlist rather than a fog of browser tabs.

Then use Ask your sources to test your selection. Good prompts are narrow and practical.

Illustrative prompts:

• "What changed this week in my sources that would matter to B2B SaaS marketers? Cite the posts."
• "Across this Space in the past week, which stories include real evidence, figures or release details?"
• "What did these sources publish about open-source evaluation tools this month?"
• "What angles are people missing in this week’s announcements?"

Because answers come only from your own sources and cite the posts, this is useful for comparison and recall. It is not a substitute for checking the source yourself.

5. Draft your context, not just the link list

Once you have your shortlist, write each item in this order:

1. What happened
2. Why it matters to this audience
3. The detail that supports the claim
4. Your judgment or next step

If you want help turning notes into clean copy, Repurai can draft from an article, idea or chat answer for Substack, Ghost or Medium, in your own learned voice with your standing instructions. That is useful after selection is done. First choose the right stories, then write them clearly.

If your bottleneck is the writing pass after curation, the follow-up article on writing a newsletter with AI in your own voice covers that step.

A selection checklist you can copy

Illustrative checklist for each story:

• Does this fit my audience promise?
• Did I check the original source?
• Can I state the claim without exaggeration?
• Is there one concrete detail worth pulling out?
• Will the reader know what to do, think or watch next?
• Can I add a sentence of judgment or interpretation?
• If I cut this item, would the issue get weaker?

If you answer no to several of these, do not include it.

Worked example

Say you write a weekly tech operations newsletter for small software teams.

You find five candidate stories:

1. A vendor announces a new model
2. A long YouTube interview with an infrastructure founder
3. A GitHub release with a small but important deployment fix
4. A viral social post reacting to the vendor announcement
5. A technical blog post with benchmark details

A stronger issue might keep 1, 3 and 5, skip 4, and use 2 only if the interview contains a practical detail your audience can apply.

Your blurbs could look like this:

• Vendor launch: Big announcement, but the useful part for small teams is the new API behavior and migration path, not the demo.
• GitHub release: Easy to miss, but this patch fixes a deployment edge case that affects self-hosted users.
• Benchmark post: More useful than the launch thread because it shows where performance improved and where it did not.

Notice what happened: you did not reward volume of attention. You rewarded reader utility.

When this does not work

There are limits to any selection method.

• If your audience is too broad, relevance becomes vague and every story seems defensible.
• If your sources are poor, good selection cannot rescue weak input.
• If you rely on summaries alone, you will miss conditions and caveats.
• If a YouTube video has no captions available, transcript-based summarizing may not be possible.
• If your issue is pure entertainment or personality-driven reading, usefulness may matter less than taste. Even then, you still need a clear editorial lens.

If your main problem is gathering the week’s material in the first place, you may also want the follow-up on how to curate content for a newsletter.

If you want your sources, summaries and drafts in one place, you can try Repurai free.

FAQ

How many stories should a curated newsletter include?

Usually fewer than you think. Include only the number you can explain with real context. For many weekly issues, one lead story and three to five supporting items is enough.

Should I include a story everyone else already covered?

Yes, if your audience still needs it and you can add a better angle, clearer evidence or a more useful takeaway. Skip it if you are only repeating consensus without adding value.

Can AI choose stories for my newsletter automatically?

AI can help you scan, summarize and shortlist, but you still need to define relevance, verify source claims and add judgment. Selection quality comes from your editorial standard, not from automation alone.

If you want to set this up on your own sources, start here: /signup