← Back to Blog workflows

Preferred Sources in AI Search: A Content-to-Task Workflow for People Who Read Too Much

Glean Editorial · May 31, 2026 · 16 min read

# Google’s Preferred Sources feature is now live in AI Overviews and AI Mode, and early data shows users are twice as likely to click through to a trusted source. But if you read a lot of long-form content, this update creates a new problem: you’ll find more of the right articles, newsletters, and reports faster, and then you’ll have to decide what to do with them. This article walks through a Preferred Sources workflow that turns those links into tasks instead of adding them to a saved-content backlog.

Sources and Trend Signals Checked

Before we go further, here’s what we know for certain about Preferred Sources as of May 2026:

Google’s official announcement (May 2026) confirms Preferred Sources are now available in AI Overviews and AI Mode. Users who set Preferred Sources see content from those sources prioritized in AI-generated answers.
Click-through data from Google shows users are “twice as likely” to click through to a Preferred Source compared to a non-preferred source in search results. This is directional data from Google’s internal testing, not a third-party audit. For context, a typical click-through rate on a standard search result is around 2–3% for the top position; Preferred Sources may push that to 4–6% or higher, depending on the query.
Source adoption is significant: Google reports users have already selected more than 345,000 unique sources since the feature rolled out. That includes everything from major publications like The New York Times and The Wall Street Journal to niche blogs like Stratechery, Marginal Revolution, and The Diff.
Google Search agents at I/O 2026 (May 20, 2026) demonstrated how AI agents can now perform multi-step tasks across websites, including pulling data from Preferred Sources. For example, an agent could extract key statistics from a Preferred Source report and populate a Google Sheet, then email a summary to your team—all without you opening the article.

The Google Trends RSS feeds for May 31, 2026 show no direct mention of Preferred Sources as a trending topic in the US, France, UK, or India. The feature is not yet a mainstream conversation, but the adoption numbers suggest power users are already engaged. If you’re reading this, you’re likely part of that early adopter group—and you need a system to handle the incoming flood.

What Preferred Sources Actually Changes for Heavy Readers

If you read more than 10 long-form articles per week—industry reports, newsletters, academic papers, technical documentation—you already have a content management problem. You probably use bookmarks, read-later apps, or email folders to save things you intend to act on. The problem is that “intend to act on” rarely becomes “actually acted on.”

Consider a typical day: You open a newsletter from The Browser or The Morning Brew, find three interesting links, and save them to Pocket or Instapaper. Later, you search for something on Google, see a Preferred Source result from Ars Technica, click through, skim it, and save it. By the end of the week, you have 20–30 unread articles in your read-later app. By the end of the month, that number hits 100+. You feel guilty, overwhelmed, and eventually stop opening the app altogether.

Preferred Sources makes this worse in a specific way: it surfaces more relevant content from sources you already trust. Instead of sifting through generic search results, you get a curated feed of content from, say, The Atlantic, Stratechery, Ars Technica, or whatever sources you’ve selected. That’s great for discovery, but it means your read-later list grows faster.

The decision you face is simple: do you keep treating every interesting link as something to “save for later,” or do you build a workflow that converts incoming content into tasks immediately?

The Content-to-Task Workflow: A Decision Table

Here is a decision table for handling any piece of content that arrives via Preferred Sources. Use it when you open a link.

Content type | Action | Tool | Time threshold

The key threshold is 10 minutes. If you spend more than 10 minutes reading a single piece of content without creating a task, you are building backlog, not building knowledge.

Example: You open a Stratechery article about the economics of AI search. It’s 3,000 words. Instead of reading the whole thing, you skim the first two paragraphs, the section headings, and the conclusion. You find one key insight: “Google’s cost per AI search query is 10x higher than a traditional search.” You create a task: “Include Stratechery cost data in Q3 budget presentation.” Due date: next Monday. Link attached. Total time: 8 minutes. You archive the article. Done.

How to Set Up Your Preferred Sources Workflow

Step 1: Audit Your Current Sources

Open your browser history, read-later app, and email newsletter subscriptions. List every source you regularly consume. Be honest: most people have 15 to 30 sources but only actively read 5 to 7. Cut the rest.

Action: Delete or unsubscribe from any source you haven’t opened in the last 30 days. For newsletters, use the “unsubscribe” link. For bookmarks, use a bulk delete. For RSS feeds, remove any that haven’t produced a click in the last two weeks.

Concrete example: Let’s say you subscribe to 12 newsletters: Morning Brew, The Hustle, Stratechery, The Diff, Axios AM, Politico Playbook, The New York Times Morning, The Guardian Weekly, The Economist Espresso, TechCrunch Daily, The Information, and The Markup. You open Stratechery, The Diff, and The Information regularly. The others you mostly delete. Unsubscribe from the nine you don’t read. That’s nine fewer emails per day, and nine fewer chances to save a link you’ll never open.

Step 2: Select Your Preferred Sources in Google

Go to your Google Search settings. Under “Preferred Sources,” add the 5 to 10 sources you actually read. Google’s documentation says you can select up to 20, but more than 10 dilutes the signal.

Threshold: If you cannot name 5 sources you trust implicitly, start with 3. You can always add more.

How to choose your sources:

Relevance: Does this source cover topics you need for work or personal projects?
Accuracy: Has the source been fact-checked or corrected errors in the past?
Depth: Does the source provide original analysis, or does it just aggregate others’ work?
Frequency: Does the source publish often enough to be useful, but not so often that it overwhelms you?

Example list for a tech executive:

Stratechery (tech strategy)
Ars Technica (tech news and reviews)
The Information (tech industry scoops)
Harvard Business Review (management and strategy)
The Economist (global affairs and economics)

Example list for a policy researcher:

The New York Times (general news)
The Atlantic (long-form analysis)
Politico (policy and politics)
Foreign Affairs (international relations)
The Brookings Institution (think tank reports)

Step 3: Create a Content-to-Task Trigger

Every time you open a piece of content from a Preferred Source, you need a trigger that forces a decision. The trigger can be a timer, a browser extension, or a mental rule.

Example trigger: “If I have not created a task from this article within 5 minutes of opening it, I close the tab and move on.”

This is aggressive, but it works. The alternative is what most people do: open the article, skim it, feel like you learned something, close it, and forget it. That is not learning. That is entertainment.

How to implement the trigger:

Use a physical timer: Set a 5-minute egg timer on your desk. When it rings, close the tab.
Use a browser extension: Install a tab limiter that warns you after 5 minutes on a single page.
Use a mental rule: Practice the “one-touch rule”—you touch the article once, decide, and act. No second chances.

Example in practice: You open a The Atlantic article about remote work trends. You read the headline and first two paragraphs. You realize it’s not directly relevant to your current project. Instead of saving it “just in case,” you close the tab. Total time: 45 seconds. No backlog created.

Step 4: Use an App That Bridges Content and Tasks

This is where most workflows break. You read something, you think “I should do something with that,” and then you either save the link to a read-later app (which creates a backlog) or you do nothing.

Glean is the app that turns preferred articles, newsletters, videos, and posts into todos immediately instead of letting trusted content become a bigger backlog. Instead of saving a link to a folder, you save the link as a task with a due date, a priority, and a note about what to do next.

Example: You read a Stratechery article about AI search agents. In Glean, you create a task: “Summarize Stratechery AI agents article for team meeting on Friday.” You set a due date of Thursday. The link is attached. Now the content has a purpose.

Other tools that can work:

Todoist: Create a project called “Content to Process.” Add links as tasks with due dates and labels (e.g., @reading, @reference).
Notion: Create a database with fields for link, summary, action item, due date, and status (To Do, In Progress, Done).
Obsidian: Create a daily note; for each article, write a one-sentence summary and a link to a task in your task manager.

Why Glean is recommended: It’s designed specifically for this workflow. It integrates with your browser, email, and calendar. It automatically extracts the article title and URL. It forces you to assign a due date before saving. There is no “save for later” button—only “save as task” or “archive.”

Step 5: Schedule a Weekly Content Cleanup

Even with a good workflow, some content will slip through. Schedule 30 minutes every Friday to review your Glean inbox and your browser tabs. For each item, do one of three things:

Convert it to a task with a due date.
Archive it permanently.
Delete it.

Threshold: If you have more than 20 items in your inbox at the end of any week, your trigger is too weak. Tighten the 5-minute rule.

Concrete cleanup checklist:

[ ] Open Glean inbox. Sort by “no due date.”
[ ] For each item: read the title. If you remember why you saved it, assign a due date within the next 7 days. If you don’t remember, archive it.
[ ] Open browser tabs. Close any tab that has been open for more than 24 hours.
[ ] Check your email inbox for any newsletters you saved but didn’t read. Delete or archive them.
[ ] Review your read-later app (if you still use one). Delete any article older than 30 days.
[ ] Count your total unprocessed items. If it’s over 20, adjust your daily trigger.

Why Preferred Sources Makes This Workflow Necessary

Before Preferred Sources, search results were a firehose of mixed-quality content. You had to spend time filtering. Now, the filtering is done for you, but the volume of relevant content goes up.

Google’s data shows users are twice as likely to click through to a Preferred Source. That means for every search you do, you are more likely to open an article. If you search 10 times per day, you might open 5 to 8 articles instead of 3 to 4. That is a 50% to 100% increase in incoming content.

Real-world scenario: Before Preferred Sources, a product manager at a SaaS company might search “AI customer support trends” and open 3 results from various blogs. With Preferred Sources set to Stratechery, Intercom, and Zendesk, the same search now returns 5 results, all from trusted sources. The PM clicks all 5. That’s 5 articles to process instead of 3. Without a workflow, those 2 extra articles per search add up to 10–20 extra articles per week.

If you do not have a workflow, that content becomes a backlog. Backlogs cause anxiety, reduce decision quality, and make you feel like you are behind. The solution is not to read less; it is to process faster.

The Read-Later Graveyard Problem

The read-later app industry is built on a lie: that saving something for later means you will read it later. Data from multiple read-later apps (not cited here because the numbers are proprietary) suggests that 60% to 80% of saved articles are never opened again.

Your read-later list is a graveyard of good intentions.

Why this happens:

Saving is easy; reading is hard. It takes 2 seconds to save an article. It takes 10 minutes to read it. The effort is asymmetric.
Context is lost. When you save an article, you often forget why you saved it. By the time you open it again, it’s no longer relevant.
Volume overwhelms. As your list grows, you feel paralyzed. You don’t know where to start, so you start nowhere.
No accountability. There’s no consequence for not reading a saved article. No one is checking. The guilt fades.

How Glean solves this:

Glean makes the default action a task, not a save. When you save a link in Glean, you must assign it a due date or a project. There is no “save for later” option. There is only “save as a task” or “archive.”

This forces you to decide immediately whether the content is worth acting on. If it is not, archive it. If it is, assign a due date. The backlog never forms.

Example comparison:

Without Glean: You find a Harvard Business Review article on negotiation tactics. You save it to Pocket. Six months later, you open Pocket, see 200 unread articles, and close the app. The article is never read.
With Glean: You find the same article. You create a task: “Read HBR negotiation article and extract 3 tactics for next client call.” Due date: tomorrow. You read it, extract the tactics, and archive the task. The knowledge is applied.

Step-by-Step Checklist: Turn a Preferred Source Article Into a Task

Use this checklist every time you open a link from a Preferred Source.

Open the article. Read the headline and the first two paragraphs. Decide if it is relevant to a current project or decision.
If yes: Open Glean. Create a new task. Paste the link. Write a one-sentence action item. Set a due date within the next 7 days.
If no: Close the tab. Do not save the link. Do not bookmark it. Do not email it to yourself.
If maybe: Give yourself 5 minutes to skim the article. If you find one actionable point, create a task. If not, close the tab.
Archive the task after you complete it. Do not keep completed tasks in your inbox. Move them to a “Done” folder or delete them.

This checklist takes 30 seconds to 2 minutes per article. That is faster than the time you spend deciding whether to save something for later.

Example with a real article:

You search “remote work productivity 2026” and see a Preferred Source result from The New York Times. You open it.

Headline: “Remote Work Productivity Dips 5% in 2026, Study Finds.” First two paragraphs: The study from Stanford shows a small decline, but it varies by industry.
Decision: You’re writing a report on remote work for your company. This is relevant.
Action: Open Glean. Create task: “Include NYT remote work productivity data in Q2 remote work report.” Due date: next Wednesday. Link attached.
Time spent: 1 minute. You close the tab. The article is processed.

How AI Search Agents Change the Equation

Google’s I/O 2026 announcement included a demo of Search agents that can perform multi-step tasks across websites. For example, an agent can find a Preferred Source article, extract data, and populate a spreadsheet.

This is powerful, but it creates a new risk: you can now automate the collection of content without automating the decision about what to do with it. If you set up an agent to pull articles from your Preferred Sources every day, you will have a daily digest of 10 to 20 articles. Without a workflow, that digest becomes a backlog.

Action: If you use Search agents, set a maximum number of articles per day. Start with 3. If you can process 3 articles into tasks, increase to 5. Do not exceed 5 unless you have a dedicated processing block on your calendar.

Concrete example: You set up a Search agent to find all The Information articles about AI startups published in the last 24 hours. The agent returns 8 articles. You have 30 minutes for processing. You skim the headlines and pick the 3 most relevant. You create tasks for each. The other 5 are archived. You do not read them. This is disciplined curation.

Another risk: AI agents might surface content that is not actually useful, just because it matches your Preferred Sources. For example, an agent might pull a The Atlantic article about a niche historical event that has nothing to do with your work. Trust your judgment, not the agent’s.

The 345,000 Sources Problem

Google says users have selected more than 345,000 unique sources. That number is impressive, but it also means the feature is being used in a fragmented way. Most users have selected a handful of sources. A small number of power users have selected 20.

If you are a power user with 20 Preferred Sources, you are seeing a lot of content. You need a more aggressive workflow.

For power users: Set a daily limit of 10 articles from Preferred Sources. Use Glean to track how many tasks you create per article. If you create fewer than 1 task per 3 articles, you are reading too much and acting too little.

How to audit your source list:

Review your Preferred Sources every 30 days.
For each source, ask: “Did I create at least one task from this source in the last month?”
If no, remove the source. You can always add it back later.
If yes, keep it. But consider whether the source is producing high-quality tasks or just busywork.

Example: You have 20 sources. After 30 days, you find that 5 sources produced 80% of your tasks. The other 15 produced only 20%. Remove the bottom 10 sources. Now you have 10 sources, and your signal-to-noise ratio improves.

FAQ: Preferred Sources and Content-to-Task Workflows

1. How do I set up Preferred Sources in Google Search?

Go to your Google Search settings. Look for “Preferred Sources” under the “Personalization” or “Search results” section. You can add up to 20 sources by URL. Google’s documentation recommends starting with 5 to 10.

Step-by-step:

Open Google Search.
Click the gear icon (Settings) in the top right.
Select “Search settings.”
Scroll to “Preferred Sources.”
Click “Add source.”
Enter the URL of the source (e.g., https://stratechery.com).
Repeat for up to 20 sources.
Click “Save.”

Note: The feature is available in AI Overviews and AI Mode. If you don’t see the option, check that you’re signed in to your Google account and that the feature is rolled out in your region.

2. Will Preferred Sources work in AI Overviews and AI Mode?

Yes. Google confirmed in May 2026 that Preferred Sources are now available in AI Overviews and AI Mode. When you ask a question, the AI will prioritize content from your selected sources.

Example: You ask Google, “What are the latest trends in AI regulation?” If you have The Verge, TechCrunch, and The New York Times as Preferred Sources, the AI Overview will pull from those sources first. The result might include a summary of a The Verge article, a quote from TechCrunch, and a link to The New York Times.

3. How do I know if a source is trustworthy enough to be a Preferred Source?

Use the “three-article test.” Read three articles from the source. If all three are accurate, well-sourced, and relevant to your work, add it. If one of three is wrong or misleading, do not add it.

Additional criteria:

Transparency: Does the source disclose conflicts of interest, funding, or corrections?
Expertise: Are the authors credible? Do they have relevant credentials or experience?
Consistency: Does the source maintain a consistent level of quality over time?
Independence: Is the source free from obvious bias or sponsorship?

Example: You’re considering adding The Onion as a Preferred Source. You read three articles. They are all satirical and not factual. You do not add The Onion to your Preferred Sources, because it fails the accuracy test. (Though you might enjoy it for entertainment, it’s not suitable for knowledge work.)

4. What if I have too many Preferred Sources?

Cut to 5. The feature works best with a small, curated list. If you have 20 sources, you are not curating; you are hoarding.

How to cut:

List your 20 sources.
Rank them by how many tasks you created from each in the last month.
Keep the top 5.
Remove the rest. You can always add them back later.

Why 5 is the sweet spot: With 5 sources, you can process every article they publish. With 20, you’ll miss most of them, and the feature becomes noise.

5. How do I stop saving articles for later and start acting on them?

Change your default action from “save” to “task.” Use an app like Glean that forces you to assign a due date. If you cannot assign a due date within 10 seconds, archive the article.

Psychological trick: Remind yourself that “save for later” is a lie. Studies show that 60–80% of saved articles are never read. By archiving instead of saving, you are not losing anything—you are freeing yourself from guilt.

6. What if I need to reference an article later but don’t have a specific task for it?

Create a “reference” task with a due date of “someday” or a project called “Reference Library.” But be careful: this can become a new backlog. Limit yourself to 10 reference tasks at any time.

Better approach: Save the link to a reference manager like Zotero or Notion, but do not add it to your task list. Reference managers are for storage; task managers are for action. Keep them separate.

7. How do I handle videos and podcasts from Preferred Sources?

Use the same workflow. For videos, watch the first 2 minutes. If relevant, create a task with a timestamp. For podcasts, read the show notes or transcript. Extract one actionable point.

Example: You find a YouTube video from The Verge about AI hardware. You watch the first 2 minutes. The host mentions a new chip from NVIDIA. You create a task: “Research NVIDIA AI chip specs for hardware comparison report.” Due date: next week. Link attached with timestamp 1:45.

8. What if I’m not using Glean? Can I still use this workflow?

Yes. The workflow works with any task manager (Todoist, Asana, Notion, etc.) or even a paper notebook. The key is the decision table and the 5-minute trigger, not the specific tool.

Paper notebook example: Keep a small notebook on your desk. When you open an article, write the title and a one-sentence action item. Set a reminder on your phone to review the notebook at the end of the day. Transfer action items to your calendar.

When to Ignore Preferred Sources

Preferred Sources are useful, but they are not a replacement for serendipity. If you only read content from sources you already trust, you will miss new voices, contrarian arguments, and emerging trends.

Rule of thumb: Read 80% of your content from Preferred Sources and 20% from outside your bubble. Use the 20% to challenge your assumptions. Use the 80% to build depth.

How to find outside content:

Follow links in Preferred Source articles to other sources.
Search for topics you know nothing about and read the top results, even if they’re not from Preferred Sources.
Subscribe to one newsletter that is intentionally outside your field (e.g., if you’re in tech, subscribe to a history or art newsletter).
Use Google’s “Explore” tab or “Discover” feed to find new sources.

Example: You’re a software engineer who reads Ars Technica, Hacker News, and The New York Times. Once a week, you read The New Yorker for long-form journalism on topics like politics, culture, or science. This gives you a broader perspective and might spark ideas you wouldn’t get from tech sources alone.

The One Metric That Matters

Track your content-to-task conversion rate. This is the percentage of articles you open that result in a task with a due date.

If your rate is below 20%, you are reading too much and acting too little.
If your rate is between 20% and 40%, you are in a healthy range.
If your rate is above 40%, you are probably creating too many low-value tasks.

How to calculate:

For one week, count every article you open from a Preferred Source. (Use a browser extension like Toggl or a simple tally in a notebook.)
Count every task you create from those articles.
Divide tasks by articles. Multiply by 100 to get your percentage.

Example: You open 50 articles in a week. You create 12 tasks. Your conversion rate is 24%. That’s healthy.

Action: For one week, count every article you open from a Preferred Source. Count every task you create. Calculate your rate. Adjust your reading habits accordingly.

If your rate is too low: Tighten your trigger. Use a 3-minute timer instead of 5. Be more selective about what you open.

If your rate is too high: You might be creating tasks for trivial content. Ask yourself: “Will this task matter in a week?” If no, archive the article instead.

Final Workflow Summary

Audit your sources. Cut to 5 to 10. Remove any source you haven’t read in 30 days.
Set up Preferred Sources in Google Search. Add your curated list.
Use a 5-minute timer per article. If you haven’t created a task in 5 minutes, close the tab.
Convert every article into a task with a due date. Use Glean or another task manager.
Schedule a weekly cleanup. Spend 30 minutes every Friday processing leftovers.
Track your content-to-task conversion rate. Aim for 20–40%.

Glean is the app that turns preferred articles, newsletters, videos, and posts into todos immediately instead of letting trusted content become a bigger backlog. Instead of saving links to a read-later graveyard, you save them as tasks with deadlines and context. The backlog stops growing.

Try Glean Free.