Stop reading everything: how to write a personal mandate

How do you stop reading everything without missing what matters?

You stop reading everything by writing down what your reading is for. A personal mandate, one page with the three to five questions you need answered, lets you choose sources against those questions, set a reading cadence for each one, and decide in advance what deserves an interruption, what can wait for a weekly read, and what you will ignore.

What follows is the practice: writing the mandate, testing sources against it, setting tiers, a copyable template, a worked example and a monthly review.

Why does a feed get worse as you add sources?

A feed gets worse as you add sources because each new source answers fewer new questions than the last one while adding just as many items to sort. The first source on a subject teaches you a lot. The tenth mostly overlaps with the other nine, and its unique contribution is buried among the repeats.

The deeper problem is that most sources are added for the wrong reason. People subscribe because a source is good, not because it answers a question nobody else on their list answers. Quality is necessary but not sufficient. A brilliant newsletter about a subject you do not need to follow is still a cost: it takes time, and it trains you to skim.

Without a written mandate, there is no test a source can fail.

What is a personal mandate?

A personal mandate is a one-page document that states the questions your reading must answer, the decision each question supports, and the signals that would change the answer. It is the brief you would hand to a research assistant on their first day.

The key word is questions. A list of subjects ("semiconductors", "our competitors", "interest rates") tells you where to look. It cannot tell you when you have read enough, because there is always more about a subject. A question can be answered, partly answered, or left unchanged by a new item, and that is what lets you stop.

Compare the two:

• Subject: "AI regulation."
• Question: "Will the rules on AI in hiring decisions require us to change our screening tool within the next 12 months?"

The second version excludes most of what is written about AI regulation, and that exclusion is the point.

How do you write your own mandate?

You write a mandate by working backward from your decisions, not forward from your interests. Five steps are enough.

1. List the decisions you will make in the next six to twelve months that depend on outside information: a contract to renew, a product to price, a position to hold, a hire, a market to enter.
2. Turn each decision into one question with a horizon. "Will X happen before Y?" or "Is X getting better or worse for us?" Keep the ones where the answer could change what you do.
3. Name the signals that would move each answer. For each question, write two or three concrete events that would confirm or reverse your current view. These become your interruption criteria later.
4. Write a "not my question" line. Name the adjacent material you will deliberately skip, such as general commentary, the trending angle on your subject, or research that will not mature within your horizon.
5. Cap the list at five. If you have more, merge or drop. A sixth question usually means one of the first five is really two subjects pretending to be one.

The draft should take under an hour.

How do you choose sources against your mandate?

You choose sources by mapping each one to the questions it answers, then cutting anything that maps to nothing. A simple grid works: questions across the top, sources down the side, a mark where a source has a realistic chance of moving an answer.

Read the grid with three rules:

• Every source must answer at least one question. A source with no marks goes, however much you enjoy it. Move it to leisure reading if you like, but take it out of the system.
• Every question needs at least two independent sources. One source gives you no way to tell whether it is late or silent. Three outlets rewriting the same press release count as one.
• At least one source per question should sit where the answer is first written down: the company, the regulator, the data publisher, the practitioner closest to the work.

Note each source's rhythm too: a twice-yearly statistical release and a founder who posts daily belong on the same list, not in the same reading slot.

What deserves an interruption, a weekly read, or nothing?

Only a development that could change a decision before your next scheduled read deserves an interruption. Everything else either waits for the weekly read or does not get read at all. Deciding the tiers in advance is what turns a reading habit into a system.

The three tiers:

• Interrupt now. One of the signals you named in step 3 has occurred, and acting a few days late would cost you. For most knowledge workers this should happen a few times a month, not a few times a day.
• Read this week. A development that updates the answer to a question without forcing an immediate decision. These arrive in one batch, after the story has settled, with the context that makes them readable.
• Nothing. True, interesting, and outside the mandate. This tier should be the largest by far. Ignoring it is the system working, not failing.

Cadence follows the question's clock speed. A question about a quarterly earnings trend does not need a daily check. A question about a pending regulatory decision might need a weekly one until the decision lands, then a monthly one. Setting cadence per question, not per day, is how a research routine stops consuming the day.

A one-page personal mandate template

Copy this structure into a note and fill one block per question. Keep the whole thing on a single page.

• Reader and role: who you are and what you are responsible for, in one line.
• Question 1: one sentence, answerable, with a horizon.
• Decision it feeds: what you will do differently depending on the answer.
• Current answer: your view today, in one line.
• Interrupt if: two or three concrete signals that justify an alert.
• Weekly read: the kinds of developments that update the answer.
• Sources: two to four, with at least one where the answer is first written.
• Cadence: how often this question needs checking.
• Questions 2 to 5: the same fields.
• Not my questions: the adjacent topics you will skip on purpose.
• Review date: the day each month you will edit this page.

The "current answer" field is the one most people skip, and the one that matters most. Without it, you cannot tell whether a new item changed anything.

What does a finished mandate look like?

Consider a hypothetical head of green coffee buying at a mid-sized specialty roaster in the US. She signs forward contracts with importers, sets wholesale prices for cafés twice a year, and has two origins that supply most of her volume. Her mandate fits on one page.

Question 1: Will green coffee costs for next year's contracts be materially higher than this year's?

• Decision: how much volume to lock in early, and at what price.
• Interrupt if: the arabica futures market moves past the level built into her budget, or a major producing country reports a frost or drought that hits the coming harvest.
• Weekly read: crop estimates, shipping and freight changes, importer market notes.
• Sources: the USDA Foreign Agricultural Service's Coffee: World Markets and Trade report, published twice a year in June and December; two importers' market updates; a commodity analyst who covers softs.

Question 2: Will either of her two main origins become harder to buy from because of traceability or import requirements?

• Decision: whether to qualify a third origin this year.
• Interrupt if: a regulator or a large buyer publishes a final requirement that her current supply chain cannot document.
• Sources: the relevant regulators' own releases, her importers' compliance notes, one trade association.

Question 3: Are the roasters competing for her wholesale accounts changing prices or terms?

• Decision: her next wholesale price list.
• Interrupt if: a direct competitor cuts wholesale prices in her region.
• Sources: three competitors' wholesale pages and newsletters, two café owners she trusts.

Question 4: Is café demand shifting toward formats she does not roast well for, such as cold brew concentrate or decaf?

• Decision: whether to invest in a new product line.
• Cadence: monthly, because the answer moves slowly.

Not my questions: health studies about coffee, celebrity café openings, general food inflation commentary, coffee culture debates on social media.

Under this mandate, a viral post about a famous chef's new café is tier three. A frost report from her main origin is tier one. A competitor's new subscription box is tier two, read on Friday.

How do you review the system every month?

You review the system with a fixed thirty-minute session that audits four things and ends with an edited page. The review is what keeps a mandate from becoming a document you wrote once and forgot.

1. Audit the interruptions. Count the alerts you received and how many led to an action or a changed plan. If fewer than half did, the criteria are too loose.
2. Audit the sources. Mark which sources contributed to any answer this month. Three silent months in a row is grounds for removal, except for sources that publish rarely by design, such as a regulator or a twice-yearly report.
3. Audit the questions. Retire questions that have been answered or whose decision has passed. Add the question your work actually raised this month.
4. Audit the misses. Write down anything relevant you learned late, and where you learned it. Each miss points to a missing source or a missing question, and fixing it is the most valuable edit of the session.

Once a year, rebuild the grid from a blank page. Source lists accumulate habits, and starting over is the only reliable way to see them.

Where does software fit in a personal mandate?

Software fits after the mandate, not before it. A tool cannot tell you which questions matter, but once they are written down, it can do the reading that a source grid of thirty or forty entries requires every week.

The mandate translates directly into a setup. The questions decide the topics, the grid decides the sources, and the tiers decide what arrives as an alert and what arrives as a brief. Kindal is one system built this way: you choose the sources or type a topic and let it propose them, it checks them every hour, every six hours or once a day, and it writes a short brief only when something changes, with each line linked to the post, filing or video behind it. If your questions are mostly about competitors and market share, dedicated tools that monitor your market are another route.

Whatever does the reading, the page stays yours.

What changes when your reading has a mandate?

When your reading has a mandate, every item has to answer to a question, and most items cannot. That is the relief: you stop reading out of obligation and start reading because something moved an answer you care about.

The page is short on purpose. Three to five questions, a few sources each, a rule for interruptions and a monthly edit. It is a small amount of writing that replaces a large amount of reading, and it makes quiet weeks feel like what they are: weeks in which nothing changed your decisions.