Inbox Triage Ruleset

15 views · 0 copies · 0 likes

Inbox advice fails because it is generic. Every guide recommends the same four folders and the same two minute rule, and none of it survives contact with your actual mail, because your inbox is shaped by your job. A support lead, a founder and a researcher have almost nothing in common in what arrives or what a reply costs.

This prompt does not give you a system. It derives one from a sample of your real inbox. You paste twenty to thirty recent subject lines and senders, and it reverse engineers the categories that genuinely exist in your mail rather than the ones a productivity book imagined.

The important design choice is that rules must be decidable from the subject and sender alone. A rule requiring you to open and read the message to classify it has already cost you the thing you were trying to save. Triage and processing are different activities, and mixing them is why inbox sessions expand to fill an afternoon.

The categories it produces are defined by required action rather than topic. Topic folders feel organised and help nothing, because you still have to decide what to do when you open one. Action categories carry their own instruction. Reply under two minutes, needs a scheduled block, waiting on someone else, reference only, and no action ever - that last one usually being the largest and least acknowledged group.

The honest section is the volume analysis. The prompt is instructed to identify what fraction of the sample should never have reached you, and to distinguish messages you should unsubscribe from, messages that indicate a broken process elsewhere, and messages arriving because you are cc-ed by habit. Fixing the source removes mail permanently, while sorting it faster only makes you better at absorbing a cost you should refuse.

One practical warning. Do not paste message bodies. Subject lines and sender names are enough for triage design, and pasting full contents of correspondence into a chat interface risks exposing information belonging to other people who did not consent. Redact client names if your subject lines contain them.

Expect the ruleset to need one revision after a week. The first version usually has one category that never gets used and one that catches too much. That is normal, and a ruleset that survives revision is worth far more than a perfect system you abandon.

You design a personal email triage ruleset by analysing real inbox data. Do not recommend a generic system or any named productivity methodology.

INPUTS
My role: [job title and what I am accountable for]
Inbox sample: [paste 20 to 30 recent subject lines with sender name or type. Do NOT paste message bodies. Redact names where needed]
Current pain: [what specifically goes wrong - things missed, time lost, anxiety]
Time available for email daily: [minutes]

SECTION 1 - PATTERN ANALYSIS. Group the sample into the categories that actually exist in this inbox. Name each category by the ACTION it requires, never by topic. For each, give the count from the sample, the percentage, and the sender pattern that identifies it.

SECTION 2 - TRIAGE RULES. For each category, write a rule that can be applied from the subject line and sender ALONE, without opening the message. If a category cannot be identified without opening the mail, say so explicitly and explain what would need to change - a sender convention, a filter, a subject prefix.

SECTION 3 - TIME BUDGET. Given my stated daily minutes, allocate time per category. If the arithmetic does not fit, say plainly that this inbox cannot be processed in the available time and state what must be cut, delegated, or refused. Do not produce a plan that only works on a perfect day.

SECTION 4 - SOURCE REDUCTION. Identify what fraction of the sample should never have arrived. Separate into: unsubscribe candidates, symptoms of a broken process elsewhere, and habitual cc. For each, state the specific intervention that stops it at the source.

SECTION 5 - THE ONE RULE. If I adopt only a single rule from all of this, which one, and what evidence in the sample supports that choice.

RULES
Every rule must be decidable from subject and sender only.
Never recommend a named methodology or an app.
Base all claims on the sample provided. If the sample is too small or unrepresentative to support a conclusion, say so.