Writing that gets read

The main problem with workplace writing is not grammar or vocabulary. It is that the reader cannot tell within a few seconds what the document wants from them.

Put the conclusion first

The single highest-return change.

Academic and legal writing builds toward a conclusion. Most workplace writing is read by someone deciding within seconds whether to continue, and the build never happens.

State the outcome, the request or the recommendation in the first sentence. Then explain.

This feels wrong to people trained to construct an argument, and it is what a reader who may stop at any point needs.

The test: if the reader read only the first two lines, would they know what you want?

Say what you want the reader to do

An enormous proportion of workplace messages describe a situation without stating what response is expected.

The reader then has to infer whether this is information, a request, a decision, or a warning — and frequently infers wrongly or does nothing.

Stating it explicitly, near the top, costs one line: no action needed; please approve by Thursday; I need a decision on option A or B.

One document, one purpose

A message containing an update, a request and a complaint achieves none of them.

The request gets lost in the update, the complaint colours the request, and the reader responds to whichever they noticed.

Separate messages are less efficient to write and considerably more effective.

Length

The correlation between length and being read is strongly negative, and most people write long because it is faster than writing short.

Practical constraints that work: a message that fits on a phone screen without scrolling; a document with a one-paragraph summary that stands alone; a proposal where the first page contains the whole argument.

Editing to half the length is almost always possible and almost always improves it.

Concrete beats abstract

Abstraction is where workplace writing goes to die.

"We should improve alignment on delivery priorities" says nothing. "Marketing and engineering disagree about which feature ships first, and I need a decision this week" says something.

The test is whether a sentence could appear in a document about a completely different subject. If it could, it is not carrying information.

The specific habits worth breaking

Throat-clearing. The first paragraph that explains that you are writing to explain something. Delete it; the document usually starts at paragraph two.

Hedging. "It might be worth potentially considering whether we could perhaps look at" is one clear sentence wearing several coats.

Hedge where you are genuinely uncertain and say so directly.

Nominalisation. Turning verbs into nouns — "conduct an investigation" rather than "investigate" — which lengthens sentences and removes the actor.

Passive voice where the actor matters. "Mistakes were made" is the canonical example and the pattern is everywhere.

Passive is fine where the actor is genuinely unimportant. It is a problem where it is being used to avoid saying who.

Jargon used to signal rather than to communicate. Technical vocabulary is efficient among people who share it and is a barrier otherwise.

Structure for scanning

Readers scan before they read. Writing that supports scanning gets read; writing that does not gets skipped.

Meaningful headings that state content rather than category. Short paragraphs. Lists where the content is genuinely a list. Bold on the key phrase rather than on whole sentences.

The point is that a reader skimming should extract the argument from the headings alone.

The subject line

The most read and least considered part of any email.

It should state the topic and, ideally, the action. "Budget approval needed by Friday" is a subject line. "Quick question" is not.

Before sending

Read the first two sentences and ask whether they state the point.

Read the whole thing and delete the ten percent that adds least.

Check that the action required is explicit and near the top.

Three checks, perhaps ninety seconds, and they do more for how your writing lands than any amount of vocabulary.