First action: check one internal document or style page today for your organisation's plain language guidance. Teams may follow these principles without the label, such as active voice or reader needs.
Anchoring your practice to standards your workplace accepts means later revisions land well with colleagues, since they already recognise the style you are following.
What is a good first step for professional writing? Correct: reviewing an internal style document. Incorrect: rewriting an entire report from memory.
The next section walks you through the full practice loop for splitting long sentences.
Practice loop: short, linear sentences
Pick one sentence you wrote this week that runs longer than two lines, then split it so each new sentence carries one idea only.
Reread each split sentence and move the actor before the action, so the team reviewed the draft replaces the draft was reviewed by the team. Keep passive only when the doer is unknown or unimportant.
Circle any word a new colleague might not know and swap it for a common alternative, unless it is a necessary technical term your reader expects.
Check that each revised paragraph sticks to one topic; split any paragraph that covers two into two shorter ones.
Read the full revision aloud once. Note any sentence where you run out of breath as a candidate for another split.
Worked demonstration: revising one long message
Suppose a colleague wrote: Following our review of the scheduling concerns raised by several team members last week, we have determined that a revised rota will be circulated by the end of this week, which we hope will address the issues identified. That is one idea buried in 38 words.
The revision starts by splitting the sentence so each new one carries a single idea: We reviewed the scheduling concerns you raised last week. A revised rota will follow by Friday. Two sentences now carry the same content in 16 words, 9 and 7 respectively. We reviewed already puts the actor before the action, so it stays, while a version like concerns were reviewed by us would push the doer behind the action and weaken the sentence.
Then swap unfamiliar wording: circulated becomes sent out unless the team reliably uses the original term, and determined drops out once the split removes the need for it. Both sentences describe one outcome, so they stay together as a short paragraph; a separate note about who to contact would start its own paragraph.
Finally, read it aloud: We reviewed the scheduling concerns you raised last week. A revised rota will follow by Friday. No breath runs short, and the reader gets the answer on first reading, which is the aim of plain language at work.
Self-check and next adjustment
Run one observable check: count the words in each sentence of your revision, divide by the number of sentences, and note the average. General plain-language guidance puts a comfortable average around 15 to 20 words, as a benchmark rather than a strict per-sentence rule.
Then check topic focus: read each paragraph and ask whether it could be summarised in one short phrase. Generally, one topic per paragraph is the default; complex or technical material may justify a series of short paragraphs instead.
Example self-check result: the rota revision has two sentences of 9 and 7 words, so the average is 8 words. That sits below the 15 to 20 benchmark, but short averages are fine because the benchmark is guidance rather than a strict rule. The rota content is unchanged and both sentences share one outcome, so the paragraph passes. Write your own two numbers in a note; the count itself is the progress marker, not a feeling about the text.
One next adjustment: if your average runs well above 20 words, return to step 1 of the loop tomorrow and split the three longest sentences only, leaving everything else untouched. If your average already sits near 15, keep sentence length and instead adjust step 3, swapping jargon in one additional paragraph next session.
Repeat the same count after each session and compare this week's note with last week's. A falling or stable average near the benchmark, with each paragraph holding one topic, is the observable sign the routine is working.