Clear communication is judged by what the reader does next, not by how the text sounds. A notice is clear when the person it is meant for can find it, read it once, and act on it correctly. The chapters follow that test in order: who reads, how to order the page, which words to use, how to lay it out, how to check it with real readers, and what changes when a misread is costly.
Each chapter opens with the short version. Tap one to read the detail.
Plain means usable on the first read
~2 min
Plain language is writing your intended reader can find, understand, and use the first time. For federal agencies it is also law: the Plain Writing Act of 2010 requires it for documents the public relies on, though no court can enforce it.
The standard is set by the reader, not by a grade level. A page for small-business owners and a page for researchers should read differently, and both can be plain. The common misconception is that plain means simplified or less precise. It means removing effort the reader gets nothing for, such as buried verbs, undefined abbreviations, or a deadline in the sixth paragraph, while keeping every requirement intact.
That gives you a test for any rewrite. Ask of both versions, "What must I do, and by when?" If the answers differ, the edit changed the rule, not just the wording.
Federal agencies also have a legal duty. The Plain Writing Act covers the letters, forms, notices, and instructions people use to get a federal benefit or service, file taxes, or comply with a federal requirement. Regulations are excluded, though a separate executive order asks that they be plain too. Agencies must train staff and publish a yearly compliance report, but courts cannot review compliance, so the pressure comes from oversight and from readers, not lawsuits.
Rule: judge a document by whether its intended reader can act on it correctly after one read, and treat any rewrite that changes a duty as a defect.
Write for one real reader with a task
~2 min
Before writing, name the person who will act on the document and what they came to do. Write to that one reader, in their words, and split the text when different readers need different things.
"The public" is not a reader. "A spouse applying for survivor benefits on a phone at lunch" is, and naming that person settles most later choices: which terms need explaining, what goes first, how long a section can run. Find your real readers in the questions people already send you and the words they type into site search.
The usual mistake is believing a general audience is safer. An abstract audience makes every sentence feel relevant, because no particular person is there to get lost. Reading skill also varies widely, and the people who most need a notice often have the least time to decode it. Aim at the reader who will struggle; clear text costs a skilled reader nothing.
When two audiences need different things, mixing them makes everyone filter out sentences meant for someone else, and some will follow a rule meant for another group. Separate the threads: "If you are the property owner, you must… If you are the tenant, you must…"
Rule: if you cannot name the reader and the action they came for in one sentence, you are not ready to write.
Answer first, then follow the reader's path
~2 min
Put the purpose and the bottom line at the top, order the rest by the steps the reader will take, and write headings as the reader's own questions so someone scanning can land on their answer.
The common mistake is organizing by what the writer knows: history, then background, then the requirement. Readers arrive with a question and often stop after the first screen. So invert it. State what the document is for and what the reader must do at the top, and move background toward the end.
After that, follow the reader's path. Process steps read best in the order they happen, such as apply, submit, review, decide, not grouped by the office that handles each one. Put the general rule before its exceptions so most readers can stop early. A condition that decides whether a passage applies at all, such as "If you filed before June 1," goes first, so readers it does not fit can move on.
Headings are the reader's table of contents. "Applications" says a topic exists; "How do I apply?" tells readers they have found their answer. Question headings work when you know what people will ask, statement headings come next, and vague labels like "General" come last. Keep headings shorter than the text beneath them, use three levels at most, and draft them first: headings that make no sense alone reveal an organization problem while it is cheap to fix.
Rule: the first screen should answer the most common question, and every heading should name something a reader might be looking for.
Short sentences, one idea each
~2 min
Long sentences fail because readers must hold every clause until the point arrives. Give each idea its own sentence, cut words that carry nothing, keep each paragraph to one topic, and break long text into labeled chunks.
A sentence works like a queue. Readers hold its opening in mind until the main point arrives, and every clause stacked in between makes the wait longer, so something gets dropped. The common mistake is hunting for shorter words. The real fix is a split: the condition in one sentence, the result in the next, linked by a signal such as "In that case."
Then cut what carries nothing. Intensifiers like "really" and "very" go first. Wordy phrases built on prepositions come next: "in order to" means "to," and "a sufficient number of" means "enough." Paired words such as "due and payable" make careful readers hunt for a difference that is not there.
Negatives need the same scrutiny. "We will not process applications unless no balance is owed" makes readers cancel two negatives; "We process your application once your balance is paid" states it. An exception to an exception cannot be fixed by rewording at all; it needs a table or a list of cases.
Federal guidance suggests paragraphs of three to eight sentences, about 150 words, each opening with a sentence that states its point. Vary the length, though: identical short sentences read as choppy and drop words like "because" that show how ideas connect.
Rule: if a sentence makes the reader wait for its point, or cancel a negative to find it, split it or turn it positive.
Verbs that say who does what
~2 min
Active voice names the actor, and in instructions the actor is usually the reader. Free verbs hidden in nouns, prefer the present tense, write instructions as direct commands, and address the reader as "you" in a conversational but serious voice.
"The following information must be included" leaves open who includes it; "You must include the following information" settles it. In instructions the actor is the person reading, which is the case for active voice. To spot a passive, look for a form of "be" plus a past participle, then ask who acted. A common mistake is confusing passive with past tense. Passive is right when no one acts but the law, or when the actor is unknown.
Hidden verbs are the next habit to break. "We conduct an analysis of the data" buries "analyze" in a noun propped up by a weak verb. Look for words ending in -tion, -ment, or -ance, especially between "the" and "of." Freeing the verb often removes a wordy phrase too.
The present tense states what is true now; conditional and future forms add distance the reader has to unwind. Instructions belong in the imperative: "Sign the form," not "Applicants should ensure that the form is signed."
Use "you" for the reader, "we" for the agency, and contractions where they sound natural. Conversational is not casual; it holds up in a denial letter. And never soften a requirement into a suggestion.
Rule: make the reader the subject of every instruction, and put the action in a verb, not a noun.
Words the reader already uses
~2 min
Whether a word is jargon depends on the audience: a term that is efficient among specialists is noise to everyone else. Prefer everyday words, define necessary terms where they first matter, keep abbreviations few, and use one term for one thing throughout.
No word is jargon on its own. A term becomes jargon when this audience does not use it the way you do. "Riverine avifauna" may suit a field report, but on a public notice "river birds" says the same thing. The trap is that writers cannot hear their own jargon, because the terms feel ordinary to them.
Some terms must stay, such as a defined legal status or the name of a part. Keep them, and explain them in plain words where they first matter, in the sentence rather than a glossary nobody opens. Readers can call a letter clear while holding different ideas of a key term, which is how one agency learned that "service-connected disability" needed explaining.
Abbreviations save the writer a few letters once and cost the reader a lookup every time. Federal guidance suggests two or three per document at most; for the rest, a short nickname such as "the committee" carries meaning that letters cannot.
Consistency is the other half. Varying words to avoid repetition is an essay habit, and in instructions a new word reads as a new thing. If the steps say "application" and the payment page says "request," readers wonder whether they owe two documents. Choose the word your readers search with and keep it.
Rule: use the word your reader would say, define it where it first matters, and never swap it for a synonym.
Must, must not, may, should
~2 min
Requirement words tell readers what they must do. Use "must" for an obligation, "must not" for a prohibition, "may" for a choice, and "should" for a recommendation. Avoid "shall," which can state either a duty or a prediction.
"The agency shall review your file" could be a promise about what will happen or a duty placed on the agency. "Shall" carries both readings, which is why federal guidance recommends dropping it, and the federal courts replaced it with "must" in their rules of procedure. "Must" has one meaning.
The other words each need one job too. "Must not" bans something. "May not" is ambiguous: "You may not enter after 6 p.m." can mean "you are not allowed to" or "you might not." "May" grants a choice, so it has no place in what is really a requirement; "Applicants may be required to submit receipts" hides whether they must. "Should" marks a recommendation that readers treat as optional, so a real requirement written with "should" gets skipped.
Put each deadline or trigger beside the action it governs, and give a calendar date rather than a period readers must count. Keep cross-references few: each "see section 4(b)" asks the reader to leave, look something up, and come back while holding their place.
The misconception is that formal wording binds more firmly. Plain words carry the same force; what weakens a requirement is a careless edit that drops a condition.
Rule: for each requirement, check that one word tells the reader whether it is required, forbidden, allowed, or advised.
Lists and tables that scan
~2 min
Turn a sentence carrying several items or steps into a list, number it when order matters, and make every item complete the lead-in the same way. When readers need only the case that fits them, a table beats prose.
A sentence naming five things an application must include hides the count and runs the items together. A vertical list shows each item and how many there are. Number it when order matters or when you will refer to "step 3"; use plain bullets otherwise.
The common mistake is a list whose items do not fit its lead-in. Read the lead-in followed by each item: "To renew, bring your old license… a recent photo… pay the fee." The last is an action, not something to bring, so it belongs in its own sentence. Say in the lead-in whether readers must meet all of the items or just one, because a list cannot say so on its own, and write "X, Y, or both" rather than "and/or."
Tables solve a different problem. Benefit amounts that vary by disability rating and by number of dependents would take a paragraph of conditions to explain; in a table, each reader finds one cell. Headers must name the variable and its unit, cells should hold a value or a short phrase, and header cells must be coded as headers so screen readers can announce them. When each answer decides which question comes next, a short series of questions beats a mostly empty grid.
Rule: use a list when readers will act on items one at a time, and a table when they need to find the one case that is theirs.
A page built for scanning
~2 min
Many readers scan before they read, so design decides what they find. Use white space and real headings, emphasize sparingly with bold or italics, choose visuals that explain something, and write links that say where they go.
Dense pages put readers off before they read a word. White space between sections and around lists shows where one unit ends and the next begins, and a consistent layout lets returning readers go straight to the deadline or amount they need.
Emphasis works only by contrast, and the usual mistake is using too much of it. Bold or italics on the one detail that matters draws the eye; a page where every third phrase is bold has no emphasis left. Avoid underlining, which online readers expect to be a link, and all capitals, which are harder to read.
Visuals earn their space when they explain. A diagram of how ground water picks up contaminants from several sources does work that words would labor over; a stock photo does none and pushes content down. A trend reads better as a chart with a caption that states the finding. Keep words out of images, where they cannot be searched, resized, or read aloud.
Write the destination into each link, as in "renew your passport online," never "click here," because many screen-reader users browse a page's links out of context. When a link opens a file, give its format and size. Put anything readers need to finish the task in the text, not behind a link.
Rule: if someone skimming only the headings, the bold text, and the links would miss the point, the design is hiding it.
Test with real readers, not your own eyes
~2 min
Writers cannot see their own gaps, so clarity has to be tested. Paraphrase testing shows what readers think a passage means, usability testing shows whether they can find and use it, and comparative studies measure the effect at scale.
The usual mistake is trusting your own read. You see your draft with the missing context already in your head, so "Is this clear?" always gets a yes, and colleagues share much of that context too. Test with people who actually do the task, start early on drafts, and plan at least two rounds of testing, revising, and testing again.
Each method answers a different question. In paraphrase testing, a participant explains each section in their own words while you note what they say without correcting them; six to nine one-on-one sessions usually show where meaning breaks. Readers of one federal letter understood every word yet said they would phone the agency, because "shortly" gave them no date for their check. In usability testing, one participant at a time uses the real page for a realistic task while you watch. Record whether the task was completed, because people often call a task easy right after failing it. A comparative study sends the old and new versions to similar groups and counts outcomes such as calls or form errors. It shows whether a change worked, not why, so run it after the smaller tests.
Readability formulas are no substitute; they cannot see a buried deadline or an undefined term. Before anyone else reads a draft, edit in separate passes, structure first and words last, and read it aloud.
Rule: measure what readers do with the text, not whether they say they like it.
Structure that assistive technology can use
~2 min
Federal digital content must meet the Revised Section 508 Standards, and much of what they ask for is good structure: real headings and lists, a sensible reading order, meaning that does not depend on color, captions, and labeled form fields.
A screen reader navigates by the structure in the code. It can jump between headings, step through list items, and announce table headers, but only when those elements are coded as such. The common mistake is styling text to look like a heading: large bold text reads as a heading to a sighted reader and as plain text to assistive technology. Real structure also helps sighted readers scan, which is why accessibility and clarity usually ask for the same things.
Screen readers follow the order of the underlying page, not its visual layout, so content positioned with layout tricks can be read out of order. Never skip a heading level to get a smaller size; change the size with styling so the outline stays intact.
Meaning must survive without color. A form that marks errors only with a red outline tells nothing to someone who cannot see red, so describe each error in words, including where it is. Text also needs enough contrast against its background to be read at all.
Forms collect the most failures. Tie every field to a visible label, put the instructions people need before they begin at the top, and explain errors in text. Video needs captions, and audio-only content needs a transcript, which also lets anyone search or skim it.
Rule: if a page's meaning depends on how it looks, rebuild it so the structure, not the styling, carries the meaning.
When a misread costs the most
~2 min
Health and safety messages reach readers who may be stressed, sick, or rushed, and stress narrows what anyone takes in. Lead with one main message and a specific action, give numbers in forms people grasp, and check understanding by asking for the plan back.
Federal health guidance estimates that nearly 9 in 10 U.S. adults have limited health literacy skills, and even skilled readers misread health information when they are stressed or sick. Under stress, people miss nuances, remember less, and tend to believe the first message they hear. Write for a reader having a bad day.
Start with one main message at the top and a call to action: what to do, how to do it, and why it matters. "Wash your hands with soap and water before you eat" is an action; "Practice good hand hygiene" is a slogan.
Numbers need the most care. Most people grasp "1 in 4" more easily than "25%." Keep the denominator the same throughout so readers can compare without arithmetic, and say whether a number is high or low for this kind of risk. The classic trap is relative risk on its own: a risk that "falls by half" may drop from 4 in 10,000 to 2 in 10,000, which changes little for any one person. State the starting and new risk in the same terms.
In conversation, check understanding with teach-back: ask the person to explain the plan in their own words, and re-explain whatever comes back wrong. It tests the explanation, not the person. "Do you understand?" does not count, because people tend to say yes either way.
Rule: in a high-stakes message, put the action first, express risk as absolute frequencies, and confirm understanding by asking for the plan back.
Written by Keentune. We are not affiliated with or endorsed by the organizations whose documentation informs this guide, and any linked sources belong to their respective owners.
All exam, test, and product names and trademarks are the property of their respective owners and are used here for identification and reference only. Keentune is independent study practice — not affiliated with, authorized, or endorsed by any of these organizations.