Sentence-level writing principles for B2B contexts – homepage copy, outreach, LinkedIn, email, sales decks, ads. Channel-agnostic craft that applies everywhere words need to persuade a business buyer. Draws from the Fletch PMM copy framework and established principles of clear, precise writing.
B2C copywriting advice dominates LinkedIn. Most of it falls apart in B2B.
The difference: B2C aims to make the audience visualize the product. A shoe, a drink, a jacket – physical things that benefit from storytelling and sensory language. B2B SaaS must make the audience understand the product. Software is invisible and abstract. The task is comprehension, not imagination.
B2B copy requires two layers of work:
-
Strategy: What IS the product? Who is it for? What does it replace? What's the sharpest differentiation? These are positioning decisions, not writing decisions. They must be answered before anyone opens a doc. (See positioning-and-strategy.md for the full framework.)
-
Craft: Which synonym is clearest? In what order should the ideas go? Should the prepositional phrase sit at the beginning or end of the sentence? How many words can be cut without losing meaning?
Most B2B copy fails at layer 1. The team skips the hard positioning questions, then asks a writer to "make it sound good." No amount of craft can fix unclear strategy. But when the strategy is clear, craft is what makes the difference between copy that gets skimmed and copy that gets read.
The subject does the action. Clearer, more direct, and shorter than passive alternatives.
| Passive | Active |
|---|---|
| Production systems are built by our team | We build production systems |
| Your AI is made production-ready | We make AI production-ready |
| Compliance reports are generated automatically | The system generates compliance reports automatically |
Diagnostic: Search for "is/are/was/were + past participle" (is built, are generated, was created). Each instance is a candidate for active voice conversion. Not every passive sentence needs fixing – sometimes the object matters more than the actor – but most do.
Place subject-verb-object before modifiers. Readers grasp meaning when the main clause comes first, then integrate qualifications.
Weak: "By leveraging our proprietary AI models trained on millions of data points, teams can reduce processing time by 80%."
Strong: "Teams reduce processing time by 80% using AI models trained on millions of data points."
The core claim (teams reduce processing time) lands immediately. The how (AI models) follows as supporting detail.
Long sentences carry complexity and detail. Short sentences create impact. Strings of similar-length sentences become monotonous – and this is where AI-generated copy fails most visibly. AI defaults to medium-length sentences with predictable rhythm.
Read the paragraph aloud. If every sentence lands at roughly the same beat, rewrite. Break a long sentence into two. Combine two short ones. The ear catches monotony before the eye does.
Elements in a list or comparison must match grammatical form. Mismatches create a rhythm break readers hear but can't identify.
Match forms:
Bad: "We focus on planning, building, and deployment." Good: "We focus on plans, builds, and deployments." OR "We plan, build, and deploy."
Bad: "Built for reliability, handling complexity, and scale." Good: "Built for reliability, complexity, and scale." OR "Built to work reliably, handle complexity, and scale."
Match grammatical weight: If one item has a modifier, all should – or none should.
Bad: "We deliver strategic consulting, development, and ongoing support." Good: "We deliver strategy, development, and support." OR "We deliver strategic consulting, rapid development, and ongoing support."
Gerunds (-ing words acting as nouns) sound like verbs even though they're grammatically nouns. This creates mismatch when paired with concrete nouns in parallel structures. The sentence passes grammar checkers but sounds wrong to the ear.
The fix in three steps:
Clumsy: "Our recommendations come from data, not from guessing or boilerplate wisdom."
- "guessing" sounds like a verb; "wisdom" is a noun with an adjective. Technically parallel. Audibly mismatched.
Better: "Our recommendations come from data, not random guesswork or boilerplate wisdom."
- Both items now noun+adjective. Balanced weight.
Best: "We recommend based on data, not hunches or platitudes."
- Active voice. Precise nouns. Fewest words for exact meaning.
Three principles applied: swap gerund for concrete noun, match grammatical weight, switch to active voice.
Diagnostic: Scan lists and comparisons for -ing words sitting next to concrete nouns. If the -ing word can be replaced with a standard noun, replace it.
Every unnecessary word dilutes impact and wastes the reader's attention. This is especially critical in outreach where the reader has given zero commitment to keep reading.
| Bloated | Tight |
|---|---|
| We specialize in the building of sophisticated systems | We build sophisticated systems |
| Provides implementation of automated workflows | Implements automated workflows |
| Has the ability to process | Processes |
| In order to achieve | To achieve |
| At this point in time | Now |
| Due to the fact that | Because |
Diagnostic: Read each sentence and ask: can any word be removed without changing the meaning? If yes, remove it. Repeat until the answer is no.
Abstract language forces readers to translate concepts into mental images. Concrete language does the translation for them.
Abstract: "We help companies improve their operational efficiency through digital transformation." Concrete: "We cut invoice processing from 3 days to 20 minutes."
Abstract: "Our platform provides comprehensive solutions for your business needs." Concrete: "Our platform handles payroll, benefits administration, and tax filing."
In outreach especially, concrete language proves understanding. Abstract language signals that the sender could be writing to anyone.
Specificity proves expertise. Generic language could come from anyone.
Generic: "We use various methodologies to deliver results." Specific: "We use test-driven development and continuous integration to ship without breaking production."
Generic: "Our team has extensive experience in AI." Specific: "We've shipped 20+ AI systems for B2B SaaS companies across legal tech, marketing automation, and e-commerce."
The test: Could a competitor write the exact same sentence about themselves? If yes, it's too generic. Rewrite with details only the sender could claim.
Converting verbs to nouns buries the action and adds words. B2B writing is full of this – "implementation" instead of "implement," "utilization" instead of "use." The noun form sounds more corporate, which is why people reach for it. It's also why it's deadening.
| Nominalized | Direct |
|---|---|
| We provide implementation of AI systems | We implement AI systems |
| Our team performs analysis of customer data | Our team analyzes customer data |
| This enables the facilitation of faster onboarding | This speeds up onboarding |
| We offer consultation on architecture decisions | We consult on architecture decisions |
Diagnostic: Search for words ending in -tion, -ment, -ance, -ence that have a verb form. Replace the noun with the verb and restructure the sentence.
"Generally," "typically," "somewhat," "fairly," "arguably," "it seems that" – these weaken claims without meaningful qualification. Remove unless they communicate something specific.
Hedged: "Our approach generally tends to produce fairly significant results for most clients." Direct: "Our approach produces significant results."
If the claim needs qualification, qualify with data, not hedge words. "Our approach produced a 40% reduction in processing time across 12 client projects" qualifies with precision. "Our approach generally works" qualifies with vagueness.
"It's important to note," "crucially," "significantly," "it should be noted that" – if something matters, the content should show it. Emphasis markers tell rather than demonstrate. They also add words that don't carry meaning.
Bad: "It's important to note that our system handles edge cases automatically." Good: "Our system handles edge cases automatically."
Placement and context show importance. Introductory labels don't.
AI-generated copy has identifiable patterns. As AI usage increases, these patterns become the background noise of B2B communication. Copy that uses them blends in. Copy that avoids them stands out.
1. "All the X. None of the Y."
Example: "All the features. None of the friction." Problem: Extremely choppy. Trying too hard. Fix: Convert the fragment to a prepositional phrase. Rewrite: "Get all the features with none of the friction."
2. "It's not X. It's Y."
Example: "Good leaders ask for help. It's not weak, it's wise." Problem: Staccato correction rhetoric. Cutesy. Fix: Change the first sentence to a dependent clause. Rewrite: "While weak leaders won't ask for help, the best ones know it's wise."
3. "No X. No Y. No Z."
Example: "It's next-gen automation. No waste. No guessing. No restarts." Problem: Approaching children's book territory. Fix: Start with two negatives, convert the third to a positive. Rewrite: "Avoid waste and guesswork with automation that gets it right the first time."
4. "Good thing. Good thing. Good thing."
Example: "Our strategies? Simple. Effective. Easy." Problem: Dated and salesy. Fix: Skip one item, modify a noun, convert the last to a verb. Rewrite: "Trust us to deliver simple strategies that work."
AI accelerates copy trends. Sentence patterns that were fresh two years ago now read as generic because they've been mass-produced. Short, punchy, fragmented sentences – once a sign of sharp writing – now signal AI because everyone's AI writes that way.
The better question for any sentence: "does this sound fresh?" That matters more than whether it "sounds like AI." A human can write a stale sentence. An AI can produce a fresh one. Freshness is the real filter.
The fix isn't avoiding AI tools. It's developing taste – knowing what sounds like everything else versus what sounds like a specific voice. This requires:
- Reading widely outside of B2B marketing content
- Paying attention to which sentence structures have been overused
- Being willing to write sentences that don't follow the current template
- Preferring fluid, connected prose over choppy fragments
Readers are bored of staccato. They're craving something more fluid.
- Starting paragraphs with "In today's..." or "In the world of..."
- "Landscape" as a noun for any market or industry
- "Leverage" as a verb for any action
- "Robust" and "seamless" as adjectives for any product
- "Navigate" for any challenge
- "Empower" for any benefit
- Excessive use of "Furthermore," "Moreover," "Additionally" as transitions
- Lists of three adjectives as standalone sentences
- Every paragraph ending with a neat conclusion
Diagnostic: Read the copy and ask: could this have been the first draft from ChatGPT? If the answer is yes, rewrite. Not because AI is bad, but because first-draft AI output sounds like everyone else's first-draft AI output.
The principles above apply everywhere. Outreach adds constraints: the reader hasn't opted in, attention is measured in seconds, and the tolerance for wasted words is near zero.
In a cold email or LinkedIn message, the first sentence determines whether the second gets read. Front-load relevance. Show the reader that this message is specifically about them and their situation – not a template sent to 500 people.
Weak first sentence: "I hope this email finds you well. I wanted to reach out because..." Strong first sentence: "Saw that [Company] just launched [specific thing] – the [specific detail] caught my attention."
The weak version could be written by anyone to anyone. The strong version proves the sender did actual research.
Outreach that tries to cover three value propositions in one email converts worse than outreach that makes one sharp point. The reader's capacity for processing a cold message is limited. Pick the single most relevant point for this specific recipient.
Claims without evidence are noise. "We're the best at X" means nothing. Proof means something.
Claim: "We build world-class AI systems." Proof: "We built the AI system that handles 200k customer interactions per month for [Client]."
Claim: "Our team is highly experienced." Proof: "We've shipped 20+ AI systems for B2B SaaS – across legal tech, marketing automation, and e-commerce."
In outreach, one specific proof point outweighs ten general claims.
The same prospect needs different messages depending on how aware they are. (Full framework in messaging-and-copy.md, Section 3.)
- Unaware: Lead with their current situation. Don't pitch.
- Problem aware: Lead with the problem. Show understanding.
- Solution aware: Lead with capability. Show fit.
- Product aware: Lead with differentiation. Show why this over alternatives.
Cold outreach almost always targets unaware or problem-aware prospects. Leading with features or product claims to someone who doesn't know they have a problem guarantees irrelevance.
Pass 1 – Clarity: Does each sentence say exactly what it means? Is anything ambiguous? Would someone unfamiliar with the product understand this? Cut every sentence that doesn't answer "so what?"
Pass 2 – Tightness: Can any word be removed without changing the meaning? Are there hedge words, emphasis markers, or filler phrases? Is anything said twice in different words? First drafts typically run 30-50% too long.
Pass 3 – Sound: Read it aloud. The mouth catches what the eyes miss. Awkward phrasing, monotonous rhythm, and mismatched parallel structures all reveal themselves when spoken. If a sentence makes the reader stumble, rewrite it.
- Search for passive voice (is/are/was/were + past participle)
- Search for gerunds in parallel structures
- Search for "there is/are" – replace with the actual subject
- Search for emphasis markers and hedge words – cut most
- Check that every list has consistent grammatical form
- Verify the first sentence of every paragraph states the point
- Confirm the first sentence of the entire piece earns the second sentence