Matching Tone: Formal, Friendly, Persuasive
The same set of facts can be written in a dozen different tones, and getting the tone wrong is one of the fastest ways to undermine an otherwise well-written email. A perfectly reasonable request can land as cold and bureaucratic, or as pushy and presumptuous, depending entirely on word choice and sentence rhythm, not on the substance of what is being asked. AI tools are unusually strong at tone control once you know how to specify it, because unlike structure or facts, tone is exactly the kind of pattern-based language skill large language models are built for. This lesson gives you a working vocabulary for describing tone precisely, so you stop getting the AI’s default guess and start getting the tone you actually need.
Why “make it more professional” is not enough
Vague tone instructions produce vague results. “Make it sound professional” or “make it friendlier” gives the AI very little to work with and it will apply its own generic interpretation, which is often stiffer and more corporate than most modern workplaces actually want. Specific tone words, paired with a short reference point, work far better. Instead of “professional,” try “warm but efficient, the way you’d write to a colleague you respect but don’t know well.” Instead of “friendly,” try “casual and upbeat, like a quick message to a teammate you get on well with, contractions are fine.” The more concrete the reference point, the more accurately the AI hits the mark.
A tone vocabulary you can reuse
| Tone | Good description to give the AI | Typical use case |
|---|---|---|
| Formal | Respectful, no contractions, no humour, precise language | First contact with a senior executive, legal or HR matters |
| Warm professional | Polite and personable but efficient, contractions fine | Regular client or colleague communication |
| Friendly and casual | Relaxed, conversational, short sentences, light humour allowed | Close colleagues, informal team culture |
| Persuasive | Confident, benefit-led, no hedging language, active voice | Sales emails, pitches, internal proposals |
| Firm but respectful | Direct, no apologising excessively, clear boundaries | Chasing overdue payments, pushing back on scope creep |
| Empathetic | Acknowledges the other person’s position before making a point | Delivering bad news, handling complaints |
Persuasive tone without sounding pushy
Persuasive writing is the tone people most often get wrong, either too soft to move anyone or aggressive enough to trigger resistance. The most reliable version of persuasive email tone leads with the reader’s benefit, not your own need, uses active voice and concrete numbers rather than vague superlatives, and avoids hedging words like “just,” “maybe,” or “I was wondering if perhaps.” A useful instruction to give ChatGPT or Claude: “Make this persuasive but not pushy, lead with what’s in it for them, cut any hedging language like just or maybe, use one concrete number if I’ve given you one.” This produces confident writing that still respects the reader’s autonomy to say no.
Matching tone to the relationship, not just the topic
The same request, asking a supplier to expedite an order, needs a different tone depending on whether you have worked with them for five years or five days. Always tell the AI about the relationship, not just the topic: “We’ve worked together for three years and have a good relationship, so a bit of warmth and informality is fine” versus “This is only our second interaction, keep it more careful and formal.” This one piece of context often matters more to getting the tone right than any list of tone adjectives, because it tells the AI how much social capital you have to draw on.
Adjusting an existing draft instead of starting over
Tone is one of the easiest things to adjust after the fact, which is a major advantage of AI drafting over writing alone. If a draft comes back too stiff, you do not need to start again, simply reply with an instruction such as “this feels too formal for how well I know this person, loosen it up, contractions are fine, keep the same facts.” Equally, if a draft feels too casual for a serious situation, “this needs to feel more serious, remove the exclamation marks and any jokes, keep it respectful.” Tools like Claude and ChatGPT handle these follow-up tone adjustments reliably within the same conversation, because the facts and structure are already locked in and only the surface language needs to shift.
Reading tone back before you send
Because AI-generated text can sometimes drift slightly from the intended tone even after adjustment, always read the final version aloud, or at least silently in the voice you would actually use, before sending anything important. Ask yourself whether it sounds like you, not like a corporate template. If it does not, a final prompt such as “make this sound less like a template and more like a real person wrote it, keep it natural” is often enough to close the gap, a technique explored more fully in the closing lesson on avoiding the AI voice.
Tone control matters even more once you start sending similar emails to many different people at once, since a single tone applied to a hundred recipients rarely fits all of them equally well. The next lesson covers how to personalise AI-written emails at scale, so that volume never comes at the cost of feeling individually written.
