What custom instructions do
Custom instructions are a permanent set of preferences ChatGPT applies to every conversation, so you never have to re-explain who you are or how you want it to respond. There are two fields: one for who you are (your role, context, and what you work on) and one for how you want ChatGPT to respond (tone, format, length, and rules). Fill them in once and every future chat starts already tuned to you — it is the single biggest quality upgrade most people are missing.
What to put in the first field (about you)
Use the first field to give ChatGPT durable context: your profession and seniority, your industry, who your audience usually is, the tools and frameworks you use, and your current goals. For example: a B2B SaaS marketer writing for technical founders; I value data-backed, concise output and dislike fluff. The more it knows about your world, the less generic and the more relevant every answer becomes — without you typing it again each time.
What to put in the second field (how to respond)
Use the second field to set your output preferences: tone (direct, professional, friendly), default length (concise unless I ask for depth), format (use headers and bullet points; show reasoning before conclusions), and hard rules (no clichés; ask a clarifying question if my request is ambiguous; flag assumptions). These instructions shape the style of every response, turning ChatGPT from a generic assistant into one that consistently writes the way you want.
Ready-to-use examples by role
A developer might write: respond like a senior engineer; show working code first, then a short explanation; call out edge cases and trade-offs. A founder: be a blunt strategic advisor; give me the strongest counter-argument to my plan; no motivational filler. A writer: match a warm, conversational voice; avoid corporate jargon and em-dash-heavy sentences; vary sentence length. Adapt these to your work — the pattern is always role plus tone plus format plus rules.
Mistakes to avoid
Three common mistakes make custom instructions backfire. First, being vague (be helpful does nothing — be specific). Second, overloading them with contradictory rules, which the model cannot satisfy at once; keep them tight and prioritized. Third, never updating them — revisit your instructions every few months as your work and the model change. Done well, custom instructions are a five-minute setup that improves thousands of future answers.
Frequently Asked Questions
What should I put in ChatGPT custom instructions?
In the first field, put durable context about you: profession, seniority, industry, audience, tools, and goals. In the second, put response preferences: tone, default length, format, and hard rules like be concise or flag assumptions. Together they tune every future chat to you.
Do custom instructions really improve answers?
Yes, significantly. By giving ChatGPT permanent context and output preferences, every response starts tuned to your role and style instead of generic defaults. It is one of the highest-leverage five-minute setups for anyone who uses ChatGPT regularly.
Can I have different custom instructions for different tasks?
The instructions are global, but you can keep them broad and override per chat when needed, or use Projects and custom GPTs for task-specific setups. Many power users keep general instructions plus a small library of task-specific prompts.
How often should I update my custom instructions?
Every few months, or whenever your role, goals, or the model change. Treat them as living settings — outdated instructions quietly degrade your output. A quick quarterly review keeps them sharp.