You typed "write a marketing strategy for my business" into ChatGPT and got back a SWOT analysis, four bullet points about "leveraging digital channels for growth," and a competitive advantage section that could describe literally any company in your industry. You didn't learn anything from it. You couldn't hand it to anyone and have them know what to actually do on Monday morning.
That's not the tool failing you. It's the default output of a prompt with no real inputs in it.
The short version
The best ChatGPT prompt for marketing or brand strategy doesn't ask for "a strategy." It hands the model who you're actually losing deals to right now (named, not "the market"), the one real reason a customer picks you over the obvious alternative, and the actual constraint you're working inside — budget, team size, or timeline. Strategy frameworks — SWOT, the four Ps, Porter's Five Forces — aren't the missing ingredient; specificity is. Ask for "a go-to-market strategy" and you get a business-school outline. Tell it who you lose to and why, and you get a plan with a real next move in it.
Why "write me a strategy" produces a deck nobody uses
A strategy document is supposed to make a decision easier. Most AI-generated ones do the opposite — they cover every base (four Ps, three personas, five channels) so thoroughly that nothing is actually chosen. That's not the model being lazy. It's mirroring the shape of every strategy doc it's been trained on, which is written to look comprehensive to a boss, not to be used by a team.
Real strategy is mostly about what you're not doing. "We're not fighting on price, we're fighting on speed of delivery" is a strategy. "We will leverage a multi-channel approach across social, email, and content" is not — it's a description of having a marketing department. The prompt is where you force the first kind instead of the second.
Brand strategy and marketing strategy aren't the same prompt
People ask ChatGPT for "a marketing strategy" when what's actually broken is the brand strategy underneath it — and the two need different prompts. Brand strategy is the identity layer: who you are, what you stand for, why someone should believe you over a competitor selling the same thing. Marketing strategy is the execution layer built on top of it: which channels, what message, what cadence. Ask for marketing tactics before the positioning is nailed down and you get a nice-looking content calendar for a brand nobody has a reason to choose. Do the positioning prompt first, always — everything below inherits its quality.
The prompts
Each one is annotated: why it's built that way, and what to swap in for your business. All work in ChatGPT, Claude, Gemini and Grok.
1. The positioning prompt (do this first)
Act as a brand strategist. My business: [what you do]. My direct alternatives — not "the whole market," the 2-3 things a prospect actually compares me to: [name them]. Customers who chose me over those, what did they say was the reason? [paste real feedback/reviews, or your honest best guess]. Give me a one-sentence positioning statement: "For [who], [brand] is the [category] that [the one thing], unlike [main alternative], because [proof]." Push back if my answer is generic.
Why it's built this way: naming real alternatives instead of "the market" forces genuine differentiation instead of a list of adjectives. The "push back if generic" line matters — without it, the model happily accepts "we care about quality" as a differentiator. Swap in: actual competitor names and real customer language, not your guess at what customers think.
2. The customer-language prompt
Here's what I think my customer's problem is: [your assumption]. Here's [reviews / support tickets / DMs / call notes — paste real customer words]. Pull out the exact phrases they use to describe the problem and the outcome they want. Don't paraphrase into marketing language — give me their words. Then tell me where my assumption above was wrong.
Why it's built this way: copy written in your internal language ("streamline your workflow") loses to copy written in the customer's language ("stop re-typing the same email every Monday"). This makes the model mine your own data instead of inventing a persona. Swap in: any raw, unedited customer text you have — five real reviews beat zero personas.
3. The competitive-gap prompt
Here are 3 competitors and how each positions itself: [name + their tagline/homepage headline for each]. What's the gap — the claim none of them are making, or the customer none of them are clearly speaking to? Don't tell me to be "better," tell me what to say that they aren't saying.
Why it's built this way: "be better" isn't a strategy; a gap in the conversation is. This forces the model to compare actual claims instead of guessing at market size. Swap in: real homepage headlines — screenshot them if you have to, don't summarize from memory.
4. The messaging pillars prompt
Based on this positioning: [paste your positioning statement from prompt 1]. Give me 3 messaging pillars — not taglines, the 3 underlying arguments for why someone should believe the positioning. For each: the argument, one proof point that backs it up, and how it could show up in an ad, an email subject line, and a landing page headline.
Why it's built this way: pillars are the argument underneath the copy, so every channel repeats a version of the same three things instead of improvising a new message each time. Swap in: nothing structural — just make sure step 1's positioning is solid, since this prompt inherits it entirely.
5. The go-to-market prompt
I have [budget], [team size — including "just me"], and [X months] before I need results. My audience is on [where they actually spend time — be specific, not "social media"]. Recommend ONE channel to focus on for the next 90 days, not three. Justify the choice against my specific constraints, then give me the first two weeks of what to actually publish or send.
Why it's built this way: asking for "a channel strategy" gets you five channels and none of them done well. Forcing one choice, justified against your real constraints, gets you something you can actually execute. Swap in: your real budget and team size — "just me, two hours a week" is a valid and important input, not a confession.
6. The brand-voice extraction prompt
Here are 3 pieces I've written, or that represent how I want to sound: [paste samples]. Describe the voice as 5 mechanical rules — sentence length, what we never say, how we open, level of formality, humor or none. Then write [a sample email/ad/post] following those rules exactly.
Why it's built this way: "sound professional but approachable" means nothing to a model; five concrete rules extracted from real samples do. Same trick that fixes generic AI writing, applied to brand voice specifically. Swap in: your best existing writing, or writing from a brand you admire if you're starting from zero.
Which prompt for which problem
| Your situation | Use prompt | What it fixes |
|---|---|---|
| Don't know why customers actually pick you | #1 Positioning | Generic differentiation ("quality, service") |
| Marketing copy sounds like internal jargon | #2 Customer language | Copy describing you, not the customer's problem |
| Can't articulate what makes you different from the obvious competitor | #3 Competitive gap | "Be better" instead of a real claim |
| Messaging changes every time you write something new | #4 Messaging pillars | No consistent argument across channels |
| Spreading effort across too many channels | #5 Go-to-market | A plan for everything, executed on nothing |
| Brand voice is inconsistent across writers or tools | #6 Voice extraction | "Professional but friendly" (meaningless) |
Do this now
Run the positioning prompt first, even if you think you already know your positioning — most people know their product, not their positioning. Everything downstream, messaging, channel choice, voice, is only as good as that one sentence. If the model pushes back and calls your differentiator generic, that's the most useful thing it'll tell you all week.
Every prompt here is ready to copy on PromptThisOne — grab the full ChatGPT Marketing Strategy prompt collection, built for exactly this.
Related guides
- ChatGPT ad copy prompts that actually convert — once positioning is set, this is how it becomes copy that sells.
- ChatGPT business ideas prompts — for validating the offer before you build a strategy around it.
- How to write AI prompts that actually work — the constraint-based thinking behind every prompt on this page.