The operator's prompt guide.

Here is context on my business. Use it for everything in this conversation, and tell me when something I ask contradicts it.

BUSINESS: [what you sell, one plain sentence, no adjectives]
CUSTOMER: [who pays, their job title, what they're trying to fix]
STAGE: [revenue range, team size, how long you've been running]
WHY WE WIN: [the one or two real reasons people choose us]
WHY WE LOSE: [the objection or competitor that kills deals]
VOICE: [three adjectives, plus one thing we never say]
LIMITS: [budget, time, regulated language, anything off the table]
THIS MONTH: [what I'm working on right now]

Play it back to me in three sentences so I know you have it. Then ask me the one question that would most change your answers. Don't start working yet.

I'm going to paste the text of my homepage. Read it once, the way a stranger would — fast, skimming, slightly annoyed.

Then answer as that stranger, not as an assistant:
1. What do you think this company sells? Be literal.
2. Who is it for? Quote the words that told you.
3. What's the one thing that would make you leave?
4. What question do you still have that the page didn't answer?

Then, and only then, rewrite the headline and subhead three ways: one plain, one built around the outcome, one that names the customer directly.

Here's the page: [paste]

I'm pasting raw customer language — reviews, support tickets, sales-call notes, survey answers. It's messy. Don't clean it up.

Pull out:
1. The exact phrases customers use for the problem — their words, not marketing words
2. The words they use for the outcome they want
3. Objections and hesitations, ranked by how often they show up
4. Any phrase that surprised you

Give me a table: the phrase, roughly how often something like it appeared, and where I'd use it — headline, email subject line, objection handling, or FAQ.

Do not invent phrases. If something appeared once, mark it once.

Here it is: [paste]

Here's how I describe my business: [paste your current description]

Play a skeptical version of my ideal customer — [role, company size, industry]. They're busy, they've been burned by a vendor before, and they don't care about my process or my technology.

Push back on my description four times. Each round, get more specific and harder to answer. Don't let me off the hook with a vague answer, and don't be polite about it.

After the fourth round, tell me which sentence survived and which one I should cut.

Here's one thing I believe about my industry that most people get wrong: [your take, two or three sentences]. Here's my evidence: [proof, numbers, a story from your own work].

Turn it into a week of content:
- One long post, 600–800 words, that makes the argument properly
- Three short posts, each taking one angle — a story, a data point, a contrarian line
- One that's only a question to my audience
- One that shows the work: a screenshot, a process, a before and after

Rules: no hashtag walls. No "In today's fast-paced world." No manufactured vulnerability. Every piece has to stand on its own. Vary the openings — if two of them start the same way, rewrite one.

Step 1 — don't write anything yet.

I'm reaching out to [name, title, company]. Here's what I know: [paste their site copy, a recent post, a press release, their job listings].

Tell me three things about their situation that are specific enough that I couldn't say them to anyone else. Rank them by how likely each one is to be a live problem right now. If the material doesn't support a real observation, say so instead of guessing.

Step 2 — use observation number [n].

Write a 90-word email. Structure: one line that proves I did the work, one line naming the problem in their words, one line on what I'd do about it, one ask — 15 minutes, with two specific times.

No compliment opener. No "I hope this finds you well." No paragraph about my company. If this email could be sent to a hundred people, it's wrong.

A customer just bought [product or service] at [price]. The moment they'll know it was worth it is when [the specific first win].

Design the first 30 days: every message, who sends it, when it goes out, and what it's for. For each one, tell me what the customer should do next and how I'll know whether they did it.

Then tell me the single point in those 30 days where people are most likely to go quiet — and what should sit there.

Assume I have no automation and no team. Give me something I could run by hand this week.

Here are the reasons people gave when they left: [paste them raw]. Here's what I know about those customers: [tenure, plan, usage, anything you have].

Three things:
1. Group the reasons by what really happened, not what they said. People are polite when they leave.
2. Tell me which group I could have saved, and at what point I would have had to act.
3. Tell me which group I should let go — and why chasing them would cost more than it's worth.

Then: what's the earliest signal, in data I can already see, that someone is heading toward each group?

I have to tell [who] that [the bad news — a price increase, a delay, a mistake, ending the relationship].

Write it. Rules:
- Lead with the news in the first sentence. No throat-clearing.
- Say what happened without hiding in passive voice — "we missed the date," not "the date was missed."
- Say what I'm doing about it, with a date.
- One apology at most. Not three.
- Don't ask them how they feel about it.

Then give me the two sentences most likely to make this worse, so I know what to watch for in my own edits.

I'm deciding whether to [decision]. Context: [the situation, the money, the timing, who's affected].

Write a one-page memo:
- The decision in one sentence, with the date it has to be made by
- Is this reversible or not? If it's reversible, say so plainly — I'm probably overthinking it.
- The case for, in its strongest form
- The case against, in its strongest form — not a straw man
- What I'd have to believe for each side to be right
- What information would change the answer, and whether it's worth the delay to go get it

No recommendation yet. Just make both sides as strong as they get.

Here's my plan: [paste it].

You're not my assistant right now. You're the person who has to clean up after this fails.

Write the post-mortem from six months in the future, assuming it went badly. Be specific about what broke and in what order. Then rank the failure points by how likely they are and how much damage each one does.

For the top three, tell me the cheapest thing I could do this week to reduce the risk.

Don't soften it and don't end on an encouraging note.

I charge [price] for [offer]. My costs are [amount]. My customers are [type]. What they compare me to is [competitor, doing it themselves, or doing nothing].

Work through:
1. What am I really being paid for — time, outcome, risk, access, or speed?
2. What is that outcome worth to them in their own numbers? Show the math and mark every assumption.
3. Three ways to restructure this: one lower with a smaller scope, one higher with something added, one priced on the outcome instead of on hours.
4. What would have to be true for me to double the price.
5. Who I'd lose at each level — and whether losing them is a loss.

If the honest answer is that I'm priced correctly, say that.

Here's my week: [dump all of it — calendar, what got done, what didn't, what's nagging at me].

Sort it:
1. What moved the business forward. Honestly — not what felt busy.
2. What I did that someone else should be doing, or that shouldn't be done at all
3. What I avoided, and the pattern in what I avoid
4. What I'll still be dealing with in three months if I don't handle it now

Then give me three things for next week. No more than three, each with its first concrete step. If I'm carrying something over for a third week in a row, call it out.