Why "act like an expert" fails

"Act like a marketing expert" asks Claude to adopt a persona. It doesn't hand over anything true about your business, your numbers, your customers, or your content. With nothing specific to work from, Claude fills the gap with the most generic version of marketing advice that fits the sentence. That's why the output always reads like it could apply to any business.

Each prompt below is built the opposite way. It only produces a useful answer once you've filled in something real. No persona, no roleplay, just your actual inputs.

How to use this page: Copy a prompt, replace everything in brackets with your real details, paste it into Claude. Skipping a bracket or leaving it vague is the one thing that breaks all five.

01

The Competitor Pain Point Prompt

Finds the exact pain point a competitor leans on that you've never mentioned, in their words, not a paraphrase.

Prompt
Here's what I sell: [one sentence describing your offer]
Here's who buys it: [one sentence describing your buyer]
Here's a competitor or niche leader doing well in this space: [name]

Pull up their last 10 pieces of content you have context on (or ask me
to paste them if you don't). Tell me the single pain point they return
to the most, in their exact words or close to it. Then tell me where
in my own last 5 posts I could have said the same thing and didn't.

Don't summarise their content strategy in general terms. I want the
one pain point, quoted, and the one place I'm silent on it.

Why it works

It asks for one quoted pain point, not a strategy summary, so Claude can't default to generic positioning advice.

Fails without

A real competitor name and your real offer. Without them, Claude invents a plausible-sounding but made-up pain point.

02

The Dead Content Audit Prompt

Finds the exact sentence where a reader would stop reading, not a vague note to "add more value."

Prompt
Here are my last 5 pieces of content, pasted in full:
[paste post 1]
[paste post 2]
[paste post 3]
[paste post 4]
[paste post 5]

For each one, give me the exact sentence where a stranger would stop
reading and scroll away, and the specific reason: too vague, too slow
to get to the point, no reason to care, or something else you name
directly.

Do not tell me to "add more value," "be more engaging," or "hook
harder." Quote the exact line and name the exact problem with it.

Why it works

It explicitly bans the three phrases Claude defaults to when it has nothing concrete to say, forcing a real diagnosis.

Fails without

The actual posts, pasted in full. A description of your content style produces advice, not a diagnosis.

03

The Objection-to-Offer Prompt

Turns your most common "no" into a proof-backed line for your sales page, DM script, or call close.

Prompt
Here's the objection I hear most often before someone buys:
[the actual objection, in the words people really use]

Here's a specific, real result I've gotten for a client or myself
that directly answers it:
[the result, with a real number or timeframe, not "great results"]

Write the one sentence that should go on my sales page or in my DM
script to answer this exact objection, using this exact result.
No reassurance language like "don't worry" or "trust the process."
The number does the convincing, not the tone.

Why it works

It requires a real number as proof, which rules out the warm, vague reassurance Claude writes by default.

Fails without

An objection in the customer's actual words and a result with a real number. "Good results" produces nothing usable.

04

The Revenue Leak Finder Prompt

Finds which single step in your funnel is losing the most money, and the one test to run on it this week.

Prompt
Here's my funnel, step by step:
[e.g. ad -> landing page -> booked call -> close]

Here are my real numbers at each step over the last [30/60/90] days:
[step 1: number], [step 2: number], [step 3: number], [step 4: number]

Tell me which single step is losing me the most money relative to a
reasonable benchmark for that step type, and the one specific test
I should run on that step this week. One test, not a list of five
ideas. Tell me what to measure to know if it worked.

Why it works

Real numbers at each step let Claude compare against benchmarks instead of guessing where the funnel is weak.

Fails without

Actual numbers per step. "My funnel isn't converting well" gives Claude nothing to compare against.

05

The Proof-First Content Prompt

Writes one piece of content built entirely around a real proof point, instead of general advice with no receipts.

Prompt
Here's a specific result I got, for a client or for myself:
[the result, with a real number: revenue, time saved, conversion
lift, whatever it is]

Write one piece of content for [platform] built entirely around this
one proof point. The hook has to reference the actual number in the
first line. Do not generalise it into advice anyone could give. If
the post could be true without this specific result, rewrite it.

Why it works

The self-check line, "if the post could be true without this result, rewrite it," catches generic drift before you post it.

Fails without

A real, specific number. "I helped a client grow" isn't a proof point, it's the same sentence every competitor uses.

Tip

No result on hand yet? Run the Content Intel System first to find a proof point already sitting in your own numbers, then come back and run this one.

Frequently Asked Questions

Because it asks for a persona instead of giving Claude something to work with. There's no hidden expert knowledge unlocked by the roleplay. The quality of the output comes from the specifics you hand it: your real numbers, your real content, your real objections. Without those, Claude fills the gap with the most generic version of marketing advice that fits the sentence.
Yes. The brackets are the whole point. A prompt with brackets filled in forces a specific answer. The same prompt with the brackets left vague, or skipped, collapses right back into generic output.
Yes. None of the 5 prompts depend on a Claude-specific feature. They work in any chat model. Claude tends to follow the formatting instructions in each prompt more consistently, which is why this site defaults to it.
Use what you do have. Even 5 posts, 3 objections you've heard out loud, or your best guess at the funnel step that's weakest is enough to start. The prompts get sharper as your inputs get sharper, but a rough real number still beats no number.
A generic template still works without your specifics, which is exactly the problem. Every prompt here is written so it cannot produce a useful answer unless you've filled in something true about your business. That constraint is deliberate.