I downloaded a content prompt pack a while ago. Someone I respected had shared it - one of those well-designed PDFs with a clean cover page and thirty prompts organized by content type. I ran a few and watched the content appear, and for a moment I had that really excited feeling you get when you think you've finally solved the problem. 

Then I read it back and the sinking feeling started creeping back in. It looked good at first glance but was completely hollow - sentences about topics I work with, in a reasonable order, expressing a point of view that could have belonged to anyone doing vaguely similar work. I spent the next hour trying to fix it before giving up. Honestly, writing from scratch and editing that would have taken way less time.

The problem wasn't the prompt. The prompt had nothing to work with.

There are Instagram carousels going around right now selling "secret codes" for Claude. Some examples are: 

Type /ghost before your prompt and your output will sound human - it will even pass AI detectors. 

Type /godmode and you unlock a more aggressive, comprehensive response style. 

Add L99 at the end and Claude answers at the highest expert level possible. 

The posts are well-designed and they encourage you to learn more from them - either downloading their set of secret codes or joining their training…you know the deal. The captions hint that Anthropic hasn't publicly announced these codes, which is doing a lot of work to make "we invented this" sound like insider access.

None of them do anything. In the regular chat interface, Claude either ignores the prefix or treats it as part of your question. You are not getting a different, better, more human-sounding response. You're getting the same response you would have gotten anyway, maybe with slightly confused framing at the start. 

The posts spread because people want to believe the gap between the output they're getting and the output they want is one magic word away from closing. It isn't. The gap is in what you're feeding the tool, not in the tool itself.

Front end marketing - lead magnets, webinars, summits, SEO - runs on knowing your audience well. You need to understand who you're trying to reach, what gets their attention and what problem they want solving. You can build a substantial audience with strong front end marketing without having done deep work on your own voice, because the front end isn't really asking for it yet. 

The inbox is where that changes. 

Back end nurturing runs on knowing yourself - your actual point of view, not the polished version you put on a lead magnet about-me section. What you genuinely think about the thing you do, where you disagree with the wisdom in your field, what you've changed your mind about, what you keep seeing that nobody else seems to be talking about. That's what an email needs to carry - not more information, tips or hacks. The specific texture of how you think. That's what someone signed up for. 

It is also what's missing when a VA or a generic prompt pack takes over the content after the lead magnet. The audience already knows who they signed up for. If that person doesn't show up in the emails and content afterwards, they notice. They just don't tell you. They stop opening your emails.

This isn't an argument for skipping list growth. A small engaged list and a large cold one are different problems. But if you're building the list without building the voice, you're just making the back end problem bigger with every new subscriber.

I saw this clearly on a call recently. 30,000+ person email list, five-figure summit registrations, sixty-plus podcast appearances. 

Front end working well by every visible measure. He'd had a VA producing content for two years. A mutual contact described the emails as "beige AF”, well-researched, properly formatted and…completely personality-free. 

You could not tell from reading it that a specific human being was behind it. He recognized it immediately when I named what I was seeing - that the person people had signed up to hear from had disappeared from the content. He said: yeah, you’re right, after the summit, I turned into a robot. 

10% open rate on a 30,000+ email list is not a distribution problem. It's what happens when the content has nothing to do with the person who built the audience.

The instinct when content stops working is to fix the content. Better prompts, better hooks, better posting schedule, better VA. But that's like trying to fix a garden by cleaning up the flowers - trimming and rearranging what's visible when the soil underneath has nothing in it. No nutrients, no water, nothing for roots to draw from. 

You're fixing the wrong problem.

The prompt is downstream of the brand. A well-structured prompt in the hands of someone who knows exactly what they believe, how they talk and what makes them different from everyone else doing similar work - that person gets something useful back. The prompt amplifies what's already there. If nothing's there, you get nothing back. Sure, faster than you could have written it yourself but completely useless.

Most people haven't built that foundation because filling out a brand framework doesn't produce it. The clarity that makes any tool actually useful comes from sitting with the real questions long enough to stop accepting answers that sound good but could apply to anyone. That's slower than downloading a prompt pack. But if the soil is the problem, that's where the work should start. If you want to do that work, we should sit down for a chat.