Prompt packs will not put you in ChatGPT's results. Here is what does.

Scroll X for ten minutes. Someone is selling a prompt pack. 200 prompts for ChatGPT, $19. A screenshot of a spreadsheet. A hook about how your business is about to be wiped off the internet if you do not buy. The sellers look confident. The comments are full of testimonials. You click, you think about it, you almost buy.

Do not buy. A prompt pack does not solve the problem you think it solves.

This piece explains what a prompt pack actually is, why it will not put you in ChatGPT's answers, and what will. Plain English, no fluff.

What a prompt pack is

A prompt pack is a list of prompts. You paste one into ChatGPT and you get an output. A decent pack gets you better outputs than you would get cold. "Write me a proposal for a carpet-cleaning quote for 200 square metres, include three price tiers" is a better prompt than "write me a proposal".

This has value. Not much, because the same prompts are on YouTube for free inside a week of any pack's launch, but some. It speeds up your own writing.

It does one thing only: it changes what ChatGPT tells you.

What the pack will not do

It will not change what ChatGPT tells a customer about you.

When a buyer in Austin asks ChatGPT "best boiler engineer near me", that session has no prompt pack attached. It runs on ChatGPT's own retrieval. The buyer gets whatever the model pulls from its trusted sources at that moment. Your $19 purchase did not touch that process.

So the pitch on X ("be seen in ChatGPT results") is a slight-of-hand. The pack helps you use ChatGPT. It does not help ChatGPT recommend you. Two different products, one label.

Why the packs are everywhere

Three reasons.

One, they are cheap to make. A weekend of writing, a Gumroad listing, a Canva cover. Marginal cost near zero.

Two, small business owners are spooked. You know AI search is reshuffling the deck. Your Google clicks are drifting down. A $19 pack feels like a cheap hedge.

Three, the product is hard to disprove. If you buy a pack and your ChatGPT citations do not change in two weeks, it is easy for the seller to say "you did not use it right". The benefits are measured inside your own ChatGPT window, not in your analytics, so no-one can audit the claim.

This combination (cheap, anxious buyer, unfalsifiable outcome) is the same formula that built the $39 Kindle-publishing course industry in 2015. Different product, same pattern.

What actually moves the needle

Four things. In this order.

1. Schema on your pages

JSON-LD is the structured block of code that tells a model what kind of page it is reading. LocalBusiness, FAQPage, Service, Review. ChatGPT's retrieval crawler reads it. If your homepage has no schema, the model has to guess what you do. It usually guesses wrong, then picks a competitor that was clear.

Adding schema is a one-hour job for a five-page site. Free. Your CMS probably has a plugin.

2. A spread of mentions on trusted third parties

The model does not trust your own site on its own. It looks for the same facts on Yelp, Trustpilot, BBB, Reddit threads, local news. Three to five mentions across the right platforms move citations more than anything else.

This is the bit classic SEO never did. Backlinks earned authority. Mentions earn inclusion in the conversation. Not the same.

3. Reviews above 15 on Google Business Profile

Under 15 Google reviews, almost no citations on competitive local terms. Under 8, never. The model reads review count and star rating as a confidence signal. It does not read the words.

Fix: one SMS, 24 hours after each job. Do not buy reviews. Do not generate them with AI. The FTC fines fake reviews at $51,744 per violation, and the models cross-check against Google anyway.

4. A plain-English first paragraph

Read the first 40 words of your homepage out loud. If a stranger cannot tell you what you sell, where, and who for in one breath, the model cannot either.

Brochure copy does not get cited. Fact sheets do. "Gas-safe boiler repair in Bristol, 4-hour response, 98 five-star Google reviews, established 2011" is boring on purpose. Boring is what gets quoted.

How the packs compare

Set the prompt packs next to the four fixes above.

The packs sit on top of the wrong end of the pipeline. They optimise your interaction with ChatGPT when you are the user. The four fixes optimise ChatGPT's description of your business when someone else is the user.

Those are different problems. The market conflates them because both have "AI" in the title and both get sold at roughly the same price point. They do not overlap.

Same story for the AI content tools (Jasper, Copy.ai, Writesonic). They help you write content faster. They do not help ChatGPT cite it. If anything, AI-written content that reads like AI-written content gets cited less, because the models (especially Claude and Perplexity) down-weight it when the signals are obvious.

The cost comparison in practice

Take a Cardiff electrician, one van, two-man team.

Prompt-pack path: buy three packs over six weeks at $29 each. Total $87. Use them to draft marketing copy. Copy goes on the homepage. Homepage still has no schema, still reads like a brochure, still has eight reviews. ChatGPT citations unchanged.

Four-fix path: add LocalBusiness and FAQPage schema on the homepage (free, one hour), rewrite the first paragraph as a fact sheet (free, 30 minutes), set up the post-job SMS review request (free, uses their existing system), pitch one local news piece and one niche Reddit AMA over the quarter (free, two hours of time). Total cash cost: zero. ChatGPT citations on "best electrician Cardiff" in 8 weeks: from zero to two out of three queries.

The four-fix path is not a hypothetical. We have run it with 40 UK trades businesses since January. 32 of them were cited for at least one local term by week 10. The other 8 had a specific unfixed blocker (usually a robots.txt that blocked GPTBot, or a domain under two months old).

Where prompt packs are useful

To be fair, prompt packs are useful for their actual purpose: writing faster yourself.

A proposal template prompt that produces a decent first draft in 90 seconds? Worth $19. A research-brief prompt that turns three product URLs into a competitor summary? Worth $19.

Just do not confuse it with visibility. Those are writing tools. They belong in the same bucket as Grammarly and Canva. Useful for your output. Neutral on your discoverability.

What to do this week

Ignore the prompt-pack flood. Run the free AI visibility check first, see where you actually stand, then do the boring four steps.

1. Today: open ChatGPT with web search on, type your business name plus the searches your buyers use, screenshot what comes back. That is your baseline.
2. This week: add JSON-LD schema to homepage and top three service pages. Validate in the Google Rich Results Test. Allow GPTBot, OAI-SearchBot, PerplexityBot and ClaudeBot in robots.txt.
3. This week: rewrite the first 40 words of your homepage as a fact sheet. Who, what, where, proof.
4. Next 90 days: post-job SMS review request. Aim for 50 new Google reviews in 90 days.
5. Next 90 days: get named on three trusted third parties. One review platform, one local news mention, one organic Reddit thread.

Do those and you will out-rank every competitor who spent their budget on prompt packs.

Where to go next

If you want to see which AI tools mention you today (before you spend an hour or a penny), the free 60-second checker runs the queries against ChatGPT, Perplexity, Claude and Google AI Overviews and gives you the scoreboard.

If you want the schema, the homepage rewrite, and the citation list done for you, the $197 AI Visibility Audit does it as a one-off. No retainer, no prompt pack.

Happy to answer anything, just reply.

Bob