<- all tokdocs

The AI "Money Prompt" for Finding Funnel Leaks: What the Clip Shows

By , founder of Mavric and Apso

View on TikTok ->

The "money prompt" is an AI prompt that, according to a Marketing School clip posted by @levelingupofficial, reads a product's code and funnel, finds where users drop off, and flags missing lifecycle messages. The speaker says it found no activation emails, no credit top-off reminders, no offer, and one page where about 20% of people abandoned. The clip never shows the prompt, the output or the product, so none of those findings can be checked. Its real argument is that a person with funnel experience still has to interpret what the AI finds.

What is the money prompt?

The only description in the audio is this: "it's called money prompt. So what it does is analyzes your code and analyze the entire funnel, it looks at drop off rates, and it comes up with where your funnel is failing."

The clip shows no prompt text, chat window, code editor or analytics screen. A card footer reads "PROMPT 01 / 03", which suggests a three-part series. No model or tool is named, so you cannot reproduce the exact prompt from the video.

How to run a funnel audit with AI

The post description gives a five-step version you can follow:

  1. Give the AI your code and your full funnel.
  2. Ask for drop-off rates at each step.
  3. Check stall points and low-credit moments for emails, reminders or offers.
  4. Fix the biggest abandonment first.
  5. Walk your engineers through the numbers.

One gap matters if you try this. Code can show whether an email trigger exists, because the codebase either contains it or does not. Drop-off rates cannot come from code. They need event data from your analytics. The clip never says how that data reached the AI or whether the AI calculated the rates.

What leaks did it find?

The speaker says "for one of our products, you're actually not sending activation emails, if people are failing to move to the next step, you're actually not reminding people to top off their credits, if they're running low on credits, and you're not giving them a deal either."

Cards list "01 No activation emails" and "02 No credit top-off reminders". The product is never named. The mention of credits points to a usage-based product, but there is no page, changelog or email sample to check.

Is a 20% drop-off on one page high?

The audio says "you have like a 20% of people abandoned at this page over here." A card shows "ONE PAGE / 20% / of people abandoned here".

The clip does not say which page, whether the 20% is a share of visitors, signups or checkout starts, what period it covers, or where the figure came from. Without that context, there is no way to say whether 20% is high or low for that step.

How many leaks were there?

The description says "Four leaks, one pass." The audio names three missing messages plus the abandonment page, which makes four if the page counts as a leak. The card footer reads "LEAKS 02 / 03", which counts three. The clip also does not explain "one pass": how long it took, what files were involved, or whether anyone corrected the output.

The description adds that "a single prompt can now surface things that used to take a full audit." It gives no audit time or cost to compare against.

Why does experience still matter?

This is the clip's main point. The speaker says "when I walk through those numbers with the engineers, they're like, well, we don't know this stuff, right? I'm like, exactly. I can't expect the engineers to know that." He concludes "it takes years and years of experience still to detect these patterns."

Two readings fit. Engineers may not know marketing patterns like lifecycle emails and offers. Or the speaker's experience was needed to judge the AI's output. The description takes the second view, calling experience "what makes it profitable." The clip shows no case where the AI was wrong and experience corrected it.

The description makes one useful observation: these leaks are "things that aren't happening." Missing emails do not show up as errors, which is why a checklist of expected messages is worth running.

Bottom line

Asking an AI to compare your code against your funnel for missing activation emails, reminders and offers is a reasonable audit, and the five-step version in the description is easy to try. Feed it real analytics data for drop-off rates, since code alone cannot supply them. The clip's specific findings, including the 20% page, come from an unnamed product with nothing shown on screen.

Sources

Published October 31, 2026. Based on a TikTok by @levelingupofficial. Full video notes (markdown)