This is an amazing prompt engineering approach when you need to feed a large amount of data and get even a bigger output. 
Great to use for writing SEO articles, landing pages and completing complex briefs.
### INITIAL PROMPT
> Act as \[Role\]. You'll do \[TASK\]. I'll provide you with \[categories of your data\]. You'll \[PERFORM A TASK\]. Don't write anything yet. **Is that understood?** {color="green_bg"}
### FEEDING YOUR DATA (FOLLOWUP PROMPTS)
> Great, now I'll provide you with \[your data like keywords/audience segments/user persona/etc\]. My \[DATA TYPE\]:<br>\<provide your data\><br>Next I'll provide you with \[DATA TYPE\]. **Don't write anything yet. Is that understood?** {color="green_bg"}
### FINAL PROMPT
> Now, perform \[TASK\]. Don't forget about \[types of data you provided\]. {color="green_bg"}
<empty-block/>
## EXTRA TIP
Beware of the token size (the current memory with GPT-4 is around 25,000 words). To avoid expiring a token size, simply edit your last prompt instead of adding a new one.