We spent ~$80 to buy 5 premium ChatGPT prompts and ran a blind check among the many members of our advertising and marketing group to see in the event that they have been value it.
Lengthy story brief: They aren’t.
The experiment
Right here’s what I did:
- I signed up for a service promoting ChatGPT prompts and chosen 5 website positioning ones
- I created easy variations of those prompts (27 phrases on common, vs. 227 for the premium prompts)
- I entered these prompts into ChatGPT
- I took the output and did a blind ballot, asking the Ahrefs advertising and marketing group to guess which output was higher.
A complete of 15 individuals took the ballot.
Listed below are the outcomes:
![Most marketers on our team thought the simple AI prompts did better in a blind test](https://ahrefs.com/blog/wp-content/uploads/2024/01/most-marketers-on-our-team-thought-the-simple-ai-p.png)
The premium prompts solely convincingly “beat” the easy immediate for 2 duties.
Contemplating that you just’ll have to cough up cash (in some instances, so much) for these pre-engineered prompts, I’m skeptical they’re value it in any respect. I don’t suppose you’re lacking out when you don’t pay.
The anatomy of a “premium” immediate
Just about all the “engineered” prompts we examined had the same construction.
I attempted to interrupt down what I noticed:
![The anatomy of an "engineered" prompt](https://ahrefs.com/blog/wp-content/uploads/2024/01/anatomy-of-a-premium-ai-prompt.png)
Particularly, listed here are the patterns:
- Function — Asks ChatGPT to behave as an expert or skilled with particular talent units.
- Goal language — Acts ChatGPT to reply in a specific language.
- Extremely-specific directions — Asks ChatGPT to do XYZ motion in excessive element.
- Emphasis on human-like writing — Asks ChatGPT to jot down in a selected tone or fashion.
- Particular output — Asks ChatGPT to construction its response in a selected format, equivalent to markdown, tables, code packing containers, and many others.
The takeaway
This experiment is on no account educational nor scientific. It’s only a enjoyable, casual check to see if “engineered” prompts may carry out higher than a easy command—at the least in keeping with human interpretation.
Based mostly on the outcomes, although, I’m not satisfied that “premium” prompts actually try this a lot for you. It’s in all probability higher to start out off easy after which refine accordingly.
So, don’t waste your cash. Use it to improve to ChatGPT Plus as a substitute.
What do you suppose? Let me know on Twitter X.