{"id":6137,"date":"2024-01-12T18:02:31","date_gmt":"2024-01-12T18:02:31","guid":{"rendered":"https:\/\/tup.kxe.temporary.site\/i-tested-premium-ai-prompts-to-see-if-theyre-worth-it-theyre-not\/"},"modified":"2024-01-12T18:02:31","modified_gmt":"2024-01-12T18:02:31","slug":"i-tested-premium-ai-prompts-to-see-if-theyre-worth-it-theyre-not","status":"publish","type":"post","link":"https:\/\/okdesign.ca\/en\/i-tested-premium-ai-prompts-to-see-if-theyre-worth-it-theyre-not\/","title":{"rendered":"I Tested Premium AI Prompts To See if They\u2019re Worth It. They\u2019re Not."},"content":{"rendered":"


\n<\/p>\n

\n
\n
\n
\n
\n

Get the week’s best marketing content<\/p>\n<\/div><\/div>\n<\/div>\n<\/div>\n

<\/p>\n

We spent ~$80 to purchase five premium ChatGPT prompts and ran a blind test among the members of our marketing team to see if they were worth\u00a0it.<\/p>\n

Long story short: They aren\u2019t.<\/p>\n

The experiment<\/h2>\n

Here\u2019s what I\u00a0did:<\/p>\n

    \n
  1. I signed up for a service selling ChatGPT prompts and selected five SEO\u00a0ones<\/li>\n
  2. I created simple versions of these prompts (27 words on average, vs. 227 for the premium prompts)<\/li>\n
  3. I entered these prompts into ChatGPT<\/li>\n
  4. I took the output and did a blind poll, asking the Ahrefs marketing team to guess which output was better.<\/li>\n<\/ol>\n

    A total of 15 people took the\u00a0poll.<\/p>\n

    Here are the results:<\/p>\n

    \"Most<\/figure>\n

    The premium prompts only convincingly \u201cbeat\u201d the simple prompt for two\u00a0tasks.<\/p>\n

    Considering that you\u2019ll need to cough up money (in some cases, a lot) for these pre-engineered prompts, I\u2019m skeptical they\u2019re worth it at all. I don\u2019t think you\u2019re missing out if you don\u2019t\u00a0pay.<\/p>\n

    The anatomy of a \u201cpremium\u201d prompt<\/h2>\n

    Pretty much all of the \u201cengineered\u201d prompts we tested had a similar structure.<\/p>\n

    I tried to break down what I\u00a0saw:<\/p>\n

    \"The\"The<\/figure>\n

    Specifically, here are the patterns:<\/p>\n

      \n
    1. Role<\/strong> \u2014 Asks ChatGPT to act as a professional or expert with specific skill\u00a0sets.<\/li>\n
    2. Target language<\/strong> \u2014 Acts ChatGPT to respond in a particular language.<\/li>\n
    3. Ultra-specific instructions<\/strong> \u2014 Asks ChatGPT to do XYZ action in extreme detail.<\/li>\n
    4. Emphasis on human-like writing<\/strong> \u2014 Asks ChatGPT to write in a specific tone or\u00a0style.<\/li>\n
    5. Specific output<\/strong> \u2014 Asks ChatGPT to structure its response in a specific format, such as markdown, tables, code boxes, etc.<\/li>\n<\/ol>\n

      The takeaway<\/h2>\n

      This experiment is by no means academic nor scientific. It\u2019s just a fun, informal test to see if \u201cengineered\u201d prompts could perform better than a simple command\u2014at least<\/em> according to human interpretation.<\/p>\n

      Based on the results, though, I\u2019m not convinced that \u201cpremium\u201d prompts really <\/em>do that much for you. It\u2019s probably better to start off simple and then refine accordingly.<\/p>\n

      So, don\u2019t waste your money. Use it to upgrade to ChatGPT Plus instead.<\/p>\n

      What do you think? Let me know on Twitter<\/s> X<\/a>.<\/p>\n

      <\/span><\/div>\n