Unlocking ChatGPT’s Full Potential With MarketMuse Questions Data

ChatGPT can perform some amazing feats, but the challenge is getting it to do the right stuff. If you use MarketMuse, you can do that with an easy workflow taking advantage of both platforms’ powerful AI technology.

It’s a simple process that uses MarketMuse to determine what content to create and then gets ChatGPT to generate and initial draft. The keyword being initial draft, as in you’ll want to use that as a starting point for your own writing. Right?

So here’s something to get you started in further exploring ChatGPT and creating more useful prompts with the help of MarketMuse data.

Research a Topic in MarketMuse

For this exercise I chose the subject “how to create a successful content marketing strategy” because I know readers of this blog are familiar with the subject. With the help of MarketMuse Research application I could find a few dozen questions related to that topic.

Combing through that list, I settled on five I thought would be best answered in an article on creating a successful content marketing strategy. Intentionally wanting to generate a short article, I felt five questions would be enough.

Create a Prompt for ChatGPT

I assembled all the questions in my notes so I could copy and paste them as a complete prompt into ChatGPT. I find it best to assemble everything into a prompt first. You can you paste them individually into ChatGPT and it’ll respond to each one; it just depends on your workflow preference.

Here’s what the prompt looks like. It’s pretty simple — I don’t specify my audience, the purpose of this article, or any other parameters. I just ask ChatGPT to create a 500-word article (allowing for about 100 words per question) for a specific topic making sure to answer the five questions.

If find it very interesting how it stopped mid-sentence, having‌ gone about 10% over the word limit. But that’s the way it works — being a linear generation method, it lacks the ability to organize and structure anything on a high level.

Now let’s take the output from ChatGPT and put that into MarketMuse to evaluate the topical coverage.

Optimize in MarketMuse

Here is the output pasted into MarketMuse Optimize. I made sure to use the same topic as I did in Research in order to properly evaluate it.

What a surprise! While the average page takes 2,839 words to reach a Content Score of 37, this generated output only uses 560 words to hit a Content Score of 25 — that’s pretty concise. While it’s certainly doing well from a topical coverage standpoint, and there are no grammatical errors, I’d still treat this as raw input for the writing process.

Although I didn’t specify any sections within the article, the questions lend themselves well to use as headings. You could copy the questions verbatim as subheadings, but that’d be rather dry. So I’d probably change it up a little, maybe something like this:

• How to create a successful content marketing strategy? -> Creating a successful content marketing strategy.
• What types of content should be included in your content strategy plan? -> ‘X’ types of content to include in your content strategy plan.
• How do you structure a content marketing plan? -> Here’s how to structure a content marketing plan.
• How do you structure a content strategy? -> Structuring your content strategy.

Takeaways

Well-structured prompts are critical to getting good output from a large language model like ChatGPT. Part of that involves making sure the output answers important questions — the type a reader would want answered when consuming that content. MarketMuse data, like that available in Questions, can make that happen.