Beautiful, fast, real estate websites
We’ve already covered some interesting use-cases for ChatGPT in real estate.
However, I’ve also been hearing about the value of AI in keyword research for SEO, so I wanted to test that idea. Here’s how I did it…
First: I asked ChatGPT to generate a list of 20 keywords relating to Lexington, Kentucky. The resulting list was… very boring.
So I tried again. This time I specified that I wanted common search terms of various lengths (long-tail search terms and short-tail) with varying content and intent.
Here’s the result:
Second: I asked ChatGPT to “arrange the above keywords based on topic, search intent, and predicted competitiveness”. The result was a mess, and the chatbot simply turned down the opportunity to even attempt the third task. So that was a failure.
Not giving up, I then tried a few prompt variations, before trying this one:
Arrange these search keywords into three horizontal columns. Column headings should be “keyword”, “topic”, and “intent” (navigational, informational, commercial, or transactional). Order them based on expected search volume: [ list of keywords]
That got me this:
(Just showing half to save space)
As you can see, it gets the topic right, but the intent and expected volume aren’t great. (For example, shorter and more general keywords tend to have higher volume, so having “Lexington schools” below a search for “Top-rated schools” is probably incorrect)
Maybe the instruction wasn’t explicit enough, so let’s see if we can refine that…
I primed ChatGPT’s contextual understanding by asking it about search intent. After it gave an answer, I told it to use that information to make another attempt. There was some improvement, but it still listed “Lexington weather” as “Navigational” (ie: looking for a specific website).
Anyway - as you can see, AI is capable of quickly generating and then sorting keywords by some criteria, so it could be useful in your SEO expert toolkit.
But it’s also very poor at assessing the purpose or intent that might underly those search queries, which is a vital part of modern SEO.
No real surprise there. We’ve already covered some of the ways that AI can’t yet match human SEO. Still…a bit disappointing?
What do you think - was this result surprising at all?
After asking ChatGPT a few questions, I asked it to try again with the added stipulation to “incorporate what you have said about assessing search volume”. The output…pretty good I think: