A few months ago, I wrote about schema markup and the importance it’s gaining in the SEO world. With the rise of ChatGPT and related tools, it’s time to take another look: specifically, how SEOs can use ChatGPT to produce structured data in a faster – but still nuanced and useful – way.
As a reminder, structured data in SEO, also commonly called schema markup, is no longer a “nice to have.” It’s a highly recommended SEO technique that can produce higher click-through rates, greater search visibility, faster indexing, and voice search domination. Structured data makes it easier for search engines to contextualize, understand, and accurately match website content to relevant search queries.
One highly effective use case is product schema markup, which helps your product SERP result stand out from other normal search results, drive higher CTR, etc. (note the ratings, reviews, and price points shown in the result below).
Although there are many types of schema markup, another powerful type is FAQ page markup, which helps users more easily find questions to their answers in the SERP, takes up real estate, and encourages higher CTR:
Now that we’ve recapped what structured data is and why it’s an important part of an SEO strategy, let’s talk about how ChatGPT fits into the mix.
If you work in marketing, chances are you’ve been hearing all about ChatGPT lately. If you don’t yet know what ChatGPT is, let’s do a quick review. ChatGPT is a Natural Language Processing (NLP) model developed by OpenAI and built on top of OpenAI's GPT-3 family of large language models. The chatbot is trained to interact in a conversational way, and many SEOs are experimenting with using it for many tasks such as research, content development (creating briefs, blog title ideas, etc.), metadata creation, keyword research, and structured data, among others.
So, how exactly can ChatGPT be used for structured data creation? Using our examples above, you can ask ChatGPT to write product or FAQ structured data for you. However, if you don’t know what specific information you need for certain types of structured data creation, you run the risk of creating inaccurate markup. Therefore, I recommend applying a human layer of analysis and knowledge first. In other words, you still need to do a little bit of legwork.
Start with a sitewide audit to identify structured data opportunities (product schema, FAQs, how-tos, etc.) for your website. This means knowing what kind of structured data options you have available, then assessing which make sense on each relevant page. You’ll also need to know what information you need for each type of structured data.
For example, with FAQs, you could ask ChatGPT to “write JSON-LD for the following FAQs.” You would still need to type out the questions and answers you want to use. However, if you had an FAQ page to copy or a list of FAQs already typed out, this could be a really quick task.
For example, I asked ChatGPT to write structured data around my coffee FAQs as shown below:
And here is the output:
For product schema markup, you would need to gather a little more information before asking ChatGPT to help you write the structured data. For example, you’ll probably want to gather your product’s name, description, offer price, inventory, review rating, etc. Then, you can ask ChatGPT to “write JSON-LD structured data for the following product” and list out all of those details you collected on the product.
For example, I asked ChatGPT to write structured data for my prettiest coffee mug product, as shown below with the output:
Instead of manually writing your code or digging into an automation tool, ChatGPT will do the work for you.
Then – and this is true of any process, especially those involving AI – make sure you have an expert handy to review the output. Before you implement that markup on the page, add the AI-created schema markup snippet into a structured data verification tool like the schema.org validator to make sure there are no errors or warnings in the code. If not, it’s ready to implement!
As you can see, ChatGPT also gives you the option to copy the code in the top right corner of the code snippet. In my example below, I copied the product schema markup into Schema.org markup validator, and it shows 0 errors and 0 warnings. So, this code snippet would be ready to implement on your page or in Google Tag Manager.
Although there are already other existing automation tools that can help you write structured data code, as long as you have the necessary information handy, ChatGPT could be a useful tool for more quickly automating this process for you.
As you can see, ChatGPT is not the end-all-be-all type of solution for schema markup. You will still need the human expertise to help identify and mold the markup and then to implement it afterward, but ChatGPT can at least help to speed up part of the process.
If you’re interested in hearing more about JDM’s thoughts on and approach to using AI and a heavy dose of human expertise in our marketing campaigns, drop us a line.