AI-Driven Pumpkin Spice

This data was originally featured in the August 30, 2023 newsletter found here: https://www.trustinsights.ai/blog/2023/08/inbox-insights-august-30-2023-foundations-of-innovation-ai-driven-pumpkin-spice/.

This week, it’s pumpkin spice season! Well, specifically, it’s the peak of search inquiries about all things pumpkin spice. However, we’re going to do things slightly differently this year, rather than just our usual roundup. That’s a good place to start though; using our basket of common pumpkin spice-related search terms, let’s see what the actuals and forecast look like for 2023:

What we can observe, especially with the dark trend line, is that after the first year of the pandemic, all things pumpkin spice have been ascendant. Searches have increased year over year.

Now, that’s helpful to understand; certainly, if we were a merchant interested in selling pumpkin spice-related merchandise, this would help us understand the timing of when to push our marketing hardest. But suppose we wanted to understand things at a more granular level? Maybe we were in charge of SEO for the merchant – what keyword anomalies could we spot in the data?

There are innumerable ways to do anomaly detection in a dataset, and trying them all could be pointless. What’s a data-driven marketer to do? Ask an AI, of course. We took our dataset and loaded it into ChatGPT’s Advanced Data Analysis module (formerly Code Interpreter) and asked it for its suggestions:

While I instinctively leaned towards ARIMA-based methods, it was nice to see the machinery make similar recommendations. I then asked it to draft the code, modified it to work the way I wanted it to, and this is what it came up with:

What’s interesting here is the dramatic jump in basic informational queries like “what is psl”. For content marketers looking for an advantage, this might be one way to identify popular but underserved topics in any domain, then produce content for that topic rapidly.

Our generative AI tools are good at more than just cranking out content. With the more sophisticated data analysis capabilities built into many of them, they can be powerful allies for analytics as well.