How General-Purpose AI is Making Podcast Discovery Harder (And What You Can Do About It)

by Dr. W. Lewis Johnson and Geoffrey Rich

As Shelly Palmer noted in a recent blog post, conventional search traffic is down 30-45%, and is not likely to come back. Instead, people are relying more on general-purpose AI tools such as Google AI Overviews to find information. Users increasingly read the AI-generated summaries and click through to the content less often. This combination of trends is bad news for podcasters who want listeners to discover and listen to their podcasts.

Conventional search engine optimization (SEO) has always had its limitations for content discovery. The new general-purpose AI tools make the situation worse. In this blog we explain why this is so, and what podcasters can do about it.

Fortunately, a new AI technology called content navigation is now available that is specifically designed to promote content discovery and does not suffer the limitations of other tools. We have applied it successfully to the Dr. Laura Schlessinger program, an established show with thousands of podcasts and videos. It can be readily applied to any podcast to improve discovery.

SEO works well only when the listener is able to specify what they are looking for as a set of keywords, and the podcaster is able to anticipate those keywords. In this blog we will focus on the following example query that does not meet these criteria: “I am interested in Dr. Laura Call of the Day podcasts about managing money for married people.” When this query is given to Google it returns a range of irrelevant content, including content that is not by Dr. Laura, that does not relate to married people, etc. Note that the query does not make clear what the listener is really interested in. Are they concerned about how to address financial matters as a team? How to communicate effectively to avoid financial disagreements? A search engine is not able to ask follow-on questions to understand what the listener is really looking for. 

Google’s AI Overview feature does a bit better at understanding the listener’s query, but it still makes no attempt to clarify the listener’s intent. It doesn’t know whether the listener is referring to Dr. Laura Schlessinger or Dr. Laura Dabney, and so it includes some of both podcasts. And it does not help listeners to actually listen to the content. It simply summarizes episodes and says “You can find this episode on Apple Podcasts.”

And then there are cases where general AI tools provide information that is simply not true. For example, if you ask ChatGPT “How many Dr. Laura Call of the Day podcasts are about managing money for married people?” It replies with “There appears to be one Dr. Laura Call of the Day episode that explicitly focuses on money issues between married couples.” There are in fact many Dr. Laura podcasts dealing with aspects of this problem. It is concerning when tools like ChatGPT provide misinformation about podcasts.

Content navigation like. Stacy, Dr. Laura’s AI Assistant, avoids these problems. She asks probing questions as needed to better understand the listener’s question. In this example, Stacy asks whether the listener is interested in issues where one spouse controls the finances or there are disagreements about spending and budgeting. Stacy provides links to content and is able to explain why it is relevant to the listener’s question. The content is automatically transcribed and summarized so that Stacy has accurate information about every podcast and can explain its relevance. For more information read this blog post, or contact alelo.com.