Speech-to-text transcript of the AutoContent API example "blog-post-to-corporate-podcast" (Whisper large-v3-turbo, not edited; no speaker labels). Welcome to today's deep dive. For you, the marketing and SEO professionals listening, our mission today is to explore how to navigate programmatic SEO in the era of AI overviews using a July 2026 article from Auto Content API. So, okay, let's unpack this because AI overviews haven't actually killed programmatic SEO, right? They've just made, well, weak programmatic SEO a lot easier to ignore. Yeah, exactly. I mean, the old strategy was essentially just this giant find and replace machine. You'd swap out a city name, like changing Austin to Dallas, or swap a few adjectives and then just call it a day. Right, exactly. So since the landscape has totally shifted, how do we actually define the new baseline for a successful page? Does it require fundamentally changing the page's entire purpose? Well, what's fascinating here is that the new bar is something called source-backed usefulness. So, for example, if you have a page about, say, the best CRM for agencies, it can't just repeat keywords anymore. It really has to cover natural follow-ups. Oh, you mean like migration costs or, I don't know, data ownership? Yeah, exactly that. And this introduces the useful page test. Basically, if a reader finishes your page and immediately has three logical follow-up questions, well, does the page already contain the evidence and the next steps to answer them. If not, it fails. Oh, wow. Okay, wait, if answering all those follow-up questions is the goal, where does legitimate automation end and like actual spam begin? Because here's where it gets really interesting. Google's scaled content abuse policy is completely format neutral, isn't it? It is, yeah. It applies to AI, scraped feeds, and even hand-built pages if they just lack value. So there are three rules for a safe page. First, it needs a unique source set, like a custom data set or a benchmark. Right, so it can't just be rehashed info from somewhere else. Exactly. Second, it has to solve a unique user problem. And third, it needs a visible reason to exist, maybe an interactive calculator or a checklist. Honestly, if a page cannot name its unique evidence, you just do not publish it. So what does this all mean for our publishing calendars? I mean, knowing what to avoid, how do we plan content around the way AI search expands queries? You know, what Google calls query fan out. Should we still be targeting long tail variations? Absolutely not one by one. That is a trap. Instead, you need to group those fan out queries by the user's decision stage. So structuring it around things like definition, comparison and implementation. Yes, and also evidence and next action. And this leads to the safe generation checklist. You have to map every page to first party sources, use one canonical page for a topic cluster, and always ensure human review. Oh, and measure conversions and URL cannibalization instead of just page count. All right. Page count is just a vanity metric now. So synthesizing all this, we really need to stop creating duplicate pages for every single wording variation. Just build one canonical source back page and then repurpose it into supporting assets like slide decks or X posts. Spot on. If we connect this to the bigger picture, you're building a hub. You are consolidating authority and making sure the AI search actually points back to you as the primary source. Well, that gives us a lot to think about. So I want to leave you all with a final question to ponder. If your automated SEO pages can only answer questions that someone else has already answered the exact same way, is your content actually solving a problem for the user or just taking up server space?