You shouldn’t use AI to generate something unless you know how 80% of the output should look.” — Sara Taher
Sara Taher is the founder of SEO Riddler, a fast-growing SEO newsletter and community, and a 12+ year SEO veteran who currently works in-house at a major publisher. She’s also the founder of a women-inclusive (but open-to-all) SEO community that hosts a popular monthly meetup, and the builder of several independent SEO tools, including a schema tool built for the agentic web. Sara’s path into SEO started by accident — through a Y Combinator-backed local marketplace startup — before she found her footing as one of the industry’s most hands-on voices on data, schema, and AI literacy.
Episode Summary
In this episode of Beyond the Search Bar, host Russ Macumber talks with Sara Taher about why she believes most SEO experiments are fundamentally broken — from unreliable Search Console data to Google’s own patented use of social engineering against site owners. Sara unpacks her nuanced take on schema markup, walks through her accidental entry into SEO via a YC-backed local commerce startup, and explains what led her to build one of the industry’s most active SEO communities. The conversation turns to AI: Sara shares hard-won lessons from her master’s program on why AI output can’t be trusted blindly, why understanding your data is a prerequisite to using AI on it, and how businesses like Starbucks and Amazon are already learning costly lessons about over-automating. It’s a candid, experience-driven conversation about skepticism, adaptability, and staying relevant in an industry that keeps changing the rules mid-game.
Key Takeaways
- Most SEO Experiments Are Broken: Sara argues that flawed methodology, incomplete Search Console sampling, and inaccurate metrics undermine a lot of published SEO “findings.”
- Google May Be Using Social Engineering: Patents suggest Google can show sites a temporary traffic drop to see how site owners react to spammy backlinks — a tactic Sara says complicates any simple cause-and-effect testing.
- Schema Has Value, But Isn’t Always a Priority: Sara sees genuine value in structured data but deprioritizes it when it competes with higher-impact dev resources, in part because its ROI is hard to prove.
- Entrepreneurial Experience Shapes Better SEOs: Sara’s accidental path into SEO — via a failed but formative YC-backed startup — gave her an instinct for tying SEO effort to business return that she believes separates strong SEOs from average ones.
- Community Over Cliques: Sara built her SEO community and monthly meetup to counter what she sees as an industry that too often rewards existing relationships over undiscovered talent.
- AI Literacy Requires Data Literacy First: Citing her master’s program professor, Sara stresses that using AI well requires already knowing what a good output looks like — otherwise there’s no way to catch errors, especially with tabular data. Automation Missteps Are Getting Expensive: From Amazon’s AI-related outage to Starbucks walking back over-automation, Sara points to real-world cautionary tales showing companies are still learning where AI helps and where it hurts.
- AI Can Expand Teams, Not Just Replace Them: Echoing a conversation Sara had with fellow guest Fery Kaszoni, she notes that AI-driven productivity can let businesses serve more clients affordably — leading to more hiring, not less.
Timestamped Summary
- 01:24–02:40 – Introduction and housekeeping; Russ sets up the conversation with Sara Taher.
- 02:40–04:55 – Sara’s hottest take: most SEO experiments are broken, and Search Console data can’t be trusted at face value.
- 04:55–06:24 – The Google patent suggesting social engineering tactics against sites with spammy backlinks.
- 06:24–08:43 – Russ and Sara compare notes on how UX and on-page experience seem to influence ranking stability in ways Google won’t confirm directly.
- 08:43–10:30 – Sara’s real take on schema: valuable, but not always worth prioritizing, and easy to abuse.
- 10:30–12:13 – Where to find Sara: her newsletter, SEO Riddler, and LinkedIn.
- 12:13–16:05 – Sara’s accidental path into SEO, and the story of StoreLeaks, her YC-backed local marketplace startup.
- 16:05–17:37 – Lessons learned from shutting down the startup right before the pandemic, and how it shaped her approach to measurement.
- 17:37–21:19 – The origin of Sara’s SEO community, built to counter cliquey industry dynamics and create equal footing for undiscovered talent.
- 21:19–24:40 – The best part of the community: a monthly meetup and the hidden talent Sara’s discovered through it.
- 24:40–27:27 – Sara’s current focus on staying relevant — schema tooling, upskilling, and a master’s program insight on why “the statistical prior” makes blind AI trust risky. 27:27–29:44 – Why you shouldn’t generate AI output unless you already know roughly what 80% of it should look like, illustrated by real project failures with AI and tabular data.
- 29:44–32:29 – The calculator analogy: why foundational skills still matter, and a story about a kid at a farmer’s market miscalculating change.
- 32:29–34:44 – What’s exciting Sara right now: the industry finally having honest conversations about what AI can’t do.
- 34:44–36:15 – Closing thoughts: Starbucks’ automation pullback, and how Fery Kaszoni’s AI-driven productivity gains led to more hiring, not less.
- 36:15–36:41 – Wrap-up and where to find Sara Taher.
