---
title: "Kitchen Side: The Hidden Drivers of AI Search"
id: "7571"
type: "post"
slug: "the-hidden-drivers-of-ai-search"
published_at: "2026-07-08T12:00:00+00:00"
modified_at: "2026-07-06T19:52:15+00:00"
url: "https://beomniscient.com/blog/the-hidden-drivers-of-ai-search/"
markdown_url: "https://beomniscient.com/blog/the-hidden-drivers-of-ai-search.md"
excerpt: "Listen to the podcast: In this Kitchen Side episode, Alex..."
taxonomy_category:
  - "Podcast"
  - "SEO"
---

## **Listen to the podcast:**

In this Kitchen Side episode, Alex Birkett, Allie Decker, and David Ly Khim unpack how brands should actually measure success in AI search, and why traditional attribution models are breaking down as buyer behavior shifts from search links to AI assistants and workflows.

They discuss why self-reported attribution is becoming the most reliable signal available, the different tiers of AI visibility from simple citations to category-level consensus, and why AI visibility functions more like a brand recall survey than a channel you can optimize in isolation. The conversation also covers the idea of a “post-channel marketer” who coordinates across product, customer education, and PR, and why chasing visibility tactics with no underlying business purpose rarely pays off.

## **Key Takeaways**

- Self-reported attribution is more reliable than clickstream data for AI search because most AI-driven visits never result in a tracked click.
- AI visibility behaves more like a brand recall survey than a channel, reflecting how a brand compares to competitors rather than something to optimize in isolation.
- Picking up a citation on a narrow, low-competition prompt is easy, but shifting broader category-level consensus, like being named among the best CRMs, is far harder and requires changing an entire conversation.
- Correcting outdated third-party pricing information moved AI search outputs within about a month in one client engagement, though this was a sentiment fix rather than a visibility gain.
- Some AI answers now include negative or qualifying recommendations, meaning a mention doesn’t always translate into a positive outcome for a brand.
- Different AI tools serve different roles, with ChatGPT and AI Overviews functioning as quick answer engines while agentic tools like Claude Code perform deeper multi-step research, and visibility strategies should account for that difference.
- Whether a company should publish a markdown version of its site for easier AI retrieval depends on its audience, mattering more for technical, developer-facing products than for typical B2B SaaS buyers.
- Sustainable AI search performance comes from investing in product experience, customer education, and reviews, since genuine third-party validation is difficult to fake once a brand has earned it.
- Marketing teams need a “post-channel” role that coordinates across product, customer success, and PR to influence sentiment and visibility, rather than treating this as an SEO-only or PR-only problem.
- Tactics pursued only to move an AI visibility score, with no underlying business purpose, are usually not worth the effort unless they’re addressing existing negative sentiment.

## **Show Links:**

- Connect with David Khim on [LinkedIn](https://www.linkedin.com/in/davidlykhim) and [Twitter](https://twitter.com/davidlykhim?lang=en)
- Connect with Alex Birkett on [LinkedIn](https://www.linkedin.com/in/iamalexbirkett) and [Twitter](https://twitter.com/iamalexbirkett)
- Connect with Allie Decker on [LinkedIn](https://www.linkedin.com/in/alliecdecker) and [Twitter](https://twitter.com/alliecdecker?lang=en)
- Connect with Omniscient Digital on [LinkedIn](https://www.linkedin.com/company/omniscient-digital) or [Twitter](https://twitter.com/beomniscient_)

## **Time Stamps:**

- **[00:00]** – Intro and episode overview: measuring success in AI search
- **[04:41]** – Why “is it working” has become the bigger question than “how do we show up”
- **[05:34]** – The case for self-reported attribution over clickstream data
- **[06:28]** – AI visibility as a mirror or brand recall survey
- **[07:25]** – Analyzing lead data to see how often AI referrals go untracked
- **[11:44]** – How quickly results can show up in AI search
- **[12:42]** – A study on “poisoning” AI answers with just 13 words
- **[14:40]** – Case study: correcting third-party pricing information shifted AI outputs within a month
- **[16:34]** – The ladder of AI search results, from easy citations to hard-won category consensus
- **[20:45]** – Negative AI recommendations and exclusion from the consideration set
- **[29:32]** – Why different AI tools, like ChatGPT vs. Claude Code, need different visibility strategies
- **[35:49]** – Probability engineering and the “post-channel marketer” concept
- **[44:25]** – Should you do something only to move an AI visibility score?

## Get the Field Notes

Weekly learnings from working on B2B content & SEO for dozens of companies.