Steve Toth, an SEO consultant and creator of SEO Notebook, argues that brand trust is the primary ranking factor for Answer engine optimization (AEO), noting how he grew a subscriber base of 27,000 organic search professionals by establishing a trusted identity. Toth’s experience as the strategy lead for FreshBooks, where the site ranked #1 for “invoice template” with 300,000 searches per month, proves that brand authority outweighs content volume. In-house SEO managers and agency owners should act on this, while beginner bloggers can skip it.
Who should buy an AEO strategy and who should skip it
Readers who should buy an AEO strategy
- In-house SEO Managers should act on this to protect brand visibility as search shifts to AI-generated answers.
- SEO Agency Owners should act on this to scale services by building technical capabilities in LLM research.
- Content Strategists should act on this to learn how to get cited in ChatGPT and other LLMs using multi-channel signals.
- Technical SEO Specialists should act on this to understand how to see what an LLM sees through HAR file analysis.
Readers who should skip an AEO strategy
- Beginner bloggers should skip this and focus on basic keyword research and meta tags instead.
- PPC-only advertisers should skip this and focus on Google Ads or Meta Ads tutorials.
What to do next about an AEO strategy
- Record browser sessions using the Network tab to capture HAR files for forensic LLM analysis.
- Verify LinkedIn AI titles by viewing posts in an incognito browser to compare generated H1s against the source content.
- Include a video, even a five-word video, in LinkedIn posts to increase the staying power of content for AI retrieval.
- Book an AEO session for $5.97 for 25 minutes to receive a personalized strategy and library of resources.
Brand trust is the foundation of modern SEO
Brand trust acts as the primary signal for Google to verify authority when generating AI Overviews or answering queries in Google AI Mode. If a brand has no identity, how can an LLM verify its authority? Toth argues that “perfect” content on a website without a strong brand presence will fail to surface, meaning that identity-based signals—including social media activity and YouTube—are now prerequisites for ranking. Think of brand trust as a digital passport; without it, your content cannot cross the border into the AI’s answer set. Brand building creates a moat that ensures content staying power, making it more effective than traditional keyword density.
LinkedIn’s role in AI search and organic ranking
LinkedIn provides a unique source of data for AI search optimization because the platform often generates its own AI-based H1 and title tags for ranked posts. These generated tags can be seen by viewing a post in an incognito browser, where they differ from the logged-in view. Strategy for ranking on LinkedIn includes using the first sentence of a post as the target keyword and including even a five-word video to maintain staying power. Analysis suggests that LinkedIn acts as a content feeder for LLMs, where high-engagement, multi-modal posts are more likely to be cited in AI-generated snippets.
Forensic SEO: Analyzing LLM behavior via HAR files
Forensic SEO involves using HAR files—HTTP Archive files that record every event between a browser and a server—to see how LLMs process queries. Capturing these files allows practitioners to see the specific “fanout queries” an LLM uses to retrieve and synthesize data. Toth warns that these files contain cookies and other sensitive data that can be used to spoof an identity, so they should never be uploaded directly to public LLMs like ChatGPT. Specialized parsers and local LLM hosting provide a safe way to analyze LLM queries without risking cybersecurity violations or account flags.
What is a HAR file?
A HAR file is a JSON or plain-text file that records every interaction between a web browser and a server during a session. It captures data like request headers, response bodies, and timing, allowing developers to see exactly what information an application sent or received. For SEO, it reveals the hidden API calls and data fetches that an LLM makes during a search.
Security risks and the move to local LLM hosting
Local LLM hosting allows researchers to perform deep-dive analysis on LLM endpoints without the risks associated with public AI tools. Toth and his team are building a local environment using NVIDIA DGX Spark hardware to avoid being flagged by providers like Anthropic for “suspicious” reverse-engineering queries. The initial hardware cost for this local setup is approximately $35,000 USD, with an additional $30,000 planned for Mac Studios with 512 GB of RAM, totaling a projected $65,000 USD in hardware investment. This infrastructure allows for the forensic analysis of LLM behavior while keeping private data and cookies entirely off public servers.
Scaling authority through consistent publication
Consistent publishing serves as a critical signal for maintaining audience retention and building the brand trust necessary for AEO. Toth has maintained a 7-year streak of publishing his newsletter at the same time every week to ensure his 27,000 subscribers know when to expect his content. Building a community on platforms like Reddit, LinkedIn, or Slack creates a reputation that makes a new newsletter launch successful from day one. Reliability in publishing ensures that the brand remains top-of-mind for both humans and the data-crawling agents that feed into LLM memory.
A brand manager uses HAR files to identify AI search gaps
Sarah is a brand manager for a mid-sized consumer goods company that is struggling to appear in AI Overviews. She begins by recording a HAR file while querying a popular LLM about her company’s primary product category. She uses a specialized parser to identify the specific fanout queries the LLM uses, discovering that the model is prioritizing a competitor because of their higher brand trust signals. Sarah decides to shift her strategy from creating more blog posts to a multi-channel brand-building campaign on LinkedIn and YouTube. If Sarah had instead uploaded the raw HAR file directly to a public LLM to ask for advice, she might have risked her identity being spoofed due to the exposed cookies in the file.
What Steve Toth has not confirmed about local LLM hardware and software tools
- Steve Toth has not confirmed the final total cost of the combined local LLM hardware setup.
- Steve Toth has not specified the exact software tools used in his custom parser for stripping PII from HAR files.
Frequently Asked Questions
How does Steve Toth get rankings on page one of Google?
Steve Toth achieves page one rankings by building a strong brand that Google knows and trusts, supported by social media and multimodal content. This brand-led approach ensures that his content is viewed as an authoritative source by both traditional and AI-driven search engines.
Is it safe to upload HAR files to ChatGPT?
Uploading HAR files to ChatGPT is not safe because they contain cookies that can be used to spoof your identity or grant access to your accounts. These files should be stripped of personally identifiable information or analyzed through a local LLM setup instead.
How much does it cost to build the local LLM setup described?
The local LLM setup costs approximately $35,000 USD for the initial hardware, with an additional $30,000 planned for Mac Studios. This brings the total projected hardware investment to $65,000 USD for a professional-grade local research environment.
How does LinkedIn handle titles for ranking posts?
LinkedIn often writes an AI-based H1 and AI-based title tag for posts when they are viewed in an incognito browser. Users can verify this by comparing the title shown in a logged-in session to the title displayed to an unauthenticated viewer.
Details first shown in Steve Toth 🔴 LIVE SEO Consulting Session by SEO Video Show.