Judith Lewis

Founder · Decabit Consultancy

Key takeaways

9:08AI language models hallucinate data by tokenizing language and filling gaps with plausible but false information, like ChatGPT fabricating monthly traffic numbers and bounce rates for a client report.
8:08Professional SEOs have handled ambiguity for decades through trial and error with Google's search algorithms; this trained judgment lets them catch AI errors that untrained clients miss.
21:21Create reporting templates that tie every metric back to client business goals (revenue, market share, newsletter subscribers) and use them monthly to demonstrate your strategic value, not just data.
16:16Use AI tools and trusted platforms like SE Ranking or Semrush to gather and visualize data quickly, but always double-check findings and add human commentary before presenting to clients.
23:24AI cannot understand your client's internal politics, business structure, or real needs; only human expertise can translate data into actionable insights that protect your job and client relationship.

Questions Judith Lewis answered

If we let an AI tell the story of our SEO results, how do we double-check its conclusions without doing all the work ourselves, and do you trust AI to understand contextual data?

Don't trust AI conclusions; trust data and analysis instead. SEOs have handled ambiguity for decades and can spot problems ChatGPT can't, like identifying broken Google Tag Manager tags. Train instances with guardrails, but always double-check with multiple tools like Screaming Frog, Majestic, and Google Search Console.

Judith Lewis: AI reports look good, but break under pressure26:25

What are the ethical and practical considerations when deploying AI for content creation and technical optimization, especially for search engine compliance and user trust?

AI-generated code can be unreliable and introduce security risks like injected malware. Energy use is significant—AI analysis burned substantial resources with poor results. Use AI for data visualization and crunching, but always verify code and compliance manually. Experts should remain primary decision-makers, not AI, because guardrails still fail and costs (energy, errors) are high.

Judith Lewis: AI reports look good, but break under pressure30:28

In Judith Lewis's words

“You're smarter than an AI. Yes, I said it. You are smarter than an AI.”
Judith Lewis: AI reports look good, but break under pressure0:00

“It did it really, really, really fast. I mean, if I didn't know better, I would be like, 'Oh my god, I'm out of a job.'”
Judith Lewis: AI reports look good, but break under pressure3:04

“We can trust SE Ranking. We cannot trust Chat GPT to give us the truth because it is not a tool. It is an LLM.”
Judith Lewis: AI reports look good, but break under pressure14:13

“No matter what your imposter syndrome says, you are better than an AI, believe it or not, because we need to use the right tools in the right way.”
Judith Lewis: AI reports look good, but break under pressure20:16

“You are the most intelligent tool in the toolbox. You're the sharpest tool in the toolbox. Don't let AI make you feel less.”
Judith Lewis: AI reports look good, but break under pressure25:25

Schedule a demo