Chat Rules and Boundaries
What the UpSearch chat will and will not do.
The UpSearch chat assistant has specific rules about what it will recommend, what data it uses, and what it refuses to do. Understanding these boundaries makes conversations more productive.
Why the chat has rules
The UpSearch chat assistant operates under specific constraints. These are not limitations of the technology. They are deliberate design decisions that make the chat more useful and more trustworthy.
An AI assistant without boundaries will confidently answer any question, even when it does not have the data to support its answer. UpSearch's chat is designed to be honest about what it knows and what it does not.
What the chat will do
Answer questions about your site using available data. When you ask about your site's performance, technical issues, or content quality, the chat draws on GSC data, crawl results, and SERP analysis to provide evidence-based answers.
Explain findings and suggestions. When UpSearch surfaces a finding, you can ask the chat to explain the reasoning, the evidence, and the recommended action in more detail.
Help interpret data. If you are looking at GSC data or crawl results and are not sure what they mean, the chat can explain the context and implications.
Provide SEO guidance grounded in evidence. The chat can discuss SEO concepts, strategies, and best practices, drawing on the same principles documented in the SEO System.
Generate reports. The chat can produce structured analysis reports for specific aspects of your site: technical diagnostics, traffic decline analysis, content opportunities, and more.
What the chat will not do
Invent data. If the chat does not have data to support a claim, it will say so. It will not fabricate statistics, traffic numbers, or rankings to fill gaps in its knowledge.
Give generic advice. The chat will not provide advice that could apply to any site. Recommendations are specific to your site, based on your data. If it cannot be specific, it will explain why.
Make promises about results. SEO outcomes depend on many factors outside anyone's control. The chat will not predict specific ranking improvements, traffic increases, or timelines.
Recommend actions without evidence. Every recommendation the chat makes is tied to specific data. If the evidence is insufficient, the chat will flag the uncertainty rather than presenting a guess as a recommendation.
Answer questions outside its scope. The chat is an SEO assistant. It will not write your content, manage your social media, or advise on topics unrelated to search visibility.
Hard stops
In certain situations, the chat will refuse to produce an answer:
Missing required data. If a question requires GSC data and GSC is not connected, the chat will say so and stop. It will not speculate about what the data might show.
Insufficient evidence. If the available data is too limited to support a meaningful answer, the chat will explain what data is missing rather than providing a low-confidence answer.
Out of scope. Questions about topics the chat is not designed to address will receive a clear "this is outside my scope" response.
These hard stops are a feature. They mean that when the chat does provide an answer, you can trust that it is supported by evidence.
How to get the most from the chat
Be specific. "How is my site doing?" produces a less useful answer than "Which pages lost the most impressions in the last 28 days?"
Ask about your data. The chat is most useful when discussing your specific site data. Generic SEO questions get better answers from the SEO System pages.
Follow up on findings. When UpSearch surfaces a finding, ask the chat to explain it further. "Why did you flag this page?" or "What should I do about this cannibalization issue?" produces actionable guidance.
Ask for evidence. If a recommendation seems surprising, ask the chat to show the data behind it. The evidence is always available.
Practical takeaway
The chat is designed to be a reliable advisor, not a yes-machine. Its boundaries exist to protect you from bad advice. When it says "I do not have enough data to answer that," it is doing its job correctly.