Check your company's AI visibility
Free, no credit card required.
Define the decision and audit boundaries
Ask which products, services, markets and customer groups matter. Record domains, brand aliases and the business objective. An initial diagnosis can use public data; traffic and implementation analysis require appropriate access. Explicitly identify what remains unchecked.
Build a question sample and retain answers
Include supplier discovery, comparisons, offering constraints and branded questions. Report these groups separately. Record the AI system, date, country, language and available settings. Repeat important questions at different times. One answer without a mention cannot diagnose an entire brand.
AI monitoring budget: choosing questions, systems and review frequency
Check the offering, competitors and sources
Separate brand presence from recommendations and page citations. Open cited URLs: do they substantiate the specific claim? Compare competitors within the same sample. Check whether existing pages clearly explain the offering, limitations and evidence. Google indexing and content accessibility are separate checks; neither guarantees AI recommendations.
AI citation audit: do sources really support information about your company?
Record evidence and dependencies
Describe each finding as observation → evidence → hypothesis → task → dependency → acceptance. Hypothetical example: AI omits furniture installation and the public offering does not explain it. Confirm the terms with the client and update the existing page. Do not claim the missing information definitely caused the omission.
Audit: evidence to scope of work. An observation does not automatically establish a cause. Audit decision framework. Evidence → hypothesis → action → acceptance.
Translate diagnosis into proposal scope
Separate baseline measurement, data corrections, technical work, content updates and subsequent reviews. Each item needs a deliverable, owner, required access and acceptance criterion. Price the work and agreed tool coverage, not a promised AI position. If the cause is unknown, propose diagnosis before a package of new articles.
Blog and knowledge base without cannibalisation: how to divide topics for SEO and AI
Verify accessibility and the existing URL map
For pages relevant to the sample, check the server response, readable main content, access for relevant crawlers and Google indexing signals. With Search Console access, use URL inspection; without it, state the diagnostic limitation. Check the canonical URL and duplicate main answers across the blog, knowledge base and service pages. A successful response code alone does not confirm indexing, and an access test does not prove a particular model used the page.
Separate urgent errors from growth hypotheses
An incorrect address, false price or service attributed to another company needs a separate verification path. Confirm the correct facts with the client, then identify the source and correction owner. Prioritise content hypotheses by relevance to audience decisions and available evidence. State what the agency and client will deliver and how the result is accepted. If the client has not confirmed service scope, include that dependency before pricing writing. The audit should support a work decision even if the cause of the brand omission remains unknown.
Final review and FAQ
Semly can support reviews of prompts, answers, competitors and sources; coverage depends on configuration and plan. Does a free report replace an audit? It provides a starting point, not a complete implementation diagnosis. Is new content always needed? No: first consider updating an existing URL. Separate confirmed findings, hypotheses and client-dependent work in the final proposal.
Sources and methodology
Share:
