Measurement methodology
Use each signal for the decision it can support. Technical readiness describes your page; sampled answers show what a platform returned; website analytics measures visits. None is a substitute for the others.
Research · write · review · publish
SEO and GEO audits inspect signals such as crawl access, content structure, metadata, and evidence. The result is a diagnostic score, not a prediction that ChatGPT or Google will recommend your brand. A high score does not guarantee indexing, citations, or traffic.
A prompt run records an answer from a particular platform or collection method at a particular time. Mention rate counts valid sampled answers naming the brand; citation rate counts valid sampled answers linking the owned domain. These describe the monitored sample, not every real user conversation.
A retrieved search result is not necessarily a cited source. Estimated AI search volume is modeled demand, not an exact count of people typing your tracked prompt. Google's keyword volume is also a different dataset. Use these inputs for prioritization without calling them actual AI impressions.
A verified crawler request shows page access, not a displayed citation or model training. An AI referrer can identify a visit, but some visits lose referrer data. Revenue and conversions require your analytics or CRM. Placing these datasets beside each other does not prove a particular crawl caused a sale.
Connect Search Console for owned-site search performance. Google also documents generative-AI impression reports for AI Overviews and AI Mode. Availability in Google's interface does not establish that Essel imports those dedicated reports; confirm supported integration fields before relying on that distinction.
AI answers change with model behavior, location, search results, and wording. A one-off test is a baseline, not a stable rank. Keep raw evidence, note configuration changes, and compare like-for-like samples. Never present an internal readiness target as a measured customer outcome.