In our 20 prompt test ChatGPT searched on 8 prompts and answered the other 12 from memory with no sources at all. Here is a six step check that separates a blank result from a real zero, and the agent job that keeps it running.
We ran 20 prompts across three AI surfaces and tracked where the sources came from. Five prompts found 31 percent of the source universe, ten found 54, and the twentieth still added 14 new domains. There was no plateau.
Same 20 prompts, three AI surfaces, one day. Gemini cited a median of 47 sources per answer, Perplexity 20, and ChatGPT cited nothing on 12 of 20 prompts while still naming vendors. Here are the two signals a visibility checker has to keep apart.
Search Console is free and covers 1,168 queries in a week, which is 23 times more than a 50 keyword watchlist. It is also three days late and blind to competitors. Here is which of the five ranking questions each route can actually answer.
We ran the same 50 keywords twice, at depth 10 and depth 100. The shallow check found our domain once. The deep check found it 22 times. Same keywords, same hour, same API. Here is the cost, the time, and the failure rate of each.
Google publishes no endpoint that returns your rank. We measured what its actual APIs do return: data that is three days old, an average position that differs from a live SERP result by 7.4 positions at the median, and an inspection verdict that says NEUTRAL for a URL Google has never seen.
We priced a SERP monitor from both ends: what the data costs per query, and how often a daily check produces something worth looking at. One hundred keywords daily costs 6 dollars a month in data and about 18 hours a month in review. That ratio should decide how you configure it.
We measured 76 skill files and split the cost in two: what a skill costs while it sits on the shelf, and what it costs when it runs. The median skill costs 107 tokens to keep and 1,851 to fire. That gap should decide which SEO skills you install.
We sent one prompt to the same model six times with web search on. Across the six answers, 18 distinct URLs were cited and not one of them appeared in all six runs. Here is what an AI visibility tracker has to record if it is going to mean anything.
Most ranking monitors alert on a three position drop. Across 106 queries in our own Search Console data, the median query swings 7.45 positions. Here is the calibration that turns a noisy alert into a usable one.