Network disclosure
All ten sites in this network are published by Bay Area AEO. They cover answer engine optimization in the Bay Area from ten angles and link to one another so readers can move between related topics.
ROWS
QUESTIONS
Real customer wording
COLUMNS
6
AI platforms
CELL
4
Readings per answer
LAYERS
MONTHLY
One layer per re-test
The shape of the index
Think of the index as a stack of identical grids. One grid holds one month. Its rows are the questions a business's customers ask, written the way they ask them. Its columns are ChatGPT, Gemini, Perplexity, Claude, Copilot and Google's AI Overviews. A cell is filled when a question has been put to a platform and the full answer saved.
Because every month uses the same rows and columns, the grids line up. A reader can trace one cell down through the stack and watch a single question on a single platform change over time, or read across one month to see how the six platforms answered the same customer.
What one cell holds
| Reading | Question it answers | Recorded for each answer |
|---|---|---|
| Mention rate | How often is the business named? | Named or not named, per question and platform |
| Description accuracy | Is it described correctly? | The sentence the assistant wrote, checked against the business's own facts |
| Competitors named | Who else appears? | Every other business in the answer, in order of appearance |
| Sources cited | What did the assistant rely on? | Each cited page, listing or publication |
The readings stay separate. A business can be named in an answer that gets its hours wrong, and a competitor can appear in an answer that cites the business's own FAQ page. Collapsing those facts into one score would hide exactly the detail an owner needs to act on.
Question families in the index
The documented panels on this site fall into three families. City panels come from the sample questions on each Bay Area location page. Niche panels come from the forty industry pages, four questions each. Conversational prompts are the longer, multi-part questions on the four industry hub pages.
| Family | Source pages | Questions | Typical wording |
|---|---|---|---|
| City panels | 48 location pages | 145 | “Ethiopian restaurant in Temescal” |
| Niche panels | 40 niche pages | 160 | “Emergency plumber in San Jose open on Sunday” |
| Conversational prompts | 4 industry pages | 16 | “My water heater is leaking. Is there a plumber in Hayward who can come today?” |
Reading one row
Take the San Jose question “Emergency plumber in Almaden Valley.” Its row has six cells. In each one the observatory records whether the plumbing company under observation was named, the exact words used to describe it, every other plumber the assistant listed, and the pages or listings it cited. If Perplexity cites a Yelp listing while Gemini leans on the Google Business Profile, the row shows it at a glance.
The row also shows what the customer asked for: emergency service, a named neighborhood, a plumber. Those three requirements tell the reader which signals the answer had to satisfy, and the readings show which ones the assistant found.
Adding a month
Each monthly re-test adds a new layer with the same rows and columns. Nothing is overwritten. That rule is what lets Bay Area AEO show a client the real answers before and after a change, prompt by prompt and platform by platform, in place of a single number that moves for reasons nobody can see.
The index also guides the work between layers. A cell where the business is named but described with stale hours points to listing cleanup. A cell where a competitor appears with a cited guide points to a trusted-mention gap. The index turns each observed answer into a specific next step.
Where the readings are defined
The indicator files on this site define each reading in detail. The reference property in this network keeps the canonical definitions of the underlying terms, including AEO measurement standards and AI visibility measurement.
RelatedMention rate · Description accuracy · Panel methodology · City prompt panels