METHOD / ENTRY 15
How Bay Area AEO shows progress: the real answer an AI assistant gave before the work, set beside the answer it gives after, for the same question on the same platform.
Bay Area AEO makes the before-and-after comparison part of two of its four principles. Under "Measured in real answers," it says it shows the real answers before and after. Under "Plain English, always," it says it explains what it is doing, why it matters and what it is changing, with before-and-after answers. The about page puts it most directly: the agency doesn't report vanity rankings; it shows the real answers AI gives to a business's customers' questions.
A before answer and an after answer are comparable when only time has changed. The question is the same, word for word. The platform is the same. The business is the same. Bay Area AEO's monthly re-test of a fixed prompt list produces exactly these pairs: the free visibility check supplies the first answer for every question, and each month adds another. Any two runs of the same question on the same assistant make a fair pair.
| Look at | Question to ask |
|---|---|
| Mention | Is the business named in the after answer? |
| Position | Is it among the two or three names the assistant puts forward? |
| Description | Does the answer describe its services, hours and specialties correctly? |
| Reason given | Does the assistant say why it recommends the business, in words the business uses? |
| Sources | Which listings, reviews or outside pages does the assistant now cite? |
| Competitors | Which other businesses appear beside it, before and after? |
The signals the agency builds map onto the parts of an answer. Consistent listings let the assistant state hours and location without hedging. Specific reviews give it phrases to repeat; the agency's food and drink page notes that AI pulls phrases about dishes and atmosphere, and about the occasion, straight into its answers. An answer-ready website gives it a page to quote. Structured data tells it the services and service area. Trusted mentions in local press and guides give it a reason to name the business at all.
So an after answer can differ from a before answer in several ways at once. A business can move from absent to named. It can move from named with a thin description to named with a reason: "known for gentle care with anxious patients" in place of a bare listing. Its cited sources can shift from a stale directory to its own service page.
Bay Area AEO's home and about pages carry an interactive demo of this idea. A visitor picks a question, such as the San Mateo dentist question or "Coffee shop in Sunnyvale with good Wi-Fi where I can work for a few hours?", picks an assistant, and switches on the five AEO signals one at a time. The demo's meter moves from "Invisible" as each signal is added. The agency labels it a simulated example that shows how assistants weigh information; the client reports show the real answers.
A ranking report tells an owner where a page sits in a list of links. The customer asking ChatGPT never sees that list. The before-and-after answer shows the owner the very text the customer reads. That is why the issuer singled out reporting tied to observable AI answers in its selection criteria. See the methodology entry for the full measurement and the timeline entry for when the after answers usually begin to differ.