What the engines say. What to fix.

How SparkBeacon measures a dealership

SparkBeacon starts from a prompt panel: the questions a car shopper in a specific market actually asks an assistant, from "best Mercedes dealer near Scottsdale" to "who will show me the out-the-door price on a GLC." Spark Noir Creative writes the panel for the store's market and brand, then SparkBeacon runs it against each engine on a schedule.

Every prompt is asked several times, because the same question gets different answers on different days. SparkBeacon reports the presence rate, the share of answers that name the store, with a confidence interval, and it reports it per engine. Spark Noir Creative never blends engines, because the same store can score two or three times differently between them, and never reports rank, because position inside an AI answer is noise while presence is stable.

Every run ends with a plan: what to change, on which pages, and how to verify it. The plan stays attached to the run, so the next run answers the only question that matters: did following the advice work?

The site audit

SparkBeacon crawls the store's site the way an AI crawler does and checks every page against the current standard. Each finding carries the pages it applies to, the change to make and the verification step. The checks include:

• Answer-first opening. The first paragraph names who, where and what, so an engine can quote it.
• Entity schema. Organization or AutoDealer markup that says what the business is, and one consistent name.
• Titles, descriptions, canonicals. Present, one per page, pointing at the page itself.
• Heading structure, thin pages, broken links, slow pages.
• Crawler access. Whether OpenAI, Anthropic, Perplexity and Google's AI crawlers are allowed in, and whether they came.
• Social previews and indexing directives that hide pages by accident.

The name, address and phone audit runs beside it: the store's NAP compared across the site footer, the schema, Google Business Profile and the directories the engines read, because one mismatched suite number is enough for an engine to split a dealership into two entities.

The crawler log

A one-line tag on the site reports every visit from an AI crawler to SparkBeacon: which engine, which page, when. It is the clearest early signal that an engine has started reading a store, and the clearest proof when one has not. Spark Noir Creative reads the log before it reads the rankings.

What a store sees

• Presence rate per engine, per prompt, over time, with the interval.
• The visibility ladder: retrieved, cited, mentioned, recommended, and where the store sits today.
• Competitors named in the same answers, and how often.
• Site audit with the prioritized fix list and the verification step for each.
• AI crawler log and AI-referred traffic.
• Progress report: what changed since the last run, and whether the plan was followed.