Impact

Numbers a GM can check on Monday.

Every figure on this page comes with where it came from. No relative grades, no leader set to 100, no screenshot presented as measurement. Here is what we have measured so far and how each client engagement is reported.

What the market looks like right now

In September 2026 we scanned 4,287 franchise dealer websites on the same fixed readiness checks and published every result on the Readiness Index. This is the baseline every PromptGraph client is measured against.

4,287

franchise rooftops scanned on identical checks.

Readiness Index, Sept 2026

40.7 of 100

average readiness score. Best store in the scan: 62. Scores are absolute, nobody is graded on a curve.

Readiness Index, Sept 2026

0.4%

of scorable stores had a live MCP endpoint. 96% had AutoDealer schema. Schema is table stakes; the agent layer is open.

2,699 scorable sites, Sept 2026

What we see across the PromptGraph network

21%

of discovery traffic across client rooftops is LLM driven.

PromptGraph network, 2026

48 hrs

from signup to live results in the dashboard. Structured data and the MCP server ship in the first week.

Standard onboarding

Who runs on PromptGraph

Multi rooftop groups and single stores across franchises, on Dealer.com, DealerOn, Dealer Inspire, and OEM mandated platforms.

Jake Sweeney Chevrolet Nick Mayer Auto Group Brady CDJR West Hills Kia Leif Johnson Ford of Austin Basil Mitsubishi Kruse Motors Volvo Parts Webstore

How every case study is reported

Client case studies are published here as each store passes its 60 and 90 day re-measurement. Each one follows the same format, so a competitor can rerun the test and get the same answer.

Section	What it contains
Setup	Market, competitor set, prompt set, engines, repetitions, and run dates.
Baseline	Visibility, share of voice, named first, citation share, and readiness score before any work.
What shipped	Endpoints, schema, llms.txt, and third party source fixes, with dates.
After	Same test, same prompts, at 60 and 90 days, with run to run variance reported.
Business result	GA4 LLM referral sessions, leads, and appointments.
Dealer quote	Approved by the store before publishing.

First case study in progress: Leif Johnson Ford of Austin, where a third party published a public July 2026 baseline of the store's AI share of voice. We will publish the same test, rerun 90 days later.

What we measure, and why both numbers

Score 1

Agent readiness

Can an AI agent read and act on this store? A fixed set of checks against the website: MCP endpoint, machine readable inventory, llms.txt, discovery file, AutoDealer schema, crawler access, sitemap. Deterministic. Run it twice, get the same answer.

Score 2

AI visibility

Does AI actually name this store? Non branded Find, Buy, and Service prompts across ChatGPT, Gemini, and Grok, with repetitions and a reported spread. Readiness is the cause, visibility is the effect. When the second number is low, the first one tells you why.