LEG 01 · FIND
GEO
GENERATIVE ENGINE OPTIMIZATION
Can an AI system find and cite you when someone asks for the best in your category?
SCORED /100 · 8 CATEGORIES
ACTOR: THE MODEL, AT ANSWER TIME
THE MEASUREMENT SYSTEM · FIND → CONVERT → TRANSACT
Three instruments,
scored one at a time.
Most agencies sell one number and call it visibility. We measure three independent systems — and we never blend them — because a business can win the AI citation and still lose the job ninety seconds later.
GEO gets you named in the answer. Agent Readiness is what happens when the user says “ok, book it” — and an autonomous agent tries to act on your site. Different actor, different failure, different score.
I THE THREE PILLARS
One voyage, three systems
that can each fail on their own.
A prospect's path runs find → convert → transact. Each leg is driven by a different actor, measured on its own rubric, and reported as its own score out of 100 across eight scored categories. We diagnose which leg is leaking before anyone spends a dollar on “more visibility.”
LEG 02 · CONVERT
CRO
CONVERSION RATE OPTIMIZATION
Once a human lands, does the page earn the call, the form, or the booking?
SCORED /100 · 8 CATEGORIES
ACTOR: A PERSON
LEG 03 · TRANSACT
AR
AGENT READINESS
Can an autonomous agent act on a person's behalf — book, quote, or buy — in real time?
SCORED /100 · 8 CATEGORIES
ACTOR: AN AGENT, IN REAL TIME
WIN THE CITATION (GEO)→ EARN THE HUMAN (CRO)→ LET THE AGENT FINISH THE JOB (AR)
II THE AGENT READINESS RUBRIC
The band you land in tells you
what an agent can actually do.
AR is scored 0–100 and mapped to five client-facing bands. Most sites we look at score well below where their owners assume — a polished site can still read to an agent as a brochure, not a business.
AGENT READINESS · ENGRAVED SCALE 0–100
CLIENT-FACING BANDS
85–100
AGENT-TRANSACTABLEAn agent can complete the task end to end — book, quote, or buy.
70–84
AGENT-LEGIBLEStructure and actions are readable; a handoff mostly succeeds.
50–69
AGENT-VISIBLE“Agents see a brochure, not a business” — content parses, actions don't.
30–49
AGENT-BLOCKEDAccess or interaction barriers stop the agent before it can act.
SAMPLE LANDS HERE · AR 34
0–29
AGENT-INVISIBLEThe site is effectively absent to an autonomous agent.
III WHOSE STANDARD THIS IS
Not our opinion. Google's published list,
scored with Google's own tool.
The Agent Readiness rubric is built on guidance Google published for making websites work with AI agents — and on a browser category Google shipped to measure it. We didn't invent the criteria. We assembled them into a repeatable score.
THE WRITTEN STANDARD
Google's own guidance
Google's web.dev guidance, “Build agent-friendly websites,” lays out what an AI agent needs to read and act on a page. Our eight AR categories map directly to it.1
SOURCE: GOOGLE WEB.DEV — LAST UPDATED 2026-04-01
THE MEASURING TOOL
Lighthouse, run methodically
Chrome's Lighthouse added an Agentic Browsing category — a checklist you can run against any URL. We run it methodically and record the result on every scorecard.2
SOURCE: CHROME LIGHTHOUSE AGENTIC BROWSING — ANNOUNCED 2026-06-22, CHROME 150+
Three scores from a rubric anyone can check — not a black box, and not a hunch.
HONEST CAVEAT: the Lighthouse Agentic Browsing category is described by Google as informational and unbenchmarked. We treat AR scores as a diagnostic and a trend line, not an industry-certified grade. As our own re-scoring dataset grows, we report movement over time rather than a settled benchmark.2
IV TWO MARKET WINGS
The same three-score system,
aimed at two very different jobs.
When an agent acts for a local customer, it is booking an appointment. When it acts for a manufacturer's buyer, it is sourcing a part. The rubric holds; what “transact” means does not.
WING A · LOCAL SERVICE
Eight verticals. The agent is booking.
The agent is completing an appointment on a person's behalf — or moving on to a competitor that lets it.
HOME SERVICES — HVAC CALLING WAVE
PLUMBING / ELECTRICAL CALLING WAVE
ROOFING CALLING WAVE
MED SPA & SALON CALLING WAVE
VETERINARY CALLING WAVE
DENTAL
PILATES / FITNESS / WELLNESS
AUTO REPAIR
LEGAL
Why the calling wave matters: at Google I/O (2026-05-19) Google described agents that book and call businesses on a user's behalf, starting with home repair, beauty, and pet care in a U.S. rollout through summer 2026.3 Businesses an agent can't complete a booking with get skipped — quietly.
WING B · B2B MANUFACTURING
Structurally different. The agent is sourcing a part.
The agent isn't booking — it's building an RFQ from your published capabilities.
- RFQ PATHSA machine-followable route from spec to quote request, not a “contact us” dead end.
- SPEC SHEETSTolerances, capabilities, and materials published as parseable data — not locked inside a PDF image.
- CERTIFICATIONSISO, AS9100, and ITAR status stated where an agent can read and filter on them.
- LEAD & MOQLead times and minimum order quantities exposed as structured fields.
- API / MCPPublic API and MCP-server expectations, so an agent can query capability directly — future-facing as these standards mature.4
ISO 9001
AS9100
ITAR
V PRODUCTIZED INTELLIGENCE
Report Factory — the same rigor,
packaged as fixed-scope reports.
Five productized reports for buyers who want the analysis without a full engagement. Each is built from the same scoring discipline behind the three-pillar scorecard.
R·01
Agency Gap Report
Where an incumbent agency's work leaves visibility and agent-readiness on the table.
R·02
Diligence Market Scan
A category read for investors and acquirers assessing a target's AI-era demand position.
R·03
SaaS Traffic Dossier
Where a SaaS brand is found, cited, and skipped across AI answer engines.
R·04
LocalGrade Scorecard
The three-pillar scorecard, packaged for a single local business and its market.
R·05
Franchise Territory Report
Location-by-location scoring across a franchise footprint, ranked by readiness.
VI ENGAGEMENTS
From a single scorecard
to a full four-phase build.
S·01 · FLAGSHIP AUDIT
Marketing + GEO Scorecard
The three-score deliverable: GEO, CRO, and AR, each scored across eight categories, with the leaking legs called out and prioritized. Sample numbers throughout are labeled SAMPLE.
S·02 · ENGAGEMENT
AI Visibility & GEO Proposal
A four-phase build that turns the scorecard into work:
- P1Foundation — incl. AR-1 agent-access unblock, week one
- P2Content & Structure
- P3Conversion — CRO + agent UX in one dev pass
- P4Authority & Monitoring — quarterly re-scores, call testing
S·03 · SPECIALIST
State Ad-Compliance Audit
A legal-advertising compliance review for regulated verticals, matched to your state’s rules — so visibility work never outruns the rules that govern how a firm may advertise.
Pricing model: scoped up front, billed phase by phase. Stop at any phase — every phase ends in a verifiable deliverable. No per-lead, per-booking, or contingency pricing.
VII TAKE A READING
See where your three scores land.
Request an audit and we'll score GEO, CRO, and AR against the published standard — and show you which leg of find → convert → transact is costing you the job.
OPEN WATER · SERVING ALL 50 STATES
SOURCES & NOTES
- Google, "Build agent-friendly websites," web.dev — last updated April 1, 2026.
- Chrome Lighthouse "Agentic Browsing" category, announced June 22, 2026 (Chrome 150+). Described by Google as informational and unbenchmarked; no scoring weight at this time.
- Google I/O keynote, May 19, 2026: agentic booking and AI calling businesses on users' behalf in "home repair, beauty or pet care," U.S. rollout summer 2026.
- WebMCP is a proposal in origin trial (Chrome 149) — not yet a shipped standard. We build toward it; we don't claim it's here. All sample scores on this page (GEO 62 / CRO 71 / AR 34) are illustrative and labeled SAMPLE.