Six stages. One bounded Sprint.
SurfaceGX compares your approved current facts with what AI and the retrievable web show, finds the owned-source and technical conflicts, produces five implementation-ready repairs, and documents one post-implementation recheck.
One change. One domain. Fixed fee. No subscription required.
What we guarantee, and what nobody can
We guarantee five implementation-ready fixes, dated evidence on both sides of the work, and one recheck of the same scope. We do not guarantee what a model will say next month. Neither can anyone selling you guaranteed AI rankings.
Define truth
Every Sprint starts with what is actually true now. SurfaceGX leads the intake: your approved, effective-dated old-versus-new facts, the priority pages, and the questions buyers actually ask. You name a fact approver, and the scope locks before observation begins.
What the scope holds
- One change on one company domain
- Up to 10 approved, effective-dated facts
- 10–15 priority pages, hard cap 15
- Up to 10 questions
What we need from you
- Authoritative sources for the new facts
- A named fact approver who signs off
- Approval of the facts, URLs, and questions before lock
Observe
With the scope locked, SurfaceGX records what three engines and the retrievable web currently show for your facts, pages, and questions. Every observation is dated, and the baseline is capped at 30 so the recheck can repeat it like for like.
What gets recorded
- Engine answers to the approved questions, dated
- What the retrievable web shows for each fact
- How your priority pages read to a machine
- No more than 30 observations, locked as the baseline
What it is not
- Not a blended score or a dashboard
- Not an open-ended crawl of your whole site
- Not a claim about every AI surface, only the three engines tested
Diagnose
Diagnosis compares the baseline against the approved facts and traces each conflict to something you control: a stale owned source, a page that cannot be read cleanly, a schema or configuration problem, or a missing answer. Conflict tracing is manual-assisted and founder-reviewed.
What a finding names
- The specific fact the observation conflicts with
- The page or owned source behind the conflict
- The technical cause where there is one
- The evidence, dated and linked
What we do not claim
- No claim that SurfaceGX controls an AI answer
- No blame on engine behavior we cannot evidence
- Findings without an owned-source path are labeled as such
Repair
The Sprint produces exactly five Fix Cards. Each states the problem and its evidence, the affected page or source, the exact change, an owner, priority and effort, acceptance criteria, and the validation method. The attachment is the draft work product, ready for review.
What a card carries
- Revised copy or a content brief
- Schema and JSON-LD corrections
- Code or configuration guidance
- Discovery-file updates where they apply
How it reaches your team
- Ticket or file handoff is the default and is included
- GitHub delivery only on a stack validated end to end
- Implementation beyond the handoff is custom work
Record implementation
Your team or your agency deploys the changes. SurfaceGX coordinates and records the state of each artifact: shipped, rejected, or blocked, with the deployment evidence you provide. The record is what makes the recheck meaningful.
What gets captured
- Deployment state per artifact
- Evidence: a ticket, PR, URL, or screenshot
- Decisions and blockers, dated
Who does what
- You deploy, reject, or block each artifact
- SurfaceGX is not the default implementation team
- The recheck starts only after deployment is reported
Recheck
One recheck reruns the locked scope: the same facts, pages, questions, and engines, up to the same 30 observations. The final proof report clearly states what improved, what remains unresolved, and the recommended next action, with links to the supporting evidence.
Four honest states
- Resolved
- Partially resolved
- Unresolved
- Not retested
What the report delivers
- What changed, backed by dated evidence
- What did not change, without overstating the result
- A clear recommended next action—even if the right action is “nothing yet”
Discuss an upcoming change
Tell us what is changing and when. If the Sprint is not the right fit for it, we will say so before you spend anything.