Skip to main content

Pre-Work & Guardrails

Reports run asynchronously. A live session that tries to generate five reports from scratch will run long, run out of time, or both — do the reporting before the client is in the room.

This is the single biggest structural fix to how this engagement runs: the live session is where you interpret a baseline with your client, not where you build one from a blank slate. Everything on this page happens before that session starts.


Client Intake Questionnaire

Send this before scheduling the session — the answers determine the delivery track and shape what you prepare:

  1. Are you currently using ai12z? If yes, how long has your chatbot been live, and roughly how many conversations does it see per month?
  2. What's your primary domain, and do you have a Business Plus (or higher) plan if you're an existing customer? (Some reports require it — check current plan requirements before promising a report.)
  3. Who are your top 3–5 named competitors?
  4. What are the 5–10 questions you most want a prospective buyer to be able to get a good answer to, from an AI assistant or from your own site?
  5. Which pages matter most for this assessment — pricing, a flagship product page, a comparison page, your trust/security page?
  6. Has anyone on your team already looked at AI visibility, SEO, or GEO tooling? What did they find?
  7. Who owns follow-through after this session — who's the internal owner for the 30-day roadmap?

Required Pre-Work, by Track

StepProspect / New ClientExisting ai12z Customer
Confirm brand name, domain, and plan tier in Organization Settings
Build the manual question list with the client's intake answersUse conversation data instead, if available
Run Citation Monitor's initial baseline✅ (this is the Week 1 baseline — see 30-Day Roadmap)
Run Agent-Readiness, Site-Wide Sweep, AI Footprint
Run URL Analysis on 3–5 priority pages
Run Q&A Analysis, pull Q&A InsightsNot available
Pull current AI Referral Analytics readingNot available

Lead time: reports run as background jobs, not instantly. Submit them at least 1–2 business days before the session — Citation Monitor and AI Footprint in particular can take longer with a full competitor set. Build this into your scheduling, not into the session agenda.


Privacy Rules for Live Data

If the session shows real chatbot transcripts, visitor details, or referral events (Track B only):

  • Never show another client's real data, even anonymized, even as a "typical example" — use the fallback dataset below instead.
  • Redact PII before the session, not live on screen. Names, emails, phone numbers, and IP addresses shown in a Q&A Insights conversation view or an AI Referral Analytics event feed should be blurred or replaced in advance if you plan to screenshot or project them.
  • Ask before you demo, don't assume. Confirm with your point of contact whether it's acceptable to show real visitor-level data to whoever else is in the room — internal stakeholders, a board member sitting in, etc.
  • When in doubt, use the fallback example instead of a real one. It's not worth the risk to the relationship.

Fallback Demo Dataset

Use these when live client data isn't available yet (Track A) or when a live demo genuinely isn't the right call — they're real, previously-published examples referenced throughout this guide, not fabricated numbers:

  • Citation Monitor: a real run scored 32/100 ("Emerging"), cited only on its own branded query, with a competitor (Ada) holding 19% share of voice and winning all three uncited category queries
  • URL Analysis: a real trust-center page scored 52.4/100 — ranked reasonably in Google, but zero FAQ schema, no visible dates, vague claims like "industry-standard encryption" instead of named standards
  • Agent-Readiness: a real audit scored 68/100 ("Mostly Ready") — strong on robots.txt and sitemap, zero on llms.txt, with the report generating a complete, ready-to-publish starter file from the site's own real pages
  • AI Footprint: a real audit scored 48/100 with zero Wikipedia presence and zero confirmed review-platform listings
  • GEO Trend Comparison: a real two-period comparison showed GEO Score rising 76.0 → 78.0 and IDK rate falling 12.6% → 1.8%, while competitive intent simultaneously rose — a useful example of the loop catching a new problem, not just confirming a win

Reusing the same dataset every time also means you get better at narrating it — don't improvise a new fallback story per client.


Deliverable Template

Don't invent a new format per engagement — every deliverable in this guide shares one structure, the same Critical/High/Medium/Low tiering the real Consolidated Action Plan report already produces:

FieldDescription
ActionThe specific, concrete fix
SourceWhich report flagged it
ImpactCritical / High / Medium / Low
EffortLow / Medium / High
OwnerWho on the client's team (or agency) is responsible
TimelineWhich week of the 30-day roadmap it falls in
Success metricWhat score or number should move, and by when it gets re-checked

Add the Owner, Timeline, and Success Metric columns to whatever export or PDF you hand the client — the raw report output doesn't include them, and they're what turns a findings list into an actual accountable plan.


  • Delivery Tracks — Which reports are even available to prepare, by track
  • The Agenda — How this pre-work feeds into the live session's run of show
  • The 30-Day Roadmap — Where the deliverable template gets used week by week