MEMECMO.AIPartner Playbook v1.1
MemeCMO GEO Service Methodology · Channel Partner Delivery Playbook
Two-phase closed loop: P1 Assessment ⇄ P2 Implementation · One replicable business process for any channel partner serving any brand client
GEO service = one loop, run continuously: P1 measures where the brand stands in AI answers and what deviates (with periodic reports); P2 modifies the real-world corpus sources through five standard work packages; the next P1 scan verifies the effect. Assessment shows where the gap is, implementation closes it, re-assessment proves it — the loop itself is the service, and the renewal rationale. The method layer is identical for every client; the content layer is rebuilt per brand — that split is the business model.
0. Four Ground Principles (before any process)
| Principle | Meaning | Where it shows up |
|---|---|---|
| ① Weld the ruler shut before reading it | Prompt panel, competitor set, judging baseline and scan cadence must be locked first — only then are movements credible and fit for acceptance | P1-4 freeze is P1's gate; any pre-lock number is labeled “calibration period” |
| ② Govern facts before producing content | Brand facts (names / scale / NAP / tagline / forbidden items) need one authoritative version, classified PUBLIC-READY / INTERNAL-CONTROLLED / HOLD | P1-3 fact sheet is the client's only deep-involvement step; no signed sheet, no P2 |
| ③ AI fills vacuum with whatever it finds | AI targets no one. Where authoritative corpus is missing, it fills the gap with wrong hotlines, rivals’ inflated claims, hallucinated prices. Whoever supplies the corpus owns the answer | All five P2 work packages are different pipes feeding the same vacuum |
| ④ Effects have physical lead times — never oversell | Retrieval engines respond in 2–6 weeks; parametric engines (model training) in months; single-scan swings within ±3 are noise | Every report carries noise bands and evidence tags; attribution accepts only “beyond noise + same baseline + consecutive” |
Phase P1 — Assessment: show the client where they stand in AI answers
| Step | Action | Standard practice | Deliverable (template exists) | Owner | When |
|---|---|---|---|---|---|
| P1-1 | Day-0 baseline scan | Launched in the sales meeting itself: create project → full scan → first scorecard in minutes. The strongest acquisition move — the client sees “what AI says about me” for the first time | Day-0 scorecard + PDF kick-off report | Partner sales (self-serve) | Pitch meeting |
| P1-2 | Accounts & roles | Client sub-organization under the partner org; partner = editor (can run scans), client = read-only viewer; credits configured per package | Account ledger + role matrix | Partner ops | Contract week |
| P1-3 | Brand fact sheet (the client’s only deep-involvement step) | Collect and get signed: official names (all languages) / tagline / the one canonical scale term / NAP / differentiating facts / forbidden items (price figures, internal metrics); classify PUBLIC-READY / INTERNAL-CONTROLLED / HOLD; load into brand_profile as the single grounding for judging and generation | Brand alignment table + signed Fact Sheet | Partner account manager + client | Week 1 |
| P1-4 | Freeze the measurement | Core prompt set (20) finalized in the buyers’ real query language and permanently locked; competitor set manually curated and locked (no auto-rotation); judging baseline = the signed facts; scan cadence aligned to the client’s management meeting | Freeze confirmation sheet (first Scan Manifest) | MemeCMO + partner | Weeks 1–2 |
| P1-5 | Reporting cadence live | Three delivery layers: weekly report (automatic) → monthly review (narrative + attribution table) → quarterly / milestone walkthroughs. First issue comes with a 30-minute scorecard walkthrough | Weekly (auto) + monthly template + walkthrough agenda | System + partner | Week 2 on |
What P1 promises the client: baseline score and cross-language position, a deviation list (which questions AI answers wrongly / misses / hands to competitors), content-influence recommendations, and weekly position tracking from then on. P1 sells on its own (diagnosis + monitoring subscription) — and it is P2’s sales funnel: every line in the deviation list is a quotable P2 item.
Phase P2 — Implementation: five standard work packages (WP)
P2 in essence: take every vacuum P1 found, and fill it through five pipes. Each WP has a proven template — a new client only swaps the parameters.
| Package | Scope | Standard actions | Template asset (field-proven) | Executed by | Effect window |
|---|---|---|---|---|---|
| WP1 Website GEO rebuild | Technical baseline + content pool | Eight technical items: SSR real text / robots allowlist / Schema trio (Organization · Service · FAQPage) / hreflang / 301s preserving cited URLs / NAP consistency / no access walls / performance. Content pool: FAQ, pricing-logic page (zero figures + CTA), facts page, buyer’s guide | Website GEO Technical Requirements + Crawler & Acceptance Checklist + Schema pack | Client’s web agency builds; MemeCMO audits & releases | 2–6 weeks |
| WP2 Crawler & index switches | The four AI-ingestion switches | robots.txt allowing 8 UAs (GPTBot / OAI-SearchBot / ClaudeBot / PerplexityBot / Google-Extended / Googlebot / Bingbot / CCBot) → GSC verify + sitemap → Bing Webmaster one-click import + IndexNow → business profiles | 48-Hour Sprint Runbook §D2 | Partner assistant, checklist-driven | 1–4 weeks |
| WP3 External authority sources | Knowledge graph + third-party endorsement | Wikidata entry (10 fields, each sourced) → Wikipedia (compliant drafting, client publishes, COI disclosed) → industry directories → every external NAP identical to the website, word for word | Wikipedia / Wikidata Publishing Pack + entry drafts | Partner / client publish; MemeCMO drafts | 2–8 weeks |
| WP4 Platform dynamic content | Unified social presence + media supply | Content-card mechanism (names / tagline / bio / NAP as the single wording source — paste, never rewrite) unifying FB / LinkedIn / YouTube; media publishing at baseline density: 2–4 deep articles per month (question-shaped titles answering the buyer prompt panel); steady beats bursts | 48-Hour Sprint Runbook §D1 + Content Cards A–D + density table | Partner content team | 2–6 weeks |
| WP5 Corpus correction | Kill what AI is saying wrong | From P1 accuracy verdicts, extract three target classes: wrong facts (hotlines, old addresses), false competitor comparisons, hallucinations (invented prices) → corrective content per item (authoritative site page + media corroboration); verify elimination in the next scan | Deviation list → corrective-content mapping table | MemeCMO drafts, client approves, partner publishes | 2–6 weeks / item |
Key sentence: nothing in P2 is mystical optimization — every item places authoritative corpus into one specific pipe, and every item can be verified by name in the next P1 scan. Any one of these modifications tangibly moves the engines: retrieval first (citation, presence), parametric later (training cycles run in months) — which is also the physical argument for a continuous subscription.
Closing the loop: pre-register every P2 action against a P1 metric
When a WP starts, register a row in the attribution table at the same time — write the expected metric first, then let the scan settle it. This is the standard language a partner uses to prove value to their client (born in a real engagement; survived three rounds of the client’s methodology audits).
| Action (WP) | Problem targeted | Before | After | Met? | Replicable? |
|---|---|---|---|---|---|
| Six fixed columns. “Met?” accepts only beyond-noise (±3) + same baseline + scan-ID traceable. “Not met” rows are allowed — honesty anchors the table’s credibility. Expected mappings: WP1/WP2 → citation strength, presence; WP3 → new-sources column, knowledge-graph accuracy; WP4 → share of voice, sentiment; WP5 → accuracy, top-of-mind. | |||||
Roles (RACI summary)
| Workstream | MemeCMO | Channel partner | Brand client | Third party |
|---|---|---|---|---|
| Platform / scans / reports / judging | R+A | I | I (read-only) | — |
| Client relationship / sales / coordination | C (solution support) | R+A | I | — |
| Fact sign-off / forbidden items | C (templates & audit) | R (collect) | A (sign) | — |
| WP1 website build | C (requirements + acceptance audit) | R (coordinate) | A | R (build) |
| WP2–WP4 execution | C (runbooks + verification) | R | A (authorize / approve) | as needed |
| WP5 corrective content | R (draft) | R (publish) | A (approve) | — |
The standard 90-day timeline
| Week | Milestone | Acceptance |
|---|---|---|
| W0 | Day-0 scan in the pitch meeting (P1-1) | First scorecard shown live |
| W1–W2 | Accounts, fact sheet signed, measurement frozen, weekly report live (P1-2…5); WP2 crawler switches via 48-hour sprint | Freeze confirmation sheet; first weekly report delivered |
| W2–W6 | WP1 website build + WP3 authority sources + WP4 platform unification and content supply | Website acceptance passed; Wikidata live; content cards consistent on all platforms |
| W4 / W8 / W12 | Three monthly reviews: attribution rows settled; WP5 corrections rolling | W12 review = 90-day acceptance report vs Day-0 (beyond-noise improvements only) |
90-day success criterion (partner-internal): reproduce the curve shape “citation leads, presence follows” — that is the methodology holding on a new client; then switch to the steady subscription loop. (Field reference: first full cycle — AIGVR 61→65, citation +82%, presence +7 pts, all beyond the noise band.)
The replicable asset library (swap parameters only)
| Template | Purpose | Used in |
|---|---|---|
| Brand alignment table + Fact Sheet sign-off | Fact governance and judging baseline | P1-3 |
| Freeze confirmation sheet (Scan Manifest) | Contract-grade measurement credibility | P1-4 |
| Weekly (auto) / monthly review template / walkthrough agenda | Three-layer periodic reporting | P1-5 |
| Website GEO Technical Requirements + Crawler & Acceptance Checklist | Web build and release audit | WP1 / WP2 |
| 48-Hour Sprint Runbook (content-card mechanism) | Network-wide unification and index switches | WP2 / WP4 |
| Wikipedia / Wikidata Publishing Pack | Knowledge-graph building (compliant: we draft, client publishes) | WP3 |
| Channel Operations Guide (frequency / density baseline) | Content supply cadence | WP4 |
| Attribution table (six columns) + deviation → correction map | Value proof and correction management | §3 / WP5 |
| Migration Protection Pack (301 / cited-URL inventory) | Preserving AI assets when a client changes site or vendor | WP1 variant |
New-client launch checklist (the salesperson’s pocket page)
- Day-0 scan run; scorecard shown to the client
- Commercial package agreed (P1 subscription / P1+P2 full); credits configured
- Client named a fact signer (empowered to rule on names, scale, forbidden items)
- Brand fact sheet signed; forbidden items (prices etc.) listed
- Core prompt language = the buyers’ real query language (not necessarily the client’s internal language)
- Competitor set confirmed by the client and locked (competitor / partner / self distinguished)
- Scan / report cadence aligned to the client’s management meeting
- Client’s web agency contact onboarded; technical requirements delivered
- Partner’s executing assistant holds the 48-Hour Sprint Runbook (local-language edition)
- Attribution table opened; every P2 action pre-registered against an expected metric before execution