A clear review of your contacts, follow-ups, messages, and AI tools
Each area is scored from 1 to 10. A score of at least 7 out of 10 is recommended before allowing AI to speak or make decisions without your approval. Place your mouse over a bar to see why it received that score.
What this score means: this is not a criticism of you or the people who helped build the account. The system was created for the larger business you are building, but there have not been enough leads yet to make every problem obvious. The repair list is clear and manageable. We simply need to fix the foundation before adding more AI.
EC Assistant is set to Auto Pilot and deployed to SMS, Instagram (all pages),
Facebook (all pages), TikTok, WhatsApp and Live chat. Bot Training reads verbatim:
"No triggers configured. Add a trigger to route conversations to specific knowledge bases."
Its instructions are template defaults — "Your goal is to assist the customers with their queries" — with no product
information, no pricing, no scope boundary and no compliance language. Both
"Send bot to sleep when I send a: Manual Message" and "Workflow Message" are OFF,
so it may continue replying while you are personally speaking with someone or while your scheduled messages are being sent.
This is a wellness business making nervous-system and hydration claims, selling water systems and recruiting income partners. Usage is currently near zero (1 unique contact, 0 actions, 0 bookings MTD) — which is exactly why this is cheap to fix now and expensive after traffic scales.
EC Lead Sorter (Published) runs as step 1 of the master router on every intake submission.
| AI Sorter branch | Tag the agent writes | Tag that exists in CRM | Match |
|---|---|---|---|
| NSR Lead | int-ecreset | int-nsr | No |
| Kangen lead | int-kangen | int-kangen | Yes |
| Partner Lead | int-business | int-partnership | No |
| Unsure | — no tag, straight to End Node | — | Silent drop |
The agent's prompt compounds it, instructing the model that "Workflow values must be exactly: NSR Main Workflow / Kangen Intake / Audit Path / Elevated Partnership Workflow" — none of those three workflows exist in this account. As volume rises this forks the taxonomy into populations no filter reconciles.
Comment Trigger - NSR - RESET appears under Needs review (1). Execution logs, 13 Aug 2026:
| Time | Contact | Action | Status |
|---|---|---|---|
| 9:03:01 am | Reese Grace | fb_interactive_messenger | Error |
| 9:03:00 am | Reese Grace | Instagram DM | Skipped |
| 9:02:59 am | its_allgray | fb_interactive_messenger | Skipped |
This is a design fault, not just a connection fault: a Wait step sits between the first and second DM,
so by the time the follow-up fires Meta's 24-hour messaging window has closed. Compounding it, the Instagram
integration shows a red alert badge. Comment-triggered DM is the primary top-of-funnel motion for all three pillars.
Global Workflow Settings → Notifications: OFF. This is why P0-3 sat unresolved for four days. Every future automation failure will also fail silently — the single most dangerous default in an account about to run AI agents.
Partner — Non-booker Follow-Up Sequence (14 enrolments) has two triggers — tag added
int-partnership and a funnel page view on the partnership video page — with
Allow re-entry ON, Allow multiple opportunities ON and Stop on response OFF.
Every revisit to the video page re-runs the entire 3-email sequence and can create another opportunity.
Worse: the act-bookedcall check happens once, at the top. Someone who books after Day 1 still
receives Days 2 and 3 asking them to book.
The AI Sorter's "Unsure" branch routes straight to an End Node — no tag, no note, no notification. The master router's "None" branch does the same for anyone who submits the intake form without selecting an interest. In both cases a real lead disappears and nobody is told.
The brief states partnership "should NOT automatically be offered to everyone". In practice, anyone who submits
Partners Optin, or ticks the partner option on the intake form, is immediately tagged, given an opportunity, and
entered into partner nurture — with no qualification step anywhere. The tags
int-partnership-qualified and int-partnership-clarity-call exist but appear unused, and the
Pre-Call Qualification – Version B form receives submissions that nothing reads.
| ID | Finding | Label |
|---|---|---|
| P1-1 | Kangen leads split across two tags — the quiz path applies kangen-quiz-completed, the router applies int-kangen. Two disjoint populations that never reconcile. | CONFIRMED |
| P1-2 | {{contact.radio_15uga}} and {{contact.radio_17wad}} are hard-coded in the live AI Summary Agent prompt. Rename those fields and the prompt renders blank — with no error. | CONFIRMED |
| P1-3 | The hydration quiz is captured twice: Quiz | 5 Minute Hydration Quiz (15 radio fields) and Survey | Survey 0 (13 checkbox fields) ask the same 12 questions. Two fields are both labelled "Q7."; two ask "How much water do you drink daily?" | CONFIRMED |
| P1-4 | No lifecycle, stage or temperature fields — only additive stat-* tags that nothing removes. A contact can be stat-cold and stat-hot simultaneously. | CONFIRMED |
| P1-5 | Zero Custom Values. Every booking link, offer name and brand fact is hard-coded inside individual emails and workflows — and the AI has no canonical source to reference. | CONFIRMED |
| P1-6 | Not one workflow has "Stop on response" enabled. Nothing in this account stops talking when a human being replies. | CONFIRMED |
| P1-7 | 12 knowledge bases, each containing a single FAQ question, 0 documents grouped, 12 of 15 slots burned — and none attached to the live chatbot. | CONFIRMED |
| P1-8 | The AI Summary Agent's output has no destination; its router transition is still the unconfigured placeholder text "Describe the transition condition" → End Node. | CONFIRMED |
| P1-9 | Interest Selection is multi-select feeding a first-match If/Else. A lead interested in both NSR and Kangen is silently classified as NSR only — your most cross-sell-ready leads are the most mis-segmented. | CONFIRMED |
| P1-10 | Hands On Therapy is a whole business line — own domain, website, 6 workflows, 5 calendars, 3 forms — with no tag, no pipeline, and no place in either AI agent's prompt. | CONFIRMED |
| P1-11 | Allow multiple opportunities is ON on the master router, which creates opportunities in all three branches. | CONFIRMED |
| P1-12 | Pipelines encode entry branch, not progress. 11 of 11 opportunities sit in "Routed:" stages; 3 of 3 in "Routed - New Lead". Engaged, Booked, Attended, Buyer and Converted have never held an opportunity. All values CA$0.00. | CONFIRMED |
| P1-13 | You currently have two administrator accounts — "Jen ElevatedConnections" and "Jen HandsOnTherapy". This can split your notifications, calendar information, and record ownership between two places. | CONFIRMED |
| P1-14 | You do not have saved contact lists for NSR, Kangen, Partners, Hands On Therapy clients, or leads who still need review. This makes it difficult for you to quickly see each group. | CONFIRMED |
Survey 0 13, HOT Liability Waiver 2, Optin 1)Test 1, Asd, and a contact literally named testing (the most recent router enrolment)src-* tags as a manual substitute; "NA" values in the attribution reportGeminiBackEndAiProtocol private-integration API key — active since 21 Apr, Last Used: neverKangen Water – Video Follow-Up) has 2 stranded contactselevatedconnections.ca is connected with no products attachedkangen-quiz-completed, qa-zoom-registered-PARTNERSHIP spacing; "Day-of Appointment Reminder" also sends a day-before email; "HOT New Client Confirmation" vs "Returning Client Confirmation Email"Hydra chatbot exists, switched off, purpose undocumented; Auto Save ON globally with four admin editorsThese parts should be kept while the account is cleaned up. You already have a stronger starting point than many businesses.
int-, act-, src-, stat-, type-. Better than 90% of GHL accounts and a strong foundation for AI classification, once the AI is made to actually use it.WF-EntryIntake-Router — the busiest workflow at 69 enrolments — is a genuine router: one intake form, one branch per pillar, tag + opportunity + follow-up per branch. The pattern is correct.int-nsr on entry, then at the end adds int-nsr-completed and removes int-nsr. Deliberate lifecycle state management — rare, and the right instinct.This section shows what the account is doing today, step by step, even when that is different from what was originally planned.
TRAFFIC
├─ IG/FB comment "RESET" ──► Comment Trigger - NSR - RESET
│ → Respond on comment
│ → Instagram DM ⚠ SKIPPED
│ → FB Interactive Msgr ✖ ERROR (24hr window)
│ ► MANY LEADS DEAD-END HERE
│
├─ NSR Optin form ──► EC : 7-Day Journey - Lead Nurture & Access
│ → Tag int-nsr → Opportunity (Routed: EC Reset, CA$0)
│ → Course Grant → Email Day 1…Day 7 (no exit conditions)
│ → Tag int-nsr-completed → REMOVE int-nsr → END
│
└─ FORM-Entry Intake ──► WF-EntryIntake-Router
→ EC Lead Sorter AI → tag "int-ecreset" ✖ doesn't exist
→ EC Lead Summary AI → output goes nowhere ✖
→ Branch A → Tag → Opportunity → Course Grant → Add to Workflow → END
CONVERSION: ► NONE DEFINED. No NSR calendar. No paid offer wired.
"Buyer: EC Paid" stage = 0 opportunities, all time.
TRAFFIC
├─ IG/FB comment "HYDRATE" ──► Comment Trigger - KANGEN - HYDRATE (same 24hr design risk)
│
├─ 5 Minute Hydration Quiz ──► Kangen Nurture Sequence
│ → Tag kangen-quiz-completed ⚠ NOT int-kangen
│ → Opportunity → ~7 emails with waits → END
│
├─ FORM-Entry Intake ──► Router Branch B
│ → Tag int-kangen ⚠ DIFFERENT TAG from the quiz path
│ → Opportunity → Email → Wait → Condition → Email …
│
└─ 6 separate Kangen funnels, no canonical entry point ⚠
CONVERSION: ► NONE DEFINED. No Kangen calendar. No product catalogue.
"Buyer: Kangen Product" stage = 0 opportunities, all time.
TRAFFIC
├─ IG/FB comment "ALIGNED" ──► Comment Trigger -PARTNERSHIP - ALIGNED
├─ Partners Optin form ──► Partnership | Opt-in Entry → Tag → Opportunity → Email → END
│ ✖ NO QUALIFICATION STEP
├─ FORM-Entry Intake ──► Router Branch C → Tag → Opportunity → Email → END
│ ✖ NO QUALIFICATION STEP
└─ Tag int-partnership ─┐
OR partnership video ├──► Partner — Non-booker Follow-Up Sequence
page VIEW ────────────┘ → Wait → Opportunity
→ Condition (NOT act-bookedcall) ← checked ONCE
→ Email Day 1 → Day 2 → Day 3 → END
⚠ Re-enters on every page revisit
QUALIFICATION: "Pre-Call Qualification – Version B" EXISTS and receives submissions.
Data lands in a generic "Additional Info" folder.
✖ No workflow reads it. int-partnership-qualified appears unused.
BOOKING: "Elevated Connection Partnership Fit…" round-robin calendar exists — ungated.
CONVERSION: "Converted" stage = 0, all time. All 3 opportunities in "Routed - New Lead".
TRAFFIC ──► handsontherapy.ca (2-page website)
──► HOT New Client Calendar (90 / 120 min) · HOT Returning Client Calendar (90 / 120 min)
→ HOT New Client Confirmation (7)
→ HOT New Client - 48hr Preparation (5)
→ HOT New Client Intake - Safe Exhale form (18 fields)
→ HOT Liability Waiver (form deleted; 2 orphan fields remain)
→ HOT New Client - 24hr Aftercare (0 — never used)
→ HOT Returning Client - 48hr Check-in (0 — never used)
→ HOT 60-Day No Booking Nurture (0 — never used)
→ Returning Client Confirmation Email (5)
✖ No tag. ✖ No pipeline. ✖ Not in either AI agent's prompt.
| Question | Answerable? | Why |
|---|---|---|
| Where did this lead come from? | Coarsely | Referral / Social / Direct only. No UTM campaign data. "NA" values present. 12 src-* tags require manual application. |
| Which pillar are they interested in? | No | Three competing mechanisms: workflow int-* tags, AI-applied int-ecreset/int-business, and kangen-quiz-completed. No single field. |
| What content did they consume? | No | Page views tracked in aggregate only. act-emailclicked is a binary flag with no reference to which email. |
| Did they book? | Partially | act-bookedcall exists, but the "Booked: Call / Q&A" stage has 0 opportunities — so it isn't being used. |
| Did they buy? | No | Both Buyer stages hold 0. Attribution shows CA$0.00 revenue closed, all time. |
| What did they buy? | No | No product object, no Kangen model field, Payments module unused. |
| What stage are they in? | No | 14 of 14 opportunities sit in a "Routed:" stage. Stage reflects entry classification, not progress. |
| What is their next expected action? | No | No next-action field, no task automation, no follow-up date. |
Right now, six of these eight questions cannot be answered clearly. This is the main reason I do not recommend giving AI more control yet. If AI only has part of the information, it may give you an incomplete or incorrect answer with too much confidence.
EC Assistant to Offkangen-quiz-completed into int-kangenPrimary Pillar, not tagsEC Assistant in Suggestive mode, grounded, scoped, sleep-on-reply ON, one channel onlyHydraExplicitly out of scope for 30 days: autonomous conversation handling, lead scoring, a reporting/BI agent, Voice AI, and content generation at scale — each is listed under "AI LATER" with its unblocking condition.
Jen, the questions marked ★ affect how I should organize your account, so please answer those first.
qa-zoom-registered and a draft "Q&A Campaign" exist — is there a live Q&A motion?WF-DMAbandon-AllPillars meant to do? Created and deleted the same day (12 Aug).stat-hot / warm / cold mean operationally — who sets it, and when?int-nsr on completion.EC Assistant is currently replying to people autonomously on Instagram, Facebook and web chat, with no knowledge base attached? I need to confirm this was intentional before changing it.GeminiBackEndAiProtocol? There is an active API key for it that has never been used.Hydra chatbot?elevatedconnections.ca is connected with no products attached — redirect, host, or release?