Objective
Build one truthful, measurable system that recruits, tests, ramps, and manages outbound cold callers.
Why it matters
Add reliable outbound capacity without turning Akash into the caller, recruiter, or daily traffic controller.
You need to
Confirm three defaults, then start Phase 0 — 5 min.
Project home
P-1364 · Build the GHL sales and CRM system · Cashflow
Action: reply “start Phase 0” to execute these defaults, or name the one default to change. This SOW itself does not contact candidates or prospects.
| Existing asset | Use | Required change |
|---|---|---|
| SDR recruitment system | Source → apply → voice screen → trial → ramp → active structure | Replace the old per-prospect $1-link approval queue with the approved scripted automation path. |
| Typeform questions + payload | Application intake and voice-sample gate | Add consent, availability overlap, recording acknowledgement, and rubric-ready field names. |
| SDR scorecard | Weekly operating metrics | Remove “link approved by Akash”; add suppression violations, qualified-start definition, and cohort quality. |
| Call flow | Truthful opener, video permission, callback, $1 start | Clarify the scripted send trigger and stop conditions. |
| Paid-trial offer sheet | Trial context, approved niches, boundaries, logging | Keep candidate sends human-owned; update the stale sales-message approval language. |
| Existing ramp plan | Day 1 through two-week review | Make call-review sampling and numeric gates explicit. |
A reviewer can trace every caller sentence to an approved script block or approved adaptive principle; all numeric claims have evidence; every disposition has one next action and one stop rule.
| Stage | Entry condition | Pass rule | Exit artifact |
|---|---|---|---|
| 1 · Sourced | Candidate came from OnlineJobs, referral, or an approved post. | Sent to one application form; no interview-first exceptions. | Source + source date |
| 2 · Applied | Required fields and voice sample received. | Complete submission; work window fits target hours. | Candidate record |
| 3 · Scorecard screen | Application is rubric-ready. | Green or one-clarifier yellow; no evidence invented. | Weighted score + notes |
| 4 · Async objection screen | Candidate advances from the application. | Natural opener, controlled “not interested,” correct handoff. | 90-second voice sample |
| 5 · Interview | Voice screen passes. | Understands role, volume, compensation, boundaries, feedback loop. | Interview scorecard |
| 6 · Paid trial | Akash sends the trial ask. | Completes role-play, 25 dials, follow-up, and logging. | Dial log + 2–3 recordings |
| 7 · Ramp | Trial meets the pass bar. | Two-week supervised trend clears quality gates. | Ramp scorecard |
| 8 · Active / Bench / Cut | Ramp review completed. | Decision grounded in the same scorecard. | Decision + reason |
| Criterion | Weight | A 5 looks like |
|---|---|---|
| Phone presence / spoken English | 3.0 | Clear, warm, concise, and controls the call without aggression. |
| Follows a script but adapts | 2.0 | Preserves the flow while responding naturally. |
| Volume tolerance | 2.0 | Evidence of consistent repetitive output and clean tracking. |
| Resilience / coachability | 1.5 | Applies feedback quickly; rejection does not change behavior. |
| CRM / dialer comfort | 1.5 | Has used a dialer and pipeline or demonstrates fast, accurate logging. |
| Target-hours availability | 1.0 | Reliably covers the approved US calling window. |
Score 0–5 per criterion. Missing proof is n/a, not a guessed score. The paid trial is the hiring signal; the application only decides who is worth paying to test.
| Task | Evidence | Pass bar |
|---|---|---|
| Learn the approved script | Recorded role-play | Truthful, natural, within boundaries |
| Complete ~25 live dials | Dialer log + recordings | All dials real; dispositions match recordings |
| Run three-touch follow-up | Tasks/messages staged or sent through the approved test lane | Correct timing, clean stop conditions |
| Update the CRM | Every dial has outcome, notes, next action, and recording | No missing connected-call records |
| Period | Operating target | Review | Gate |
|---|---|---|---|
| Day 1 | Access, offer map, script, dispositions, test records | Mock calls + logging check | Zero policy or logging errors |
| Days 2–3 | 15–20 supervised dials/day | Review every connect | Natural opener + correct stop rules |
| Days 4–5 | 30–40 dials/day | Review 5 calls/day | Clean CRM + improving permission rate |
| Week 2 | 50–75 dials/day, list permitting | Review 3 calls/day + exception calls | Outcome trend improves; no integrity violations |
| Day 10 review | Keep / extend / bench / cut | Full funnel cohort | Same published rubric for all candidates |
Recruiting candidates and contractor prospects must never share pipelines, tags, automations, or reporting.
| Prospect stage | Required event | Automation | Owner |
|---|---|---|---|
| Ready to dial | Fit + list + suppression checks passed | Create call task and caller assignment | Sales ops |
| Dialing | First dial attempted | Start attempt count and cadence | SDR |
| Connected | Human conversation occurred | Require recording, notes, objection, next action | SDR |
| Video permission | Prospect explicitly agreed to receive the video | Send approved deterministic text to the verified number | System |
| Video sent / watched | Delivery and view events | Create same-day callback or follow-up task | System + SDR |
| $1 started | Payment confirmed | Stop SDR cadence; start build/booking handoff | System |
| Launch booked | Calendar event confirmed | Start pre-call preparation | System |
| Live client | Launch approved and billing state confirmed | Trigger downstream-quality credit and bonus record | Sales ops |
| Nurture / DNC / Bad fit | Exact disposition | Suppress or schedule the allowed next action | System |
| Workflow | Trigger | Actions | Rollback |
|---|---|---|---|
| Candidate intake | Typeform submission | Create/update recruiting candidate, store source, flag missing proof, prepare internal scoring record | Deactivate workflow; delete test records by run ID |
| Candidate decision support | Candidate enters scorecard-ready stage | Generate internal score packet and draft next-step copy; never send | Deactivate; drafts remain internal |
| Caller assignment | Prospect enters Ready to dial | Check fit/suppression, assign SDR, create call task, stamp versions | Disable assignment; unassign batch by run ID |
| Video delivery | Recorded permission + verified phone | Send exact approved video text; log delivery and next callback | Disable workflow; mark affected contacts manual hold |
| $1 handoff | Payment event | Stop outbound cadence; tag start; create build/booking event | Replay from event log after correction |
| Scorecard rollup | Nightly or on demand | Aggregate calls, connects, video permission, watched, $1, live, errors by caller/cohort | Rebuild from raw immutable events |
| Access offboarding | Candidate/caller cut | Create access-removal checklist and verify completion | Reissue only through approved onboarding |
| Layer | Metrics | Use |
|---|---|---|
| List | Assigned, valid, wrong number, bad fit, DNC, reachable | Separate caller skill from data quality |
| Activity | Dials, connects, talk time, attempts per lead | Verify consistent execution |
| Belief path | Video permission, delivered, watched, callback completed | Diagnose script and asset quality |
| Cash path | $1 starts, launch bookings, live clients | Measure real business value |
| Quality | Refunds, disqualified starts, missed promises, suppression violations | Prevent volume from hiding harm |
| People | Per-SDR conversion, trend, coaching actions, trial/ramp decision | Keep/extend/bench/cut fairly |
This SOW defines operating safeguards; it is not legal advice. Final employment classification, compensation compliance, recording consent, and telemarketing requirements must match the applicable jurisdictions.
| Phase | Timing | Scope | Exit gate | Effort |
|---|---|---|---|---|
| 0 · Lock | Days 1–2 | Lead definition, script, claims, dispositions, comp, owners, success measures | Signed operating contract; no ambiguous field or promise | 4–6h |
| 1 · Recruitment build | Days 2–5 | Typeform, candidate pipeline, rubric, internal drafts, paid-trial packet | Synthetic applicant passes end to end without an external send | 10–14h |
| 2 · Sales pipeline build | Days 4–8 | GHL stages/fields, assignments, dispositions, tasks, message triggers, stop rules | Owned test contact moves through every stage once | 14–20h |
| 3 · Automation + scorecard | Days 7–10 | n8n workflows, event logs, scorecard, rollback, failure handling | Counts reconcile; retry creates no duplicates | 10–14h |
| 4 · Candidate trials | Days 8–12 | Source, screen, interview, paid trial for 2–3 candidates | Evidence-backed advance/bench/cut packet | 4–6h internal + trial pay |
| 5 · Ramp + handoff | Days 11–21 | Supervised dials, daily review, access controls, week-two decision | One operator can run the weekly loop without Akash traffic control | 6–8h internal |
Calendar dates start when Phase 0 inputs are locked. Candidate availability and reply speed can extend elapsed time without changing the fixed implementation scope.
| Role | Owns | Does not own |
|---|---|---|
| Akash | Defaults, candidate-facing sends/offers, final hire, irreversible live approvals, weekly performance decision | Daily applicant scoring, CRM traffic control, manual scripted sales sends |
| Builder / automation owner | GHL/n8n build, tests, versioning, rollback, documentation, data reconciliation | Changing offer economics or sending as Akash |
| Recruiting operator | Sourcing, complete-record checks, rubric scoring, shortlist packet, trial logistics | Unilateral offers or off-rubric decisions |
| Sales ops / SDR manager | List approval, assignments, call review, coaching, scorecard, comp calculation | Inventing claims or changing the approved offer |
| Outbound SDR | Dials, truthful script, permission, callback, clean logging, follow-up, $1 handoff | Closing $975/week, altering price/guarantee, fulfillment, unscripted texting |
| Item | Planning assumption | Excluded |
|---|---|---|
| Implementation labor | 48–68 hours using existing St. Pierre assets | Unrelated GHL cleanup and offer rebuild |
| Paid trials | $170–$255 for 2–3 candidates at $85 each | Candidate travel/equipment |
| Month-1 ramp | Up to $1,000 base per advancing caller + earned bonuses | Long-term payroll and benefits |
| Software | Existing GHL, n8n, Typeform, dialer/recording, email | New annual subscriptions unless explicitly approved |
| Lead data | One clean pilot batch sufficient for comparable trials and first 500 dials | Large-scale list acquisition |
No money is spent from this SOW. Trial payments, compensation, subscriptions, and paid sourcing remain separate explicit actions.
| # | Test | Pass condition |
|---|---|---|
| 1 | Submit one synthetic candidate application | One recruiting record, correct source, all fields, no candidate message sent |
| 2 | Score the same candidate twice | Same evidence produces the same rubric result; missing fields remain n/a |
| 3 | Advance a trial candidate | Internal draft and checklist created; Akash remains the sender |
| 4 | Load one owned test prospect | Fit, source, timezone, suppression, owner, and versions stamped |
| 5 | Run call/video permission path to 5477 | Exact approved text reaches only 5477; event and callback task recorded |
| 6 | Retry the permission event | No duplicate text, task, or stage event |
| 7 | Apply DND/STOP/manual hold | All future scripted sends suppress immediately |
| 8 | Post a $1 test event | SDR cadence stops; build/booking event starts once |
| 9 | Cut a test caller | Access-removal checklist completes and assignments stop |
| 10 | Reconcile the scorecard | Rollup equals raw event counts for the test cohort |
| 11 | Simulate workflow failure | Failure stays in in-app queue, retry is safe, rollback works |
| 12 | Cold-read handoff | A new operator can run recruiting, trial, ramp, and weekly review from the documents |
| Risk | Early signal | Mitigation |
|---|---|---|
| Poor list quality looks like caller failure | Wrong-number/bad-fit rates spike across callers | Version lead batches; compare callers only on comparable lists/windows |
| Commission-only model attracts weak or volatile talent | Strong voice applicants decline or churn quickly | Use paid trial + stable month-1 ramp base + quality bonuses |
| Callers copy deceptive course tactics | Prospects ask “is this a sales call?” or recordings contain invented claims | Claims register, published do-not-use list, immediate integrity cut condition |
| Akash becomes the manual bottleneck | Prospects wait for routine scripted sends or stage moves | Automate deterministic sales steps; keep only exceptions and candidate sends human-owned |
| Volume outruns fulfillment | $1 starts rise while launch/build backlog grows | Capacity threshold pauses new list assignment, not already-promised follow-up |
| Metrics can be gamed | Dials rise but recordings, notes, or downstream quality fall | Pay on qualified starts and live cohorts; audit recordings and raw events |
| Candidate and prospect data mix | Recruiting contacts receive sales automation or vice versa | Separate pipelines, tags, workflows, permissions, and tests |
Permanent project deliverable · 01-Projects/St_Pierre_Digital/SOWs/outbound-cold-caller-recruitment-sales-system-implementation-sow.html · based on the Michael Laurens course scrape, corrected to St. Pierre’s current GHL/n8n/brain operating rules.