WELLCODE · FACEBOOK & META ADS

WellCode Facebook Ads SOP

Operating procedure for planning, creating, reviewing, scaling, and recommending pauses for Meta ads

Research date: Oct 3, 2026 | Status: current local SOP

Editorial update: Oct 3, 2026 — campaign-specific acquisition, booking, nurture, and workshop plan.

Campaign plan at a glance

• ACQ — Free-course acquisition: bring new Romanian-speaking career changers into the free course. Ad → course registration → first lesson → follow-up. Judge on cost per qualified, activated learner and the cohort's later booked calls and customers.
• BOOK — Call booking: convert people ready to discuss mentorship into qualified sales calls. Ad → explanation of the paid program and who it fits → qualification → calendar. Judge on cost per qualified attended call and customer acquisition cost.
• NURTURE — Lead activation and warming: move existing registrants who are not ready to book toward course engagement and a sales conversation. Ad → relevant lesson, demonstration, or case study → next step. Judge on additional qualified bookings and sales relative to a comparison group, with course reactivation as an early signal.
• WORKSHOP — Optional event campaign: promote a specific workshop only when one is scheduled. Ad → workshop registration → attendance → call invitation. Judge on cost per qualified attended call and customers from the workshop cohort, not registrations alone.

This is the proposed campaign structure. Free-course acquisition is the documented existing funnel; separate BOOK and NURTURE campaigns are proposed roles, not a claim that those campaigns are currently live. Budget should follow evidence for each role rather than being divided equally by default.

ACQ — Free-course acquisition

• Who sees it: new prospects considering a career change, in Romania and serviceable expatriate markets. Exclude existing course registrants and customers where reliable audience matching permits; existing leads belong in BOOK or NURTURE.
• Offer and message: “Start the free programming course.” Show the first lesson, a practical beginner exercise, or how learning fits around work. Explain what the free course includes and the effort required. The destination is course registration, not a sales calendar.
• Measure: spend / qualified activated learners as the early quality metric; registration → activation, qualification, booking, attended call, and customer rates as cohort diagnostics. Define “qualified” and “activated” before the test. Use cost per customer after the normal sales lag to judge economics; cheap registrations alone do not justify scaling.
• Enable / keep / scale: launch when registration, first-lesson access, lead attribution, and follow-up work end to end. Keep a test running within its agreed loss limit while the cohort matures. Scale when qualified learner acquisition meets its agreed ceiling and mature downstream results support the acquisition cost.
• Disable / revise: pause immediately for broken registration, tracking, or handoff. Pause or revise the offer/audience when mature cohorts fail the agreed qualification or booking targets, or the test reaches its loss limit. If registrants fail to start the course, investigate onboarding before concluding that the ad is the cause.

BOOK — Qualified call booking

• Who sees it: engaged course participants, relevant site visitors, and other warm prospects who have not booked. Exclude existing bookings and customers where matching permits. Cold direct-to-call acquisition is a separate test cell; report it separately from warm booking results.
• Offer and message: “Discuss whether WellCode mentorship fits your goals.” Use a relevant student case study, mentor demonstration, or objection-handling ad. Explain that the call concerns a paid program and who it is suitable for. The destination includes qualification and available calendar slots; do not advertise this as another free-course signup.
• Measure: spend / qualified attended calls as the primary campaign metric. Also track cost per qualified booking, booking → attendance, attended call → customer, and customer acquisition cost. Define qualification consistently and deduplicate reschedules. A cheap booking that does not attend or qualify is not a win.
• Enable / keep / scale: launch only when qualification, booking attribution, confirmations, and sales capacity are ready. Scale when cost per qualified attended call is below the approved ceiling and mature close rates support the customer-acquisition target.
• Disable / revise: pause for a broken calendar, missing attribution, or no sales availability. Pause or revise when the loss limit is reached or mature results miss the qualified-attendance or customer-acquisition targets. Diagnose poor-fit bookings, no-shows, and weak closing separately before changing media.

NURTURE — Existing-lead activation and warming

• Who sees it: existing leads who have not booked, separated by the next missing step: registered but inactive, started but stalled, or engaged with unresolved objections. Exclude customers and people with a booking. Use time windows based on actual course progression and sales lag, not an invented universal number of days.
• Offer and message: inactive leads get a first-lesson reminder or short demonstration; stalled learners get help with a common obstacle; engaged leads get a relevant case study or explanation of mentorship. Send each ad to the specific lesson/content it promises. Once the main CTA becomes “book a call,” classify and measure that activity under BOOK.
• Measure: incremental qualified bookings or attended calls versus a comparable group receiving the usual email/sales follow-up without the extra paid nurture. Use reactivation and lesson progress as leading indicators. Video views, likes, frequency, and cheap clicks diagnose delivery; they do not establish business value.
• Enable / keep / scale: test only when there is a reachable existing-lead pool, a specific progression bottleneck, relevant content, and a way to track the next step. Coordinate with email and sales so paid nurture addresses a gap. Continue or scale when added downstream outcomes justify the extra spend within the agreed test budget.
• Disable / revise: pause if the intended audience or destination is wrong, or if the test-loss limit is reached. Stop or revise when a sufficiently observed comparison shows no useful lift, or repeated exposure brings no further progression. If the audience cannot support a meaningful paid test, prioritize existing follow-up and revisit later.

WORKSHOP — Optional, time-bounded

• Audience and destination: prospects suited to the workshop topic; send them to event registration, then attendance reminders and the call offer. Separate new-prospect acquisition from existing-lead promotion in reporting.
• Measure and decisions: enable only with a confirmed event, working registration/reminders, and sales follow-up capacity. Stop acquisition at the registration deadline or capacity limit. Evaluate attendance, qualified calls, and customers after the follow-up window before repeating; do not declare success from low-cost registrations alone.

Decision rules shared across campaigns

• Set ceilings before spending: agree on maximum acquisition cost, acceptable test loss, qualification rules, and a review window that accounts for sales lag. As a planning estimate, allowable cost per qualified attended call = target customer acquisition cost × mature qualified-attended-call-to-customer rate. For ACQ, allowable cost per qualified activated learner = target customer acquisition cost × mature qualified-activated-learner-to-customer rate. Allocate room for nurture and other paid touches; do not let every campaign spend the full funnel allowance independently. These are planning relationships, not supplied numerical targets.
• Avoid double-counting: acquisition, nurturing, and booking may all influence one customer. Report campaign-stage outcomes separately, then reconcile deduplicated customers and total funnel spend. Do not add attributed conversions from overlapping campaigns as though they were different customers.
• Re-enable: verify the original pause reason is resolved, document what changed, and approve a new bounded test with a loss limit and review point. The SOP agent recommends actions; a human operator changes campaigns and ads.
• Research: each weekly review checks official Meta sources and CT the Disrupter, and records which campaign role each finding affects. Practitioner advice remains a test proposal until supported by relevant evidence.

1. WellCode business and funnel context

• WellCode teaches Romanian-speaking people to code and become software developers. The priority audience is employed adults considering a career change, including Romanian expatriates; secondary segments are women on maternity leave and school students.
• The core paid funnel is Meta ad to free course, followed by self-booked or salesperson-booked sales calls and a high-ticket sale. Occasional live workshops use email promotion and a call-booking offer.
• Operational principle: optimize for qualified prospects and profitable customers, not lead volume or cost per lead alone.

2. Objectives and success measures

• Primary business outcomes: enrolled customer volume and customer acquisition cost within an approved payback and margin model.
• Diagnostic KPI ladder: spend -> lead -> reachable lead -> qualified lead -> booked call -> attended call -> sale -> collected revenue.
• Report both counts and rates at every transition. Break out Romania and expatriate markets where sample size permits, because language fit does not imply identical economics.
• Required inputs before numeric thresholds are approved: currency; average collected revenue and gross margin; refund/default rate; target payback; sales capacity; qualification definition; booking, show, and close rates; acceptable loss per test.
• Historical CPL 40 and cost per booking 800 from the prior 'Ads SOP' are unvalidated reference values only. Do not use them as universal stop/scale rules until their currency, calculation, and relationship to current unit economics are confirmed.

3. Audience and message strategy

• Default to career changers. Build Romanian-language messages around credible transition mechanics: learning while employed, time requirements, support, proof of student work, and realistic next steps.
• Include expatriates through eligible global geographies where WellCode can sell and support prospects; do not restrict the strategy to Romania by habit.
• Use broad or Advantage+ audience delivery as a controlled default candidate, with location, language, legal, and genuine serviceability constraints retained. Use manual interests or lookalikes as comparison cells only when the test has a clear hypothesis.
• Do not imply knowledge of personal attributes in copy. Avoid phrasing such as 'You are stuck in your job' or 'Are you unemployed?'. Frame the situation generally: 'For Romanian speakers considering a move into software development.'
• Keep secondary audiences in separate campaigns, messages, landing experiences, and reporting so their economics do not contaminate career-changer decisions.

4. Campaign and ad-set planning

• Use the opening campaign plan as the operating map: ACQ brings in new course users; BOOK asks ready prospects to schedule; NURTURE moves existing leads to their next step; WORKSHOP is optional and event-specific. Creative formats such as video or carousel are executions within these roles, not campaign types.
• Record each campaign's role, audience, exclusions, destination, optimization event, primary business KPI, budget, loss limit, review window, and owner before launch. These roles define distinct promises and measurement; they do not require separate campaigns for every small audience or creative variation. Launch only the roles that have the audience, budget, tracking, and follow-up to support a useful test.
• Choose the objective and performance goal closest to the deepest event that occurs often enough and can be measured reliably. Prefer qualified-lead or downstream optimization once CRM feedback is trustworthy; do not optimize indefinitely for cheap registrations.
• Use a consolidated structure with minimal overlapping ad sets. Split only for a meaningful constraint or hypothesis such as market group, funnel, offer, or audience category.
• Make Advantage+ leads with Advantage+ audience, placements, and campaign budget the default test candidate, not an automatic winner. Maintain a clean manual-control test when budget and volume support a fair comparison.
• Use Advantage+ placements unless a placement is legally, operationally, or creatively unsuitable. Diagnose placement quality with downstream CRM results, not click-through rate alone.
• Define before launch: hypothesis, primary KPI, guardrails, test budget, minimum evidence, decision owner, and what action each possible result triggers.

5. Creative and copy workflow

• Build diversity across concepts and motivations, not merely color or headline variants. Core concept families should include career-transition story, learning process, objection handling, student proof, instructor credibility, and concrete free-course value.
• Produce genuinely distinct formats: founder/instructor video, student story with permission, demonstration or screen recording, static proof, carousel, and native 9:16 video with audio and safe-zone checks.
• Match the promise from ad to landing page and free course. State time, effort, prerequisites, and limitations clearly; never guarantee employment, salary, or a fixed transformation timeline.
• Use AI tools for ideation and production only with human checks for truthfulness, Romanian fluency, brand fit, rights, artifacts, and unintended claims. Record when an asset is AI-created or materially edited and preview the resulting Meta transparency label.
• Keep a creative register with concept, hook, proof, format, audience, launch date, spend, qualified outcomes, and learning.

6. Launch quality checklist

• Correct business, page, Instagram account, pixel/dataset, domain, conversion location, objective, performance goal, budget, schedule, market, language, and exclusions.
• Pixel plus Conversions API coverage for website events; CRM Conversions API mapping for qualified lead and later stages where feasible; deduplication, timestamps, consent, and event match quality checked.
• UTM convention and immutable campaign/ad identifiers captured in CRM; test lead verified end to end from form through salesperson view and Meta event receipt.
• Landing page and form work on mobile, load acceptably in major markets, use Romanian copy, explain the free course, collect only needed data, and link privacy information.
• Creative renders correctly in Feed, Stories, and Reels; captions and key elements remain in safe zones; comments and destination links are checked.
• Sales follow-up capacity, ownership, scripts, working hours, and lead-routing alerts are ready before spend begins.

7. Measurement and attribution

• CRM is the operational source of truth for lead quality, booked calls, show rate, sales, revenue, and cohort outcomes. Ads Manager is a delivery and attribution view, not the sole revenue ledger.
• Send consistent funnel-stage events back to Meta, including stable identifiers and meaningful timestamps. Start with a clearly defined qualified-lead event; expand to booking, attended call, and sale only when stage definitions and data hygiene are reliable.
• Maintain a weekly reconciliation of Meta, analytics, form, CRM, and payment totals. Investigate missing IDs, duplicate events, time-zone differences, attribution windows, and late conversions.
• Meta changed website and in-store click-through attribution so only link clicks count as click-through; other social interactions move to engage-through attribution, and video engaged-view was shortened from 10 seconds to 5 seconds. Annotate trend breaks and never compare pre/post results as though the definitions were unchanged.
• Use A/B tests for isolated campaign decisions. Use Conversion Lift or another incrementality experiment when spend and eligibility justify it; use incrementality as a calibration layer rather than treating platform-attributed conversions as automatically causal.

8. Review cadence and testing windows

• Check delivery, disapprovals, broken tracking, and lead routing daily. Review quality and funnel outcomes at least weekly; review cohorts and unit economics monthly or after the sales lag has matured.
• Do not make a decision merely because 3–4 calendar days elapsed. Predefine evidence in business terms: enough spend relative to the approved acquisition target, enough relevant events, and enough time for normal lead-to-sale lag.
• Avoid unnecessary edits during learning. When a major edit is required, log it and restart the evaluation window.
• Treat small samples as uncertain. A result can be directionally promising without qualifying for scale.

9. Rules for recommending changes or scaling

• Scale only when tracking is healthy, lead quality is acceptable, downstream outcomes meet the approved economics, sales capacity can absorb volume, and results are stable across a representative window.
• Budget changes must be sized to the account's recent delivery stability and risk tolerance. The historical 20% step is a candidate operating test, not a universal law.
• Prefer scaling winning concepts into additional credible executions and placements before multiplying near-duplicate ad sets.
• Document the decision, baseline, expected effect, downside guardrail, review point, and rollback condition. Change one major variable at a time when causal learning matters.

10. Rules for recommending that ads be turned off

• Recommend an immediate pause for policy risk, broken destination or tracking, wrong audience or geography, deceptive claims, material brand error, or unserviceable lead flow.
• Recommend an economic pause when a preapproved loss limit is reached without sufficient downstream evidence, or when mature cohorts show unacceptable qualified-lead, booking, show, close, or acquisition economics.
• Do not pause solely because CPL exceeds a fixed number if downstream quality is stronger; do not keep an ad solely because CPL is low if contacts are invalid, unqualified, unreachable, or do not progress.
• For zero-booking cases, first verify routing, contact speed, CRM status discipline, sample size, and sales capacity. Separate an advertising problem from an operational follow-up problem.
• Record whether the action is pause, budget reduction, creative replacement, audience/funnel test, or tracking repair, and define the condition for reconsideration.

11. Free-course lead follow-up and sales-call handoff

• Route leads immediately to a named owner and track time to first attempt. Use an agreed multi-touch sequence across permitted channels and local time zones; never assume Romania-only availability.
• Give sales the ad, creative angle, landing version, source market, and form answers so the conversation matches the promise.
• Standardize CRM stages and reasons: invalid, unreachable, not eligible, not ready, qualified, booked, attended, no-show, won, lost, and disqualification reason.
• Marketing and sales review lead-quality patterns together. Feed durable qualified-stage signals back to Meta and turn recurring objections into creative or offer tests.

12. Live-workshop campaign considerations

• Treat workshops as a separate funnel with registration, attendance, offer exposure, booking, show, and sale stages.
• Coordinate paid retargeting with email timing and suppress or segment registrants where appropriate. Confirm event time zones for expatriates.
• Assess incremental bookings and sales, not registrations alone. Preserve enough follow-up time before declaring the campaign unsuccessful.

13. Policy and compliance checks

• Review Meta Advertising Standards before launch and after material policy updates. Pay particular attention to personal attributes, misleading or unrealistic claims, privacy, data use, and landing-page consistency.
• Career-change education is not automatically an employment ad, but creative that promotes a specific job, recruitment opportunity, or employment outcome may trigger employment-related requirements. Escalate ambiguous cases for policy review.
• Use consented first-party data and honor applicable privacy and deletion requirements for prospects across jurisdictions. Do not upload or infer sensitive data.
• For AI-generated or significantly AI-edited assets, anticipate Meta's 'AI info' transparency treatment, retain source/rights records, and require a human factual review.

14. Test backlog

• Priority 1: connect CRM qualified-lead outcomes to Meta and compare qualified-lead optimization with lead-volume optimization.
• Priority 2: Advantage+ leads versus a clean manual control, judged on cost per qualified lead, booked call, attended call, and sale.
• Priority 3: website/free-course flow versus higher-intent instant form, with matched follow-up and downstream measurement.
• Priority 4: distinct creative concepts and formats, including native 9:16 Reels assets, measured beyond CPL.
• Priority 5: Romania versus expatriate market groups, using serviceability and mature customer economics.
• Priority 6: follow-up speed and contact-sequence experiment, separated from media changes.
• Priority 7: incremental attribution or Conversion Lift pilot when eligibility and event volume are sufficient.

15. Sources and dated change log

• Official source: Meta, Advantage+ leads campaigns and opportunity score. Evidence: early multi-region tests included education and reported lower cost per qualified lead; confidence for WellCode: medium until replicated on its funnel.
• Official source: Meta Blueprint, Conversions API for CRM. Guidance: send first-party CRM outcomes so delivery can optimize for lead quality; confidence: high for process, performance impact must be tested.
• Official source: Meta Blueprint, creative diversification training. Guidance: vary concepts, motivations, visuals, messages, and formats; confidence: high as a workflow principle.
• Official source: Meta, simplifying ad measurement for a social-first world. Guidance: link-click attribution definition changed, other social interactions moved to engage-through, and the video threshold changed; confidence: high.
• Official source: Meta, 2026 AI-driven ads update. Evidence: continued ranking and incremental-attribution investment; confidence: high for platform direction, medium for expected WellCode lift.
• Official source: Meta, generative-AI ad transparency update. Guidance: Meta and detectable third-party AI edits can receive AI information labels; confidence: high.
• Official source: Meta for Business, Reels ads guidance. Guidance: use Advantage+ placements and native 9:16 video with audio and safe-zone checks; confidence: high for production practice, medium for WellCode performance.
• Change log — Oct 3, 2026: created the canonical SOP; replaced fixed-day and fixed-percentage rules with evidence and economics gates; added CRM outcome feedback, Advantage+ leads testing, creative diversification, attribution reconciliation, policy/AI review, lead handoff, and a prioritized test backlog. Classification: official guidance plus test recommendations. Overall confidence: high on direction; numeric thresholds remain unresolved.
• Change log — Oct 3, 2026: added a quick-reference summary of the documented funnel, recommended creative types, metrics, and launch/pause/restart decisions; made CT the Disrupter a required weekly research source at the user's request. Basis: sections 1–14 of this SOP and the user's workflow instruction; classification: operating summary and test recommendations, not a new platform finding. Confidence: high that the summary reflects the existing SOP; live campaign status, measured performance, numerical thresholds, and practitioner claims remain unverified. Relevance: keeps high-ticket career-changer decisions tied to qualified calls and sales. No new research review was performed.
• Change log — Oct 3, 2026: replaced the general opening summary with a proposed ACQ / BOOK / NURTURE / optional WORKSHOP campaign plan, each with audience, offer, destination, metrics, and enable/disable criteria. Added restart rules and shared-budget economics. Reason: the user requested concrete campaign roles rather than creative formats. Basis: the user's requested structure and existing SOP sections 1–14; classification: proposed operating plan and test recommendations. Confidence: high in alignment with the requested funnel roles; performance and account implementation remain unvalidated. No new platform research or live-account audit was performed.

Import provenance

Imported on 2026-10-03 from WellCode Facebook Ads SOP, last modified 2026-10-03T10:22:02.176Z. All 15 sections and seven source links were preserved. This import is not a new research review; source claims and confidence levels are inherited from the document.

The local Markdown file is now canonical. Google Docs is retained only as the historical import source.