Meta traffic
Known delivery, proxy clicks, bounded budget.
🐴 Lovasember · review-gated LLL
Verification and Constraints
● READY FOR REVIEWBuild a measured Meta lead system, preserve the working baseline, and improve campaigns only on evidence of real enquiries.
Next executor: Sol High, only after later “Launch”
Goal and finish line
End goal: Lovasember receives more suitable enquiries from a reliable, measurable Meta path within its agreed spend.
End state: Paid self-knowledge booking works, completed outcomes can be reconciled, and a bounded Meta setup has a control, one interpretable test and a decision rhythm. The separate free leader offer waits for its own ready path.
Review gate
This is a plan. No executor, campaign, website or tracking change has been launched for this run.
Matt reviews this board and comments. A later explicit “Launch” starts one Sol High executor.
0 / 11execution checks passed
5deliverables
4,000 Ftcurrent configured daily budget
End goal system
Coherent offer
5,000 Ft self-knowledge booking path.
Real outcome
Confirmed, cancelled and qualified stages.
Better allocation
Spend follows verified lead quality.
Route to the end state
Read exact baseline
Matched windows, ads, copy pack, targeting.
Repair and prove
CTA to booking to private receipt.
Choose one test
Creative first, audience next, within cap.
Read business result
Keep, revert or refine from real leads.
What the evidence can say now
CONFIRMED
Meta account and lead campaign were active at 4,000 Ft/day on 26 Sep. Two active ad sets and 16 active ads. Both optimize the CTA-click proxy. Pre-interruption v9 ads carried most historical proxy actions. Post-resumption Meta shifted delivery toward v1.
PROPOSED
Keep the budget ceiling and paid baseline. Use the Sanyi-and-horse v9 concept as a traffic-proven control after claim review; compare a distinct factual photograph. Test women 35–55 in Hungary later as a hypothesis.
OPEN
Actual bookings and qualified lead counts. Which of five text assets within each active ad worked. Whether the current 180-day “conversion” exclusion event, Lovas_coaching_lead, reflects real clients or an earlier proxy.
Audience read: Before the payment restriction, the 10% lookalike produced more CTA-type actions (123 versus 58), while the interest set had about 15% lower cost per such action. Since resumption, spend has shifted toward the interest set. Neither audience has a verified booked-lead or quality winner. Individual interests cannot be ranked from one combined ad set.
Creative evidence, without a false winner
Sanyi and horse. The two high-volume v9 ads used the same image, so this is one visual reference. Inner-peace claim needs review before reuse.
Sunset and swans with a weekend-retreat message. Meta concentrated spend here after resumption, but no received-lead win is verified.
All 16 currently active ads share the same five primary texts and five headlines in Meta's asset feed. Their individual copy combinations have no separate outcome rows. Underdelivered images and individual interests cannot honestly be labelled losers yet.
Proposed settings for the first move
| Decision | Proposed setting | Proof or condition |
|---|---|---|
| Serving baseline | Existing Leads campaign 120234472897790312 with its two active ad sets and 4,000 Ft/day campaign budget. Keep the sales campaign paused. Record seven full stable measurement days after path repair. | Read exact IDs, budgets, statuses and dated Activity History before any write. |
| Offer and destination | Paid self-knowledge consultation, 5,000 Ft, current Google booking route. Correct dead/contradictory CTAs. No leader offer on this path. | Rendered mobile/desktop journey and marked private booking receipt. |
| Optimization event | Keep Pixel 755189613954638 and current appointment_cta_click optimization, labelled proxy, until a real completion transport, volume and Meta eligibility are proved. Preserve current placements. | Calendar booking ID/status is the private receipt. GA4/Meta completion and attribution stay PENDING if the source join cannot be proved. Read exact attribution/bid fields before changes. |
| First image test | After seven full stable days, prefer a native randomized Meta A/B: v9 Sanyi/horse factual photo versus one distinct factual photo. Same one body, title, CTA, interest-set audience, destination, event, placements and bid. Two cells at 2,000 Ft/day each for 14 full days. | Prepare paused. Temporarily pause the original 4,000 Ft/day campaign only during the verified experiment, preserve rollback, and confirm active configured budgets remain 4,000 total. If native setup or lead volume is insufficient, keep baseline and mark test INCONCLUSIVE. |
| Copy test | After the image result, test one primary text or headline change separately, with one text per ad. No current five-body/five-title winner can be named. | Current text asset feed has no separate outcome rows. Keep all other variables equal. |
| Audience test | Keep current HU/all-gender interest-set comparator 120244897438340312. Later test HU women 35–55 on the same offer/creative. Free leader route stays off until independently ready. | Check whether age/gender are strict controls or expandable suggestions and inspect actual delivery. Real lead quality, not proxy clicks, decides. |
| Exclusions | Audit 180-day audience 120234799674220312 before copying or changing it. Its current rule is Lovas_coaching_lead, not the CTA-click event. Any remove/replace decision is separate. | Trace the rule to actual business receipts. Do not exclude underdelivered interests or demographics. |
| Decision pace | Day 7 tracking and spend check, day 14 test decision, day 28 subsequent business review. No automatic test extension or budget increase. | Matched provider and business outcomes, or honest UNKNOWN. Future dates remain pending with an owner. |
Execution proof checklist
0 of 11 verified · executor checks only after proofGround creative and audience choices
Decide what has evidence, what is only a proxy, and what cannot yet be ranked.
PENDING 0 / 3
Criteria, constraints and proof
Observable tests
Protect
- Never mix full-lifetime and short-window results as one test.
- Keep raw account exports and customer records off this public board.
Proof: Meta API readback, matched-window calculation and creative mapping; dated private receipt.
GUI screenshot and blind Luna High description: pending execution
Repair the paid enquiry route
Keep the paid self-knowledge offer coherent from ad click to confirmed booking.
PENDING 0 / 2
Criteria, constraints and proof
Observable tests
Protect
- Preserve the current paid self-knowledge booking process and unrelated page content.
- Keep the separate first free leader consultation off this paid path.
Proof: Rendered post-deployment screenshots at both widths, blind Luna High descriptions, URL and private receipt readback.
GUI screenshot and blind Luna High description: pending execution
Measure completed enquiries reliably
Separate clicks from completed and qualified outcomes in Meta, analytics and the business record.
PENDING 0 / 2
Criteria, constraints and proof
Observable tests
Protect
- Keep CTA clicks diagnostic and never rename them real leads.
- Do not invent a Calendar-site GTM completion callback or expose lead identities.
Proof: Controlled session, Calendar booking ID/status, event details, dedup check and business receipt; post-test screenshots with blind Luna High descriptions.
GUI screenshot and blind Luna High description: pending execution
Configure a controlled Meta test
Keep current spending bounded and make the first image comparison interpretable.
PENDING 0 / 2
Criteria, constraints and proof
Observable tests
Protect
- Preserve baseline objects and rollback; only the planned experiment may temporarily pause it.
- Do not reactivate paused sales campaigns or change multiple test variables together.
Proof: Frozen object manifest, API and UI before/after readback, rollback values, post-change screenshot and blind Luna High description.
GUI screenshot and blind Luna High description: pending execution
Review real lead outcomes
Make the next scale or revert decision from received enquiries rather than platform clicks.
PENDING 0 / 2
Criteria, constraints and proof
Observable tests
Protect
- Never promise a lead uplift or treat a marked test as a customer.
- Do not send a client email without Matt’s explicit send authority.
Proof: Dated business reconciliation, provider readbacks and final board. Longer 14/28-day checks remain scheduled or pending, not fabricated.
GUI screenshot and blind Luna High description: pending execution
Open decisions and authority
To resolve from evidence
- Actual received/booked/qualified lead history, Calendar owner and receiving inbox access.
- Precise ad-set age, audience expansion and placement fields, plus whether the 180-day exclusion represents customers or a proxy.
- Whether Calendar completion can carry consented attribution to GA4/Meta. Site GTM cannot be presumed to observe a Google-hosted booking.
- Asset rights and factual claims for any reused or new photo.
Guardrails
- No production execution before later explicit Launch.
- No higher daily configured budget, paid test booking or new client commitment without authority.
- No login, token, VPN or other new security layer on hosted pages.
- Client email stays a draft; no send.
Public source limit: Google documents appointment schedules, paid appointments and booking forms and confirmations. These do not prove this account's attribution path. Meta deduplication and audience behavior still need authorized account readback; public documentation was rate-limited or login-gated during review.
Future-state email to SanyiHypothetical, unsent, edit against receipts
This is the message we would want to be able to send after execution. Past-tense claims must be checked and revised against what actually passed.
Tárgy: A Lovasember Meta kampányainak mérhetőbb alapjai
Szia Sanyi!
Átnéztük a Lovasember korábbi és friss Meta-eredményeit, és különválasztottuk a kattintásokat a tényleges foglalásoktól. Rögzítettük a kiinduló beállításokat és az elvégzett változtatásokat. A kreatívok összehasonlításához olyan tesztet készítettünk elő, amelyben egyszerre egy dolgot változtatunk, így a következő döntésünk ellenőrizhető adatokra épülhet.
Végigellenőriztük az önismereti jelentkezés útvonalát. A gombok a megfelelő helyre vezetnek, az 5 000 Ft-os konzultáció feltételei egyeznek az oldalon és a foglalásnál. A foglalási nyilvántartásban külön kezeljük a visszaigazolt és a később lemondott alkalmakat. Az ellenőrzött mérési kapcsolat segítségével a kattintásokat és a tényleges foglalásokat is külön tudjuk értékelni.
A napi beállított hirdetési keret 4 000 Ft maradt. A következő értékeléseknél a ténylegesen beérkezett és megfelelő érdeklődéseket nézzük, és ezek alapján döntünk a kereten belüli elosztásról vagy a következő tesztről. Ha még kevés az adat a biztos döntéshez, ezt egyértelműen jelezzük.
Üdv,
Mátyás
**Limited tracking variant:** replace the last sentence of paragraph 2 with: „A foglalásokat a naptár nyilvántartásából ellenőriztük. A hirdetésekhez való automatikus hozzárendelés még nem igazolt, ezért annak eredményét egyelőre nem állítjuk biztosnak.”
Parent ↔ executor updates
Executor not launchedPARENT · Review gate26 Sep 2026
The public board and Sanyi draft are ready for Matt's review. After comments are reconciled, only a later explicit Launch creates the Sol High executor. Its progress and evidence go here in blue.
Parent read setAfter launch, small files first
The future executor will maintain a deliverables list, main prompts and session ID. On completion, the parent reads those with a clean gist and checks the actual board and outputs.
Recovery state
No executor exists for this run. Current resume point: Matt reviews and comments on this board. The goal Markdown and private Meta snapshot are saved locally; the board remains the public contract.