Five actions for Gloria Marcott's account, ranked by money impact. Read-only audit: nothing in the account was changed.
The ads bring in applications at $11.20 each. Meta has never seen a booked call, because both calendars send bookers to pages that do not report the booking. In 14 days the campaign spent $650 and got 58 applications, up from 6 at $14.51 on 2 Oct. The ad set is told to find women who book, but 0 booked calls have reached it. Until a real booking lands on /success, the $50 a day is aimed at an event Meta never sees.
Ranked by money. At Gloria's own numbers (about 80% of booked calls show, 8 in 10 qualified women buy, $3,997 all in), each booked call is worth about $2,500 in expected revenue. Everything below protects or adds booked calls.
What is wrong: the booked-call event (Schedule) only fires when join.soulpunch.us/success loads. No real booking ever gets there. The calendar on /methodology ("Ways to work with Soul Punch [ORG]") redirects to soulpunchpro.com/find-trainer, the Trainer Portal, which has no Meta pixel. The calendar on /book ("Work with Soul Punch [HSL]") redirects to go.soulpunch.us/confirm, which runs a different pixel (1434515315487961) and only sends a page view. The 13 Schedule events the pixel has logged since 29 Sep cannot have come from either calendar.
The fix: in Calendly, set the confirmation redirect on both event types to https://join.soulpunch.us/success. Then book one real slot with Events Manager Test Events open and check Schedule arrives on Soul Punch Pixel #1. If the confirm page must stay, add Soul Punch Pixel #1 and a Schedule event to it instead.
Expected impact: Meta finally sees who books. Every booked call (about $2,500 in expected revenue) becomes a signal the ad set can chase, and the cost per booked call becomes a real number.
What is wrong: the ad set optimises for Schedule and has received 0 in 14 days and $650. It cannot learn. Even after fix 1, $50 a day is unlikely to produce the 50 or so booked calls a week Meta likes to learn from.
The fix: export every Calendly booking since 29 Sep from both calendars and match emails to applicants who came in with utm_source=fb_ad. Spend divided by those bookings is today's true cost per booked call. Then duplicate the ad set with the goal set to applications (SubmitApplication: 58 in 14 days, about 4 a day), same creative and audience. Keep judging it on booked calls.
Expected impact: Meta gets an event it sees every day, so delivery stops drifting to cheap, cold views (cost per 1,000 views fell to $19 to $41 since 10 Oct while clicks fell). If 1 in 10 applicants books, $11.20 per application means about $112 per booked call against about $2,500 in expected revenue.
What is wrong: the funnel uses two different calendars. The /methodology one is 30 minutes with 6 required questions (phone, services wanted, one win, how she heard, social media profile, a punctuality agreement). The /book one is 45 minutes but only shows 3 days of slots. Its confirm page opens with "Wait! Your booking is not yet complete!", which adds doubt right after she commits. The thank-you page also shows a "Digital Education Company, Scott Weddell, Australia" heading that is not Soul Punch.
The fix: keep one calendar on /methodology, straight after the alignment check. Cut it to phone plus one question. Open 7 to 14 days of slots. Confirm on /success with the homework. Text and email the booking link within 5 minutes of every application. Remove the stray heading.
Expected impact: fewer women drop between applying and booking. Even 2 extra booked calls a month is about $5,000 in expected revenue.
What is wrong: every event reaches the pixel from both the browser and the server: applications 125 and 152, video page views 151 and 156, booked calls 6 and 7. The booked-call and video page events are hand-written page-load scripts with no event ID, so a refresh counts again and Meta may not be able to pair the browser copy with the server copy. The 2 Oct "always even" pattern has eased (18 of 95 hours now odd) but the double send is still there.
The fix: read the deduplication rate for Lead, SubmitApplication and Schedule in Events Manager. Give the browser and server copy of each event the same event ID, or switch one copy off.
Expected impact: the cost per application and per booked call you see becomes the real one. About 278 application events site-wide is likely about 140 real women.
What is wrong: Video A ("Be careful isn't enough") costs $38.50 per credited application against $15.63 for Video B, Gloria's origin story. Women 25 to 34 cost $32.95 per application. Audience Network took $52 for 4 applications, and only 4% of its visitors apply against 19% from the Facebook feed. Facebook in-stream video cost $31.56 for 1. Nothing brings back the women who applied but did not book.
The fix: in the new application ad set, swap Video A for a second cut of the origin story or Gloria's past "personal development" winner. Keep the "be careful" copy ($17.60) and both headlines. Remove Audience Network and in-stream. Add a $10 to $15 a day ad set for applicants and video page visitors who have not reached /success, with a "book your call" message.
Expected impact: moving Video A's $154 to Video B's rate would have bought about 6 more applications. Warm applicants book far more cheaply than cold women.