Pop-ups and welcome emails, ready for your and Nick’s go

G = MARKETING AREAS/META ADS/CREATION/GUIDEBOOK-GIFT-POPUP-2026-10-14, W = MARKETING AREAS/META ADS/WELCOME BY ANGLE, X = MARKETING AREAS/META ADS/CREATION/WELCOME-REVIEW-2026-10-16 (this page’s folder).

PACKET-LP = G/results/PACKET-LP.md. No form PATCH, ever; no beta calls on a live flow.

1. Prep, GET only. BEFORE files: forms list, TFV34a, X5yXrA with additional-fields[flow]=definition, segment QW7Nte, and the public full-forms feed from a headless opus.life load with trackers blocked (PACKET-LP 7.1).

2. Prep, the flow file. python3 -I tools/lp/flow_edits_2_3_7.py <fresh X5yXrA GET> <out> for edits 2, 3 and 7 (the read-back count, PACKET-LP 7.5: /pages/broq 18, /pages/nmn-99 18, /pages/spmd-99 6). Edit 3 is what makes an ad page count as its product’s page in the lead pick.

   Edits 1, 5 and 6 at go-live (the new form id; step 119237299 to the W9Ld87 LP copy with its subject; the name). Edit 4 (email 1) is replaced by the angle block, made by X/tools/angle_block.py and checked by X/tools/angle_check.py (notes in X/klaviyo-ready/flow/README.md), which dry-run on the file:

   • Right after delay 119237316, in place of subject split 119237349 (and its three 0% copies) and email step 119237529: split 1, a multi-branch split on the trigger event’s page (metric-property, metric SWhwdW, field page, contains, the shape X5yXrA’s lead split and edit 3 use). Its 14 paths: first /pages/stack-results-or-refund (the stack’s blood test ad page, live since midnight) to FALLBACK, ahead of TOPLIST’s /pages/stack; then one path per angle, its pages ORed (12 angles); then FALLBACK for everyone else.

     Nested page splits inside BLOODTEST, NMN, SIXYEARS, LOOK and TOPLIST, longer names first, as X5yXrA’s lead split nests. If the POST refuses nested splits, one path per page fits only with BLOODTEST’s pages sent to FALLBACK (20 paths with the stack-results-or-refund line, Klaviyo’s limit).

   • Each page: one update profile step (signup_angle for reports, welcome_skip, and welcome_lead set to the angle’s product on single-product pages; TOPLIST, LABTESTED and FALLBACK leave it to the lead split), its email 1, then the 1 day wait 119237435, where FALLBACK’s email 1 also goes.

     Then today’s placed-order check and the one lead re-check (119237620), then split 2: the same page lines, the stack-results-or-refund line first, one path per angle email 2 (17), and everyone else to today’s email 2 (step 119237637, template QRH4Yq): 19 paths, then on to the wait before email 3.

   • Before emails 3, 4 and 5 (steps 119237665, 119237692, 119237707): a split on welcome_skip that passes that email by to the wait after it. Values: 3 for SECRET, SIXYEARS, FASTING and the NMN back to 25 page; 4 for TOPLIST; 5 for LAB, PURE and LABTESTED.

   • BLOODTEST is held but built: its two email 1 steps carry FALLBACK’s email 1 and its two email 2 steps carry QRH4Yq. Releasing it means swapping the template and subject on those 4 steps on screen.

     Expected counts, held or released: 14 paths in split 1 and 19 in split 2 (under Klaviyo’s 20; 18 is the most proven in a live flow, so the POST confirms it), 38 new email steps, 3 skip splits. Every email step: sender “Nick Carter, Opus Longevity” <nc@opus.life>, subject and preview text as on this page.

     Every email 1 step (the angle email 1s and FALLBACK’s) copies step 119237529’s settings: Smart Sending off, UTM tracking on, status live.

     Every email 2 step copies step 119237637’s (Smart Sending on). The dry run checks all of them, and passed held and released.

3. Images and templates. PACKET-LP 7.2 and G/results/PACKET.md section 6 step 2 (the covers, cover-480x720.jpg for the emails). POST /api/templates, 43 in all.

   The 38 angle emails come from X/klaviyo-ready/emails/, done Thursday night by X/make_klaviyo_ready.py (list in X/klaviyo-ready/MANIFEST.txt): W’s builds with Klaviyo’s {% unsubscribe 'Unsubscribe' %} and {% manage_preferences 'Manage preferences' %} in place of the #unsubscribe and #preferences links, nothing else changed, and look email 1 for the skin page taken from X/email-fixes/ (fixed Thursday night; proof in X/email-fixes/PROOF.md). The cover’s src is still set to the uploaded cover’s Klaviyo address at upload.

   The upload stops if any #unsubscribe or #preferences is left, and Klaviyo’s renderer shows both links working. Then the LP email 1 and the W9Ld87 LP copy (PACKET-LP 7.3, tools/lp/build_email_lp.py with COVER_URL), the site email 1 and W9Ld87 guidebook copies (G/results/PACKET.md section 6, Email 1 steps 1 to 3), and the guard copy (PACKET-LP 7.3), made from the new site email 1, since that is what step 119237529 carries once the site emails are swapped.

   Klaviyo’s renderer for each; no “15%”, “OPUS15” or “code” in the ad page email 1s.

4. Site emails, then site pop-up. G/results/PACKET.md section 6, Email 1 step 4 (one preview to mike@opus.life; swap on 119237529 and 119237299 on screen; old ids kept), then steps 3 to 7 (Edit Content; the round 8 edits in A and B per G/results/diff-check.json; the CSS with the Download button’s real id; hidden fields; publish 50/50; the checks after), with PACKET-LP 7.7 in the same sitting (TFV34a gains */pages/stack*, */pages/top-5-supplements*, */pages/top-3-supplements*, */pages/lab-tested*).
5. Ad page form. PACKET-LP 7.4 (editor clone of TFV34a, matched to G/preview/payloads-lp/lp-0.json and lp-1.json), display rules per PACKET-LP section 5, except the show list starts as *opus_test=go* only. Publish.
6. New flow. X/klaviyo-ready/flow/README.md steps 3 to 8 (post-body.json, actions by temporary_id; PACKET-LP 7.5’s own POST has the same gap): POST /api/flows (draft), trigger form_id equals the new form; read back; counts; set Live on screen.
7. Guard. PACKET-LP 5b on X5yXrA, by hand; python3 -I tools/lp/guard_check.py <BEFORE> <AFTER> <new flow id> must PASS; else 5b.10 and fallback (B).

8. Segment and hold. PACKET-LP 7.6 (QW7Nte groups 0 and 3 gain the new form id; read back).

   PACKET-LP 7.8: SITE BUILD removes the “91 STACK” hold from layout/lp-stack.liquid and the top 5 and top 7 templates, after TFV34a’s list has the stack patterns and before the form’s real list. SITE BUILD ships it as a theme publish and keeps the previous theme id for the undo.

9. Tests. One sign-up per page on ?opus_test=go, each in a fresh browser context (chrome-mike background tabs, as in G/COMMON.md): the 20 angle pages, /pages/stack-v2, /pages/broq and /pages/stack-results-or-refund. The Meta pixel, analytics and TikTok are blocked; only Klaviyo’s calls go through.

   Each: on list XpXgJ5, form_id the new id, $source and opus_popup_version right, in QW7Nte, signup_angle and welcome_lead right, the angle’s email 1 received with its subject (FALLBACK’s for the blood test pages while held), Download opens, no X5yXrA email. Then PACKET-LP test profiles 5, 6 and 7, with profile 6 waiting for the angle’s email 1 (not the LP email 1).

10. Open. Show list set to PACKET-LP section 1’s allow list; publish; PACKET-LP step 10’s checks (forms list, feed diff, headless renders on ad and site pages with trackers blocked, the tracked Download on a phone, in-app browsers, Gmail dark, OPUS15’s rules).
11. Undo. PACKET-LP’s Undo; old template ids back on 119237529 and 119237299; TFV34a’s round 7e versions (A 28913132, B 28913133) back; the new flow to manual, with the ad page form to draft (or its inbox line out); one angle off by swapping its two templates and subjects on screen to FALLBACK’s email 1 and QRH4Yq and clearing welcome_skip on that path’s update-profile step; the stack hold back by SITE BUILD publishing the previous theme id.

1. The angle comes from the trigger event’s page, already used live by X5yXrA’s lead split, not from a site script writing signup_angle (method A). No SITE BUILD theme change, no race: the research ranks a property of the trigger event safest, since it is written with the event (its Q2).

   It couldn’t confirm a page property from Klaviyo’s help pages, but X5yXrA’s live lead split reads page on metric SWhwdW today (PACKET-LP edit 3’s shape), and the test sign-ups prove each path. The flow writes signup_angle for reports.

2. Email 1 stays at today’s 1 minute. Method A’s 2 to 5 minutes was only there for the script’s race.
3. The angle split lives in PACKET-LP’s new flow, not in X5yXrA: the API can’t change a live flow’s steps, and site sign-ups keep today’s welcome exactly.
4. One ad page form for all angles, as both plans say; method A’s “keep the one pop-up TFV34a” can’t carry the no-15% rule on the ad pages.

5. Ad page sign-ups get their angle’s email 1, not PACKET-LP’s LP email 1. Everyone else gets W’s FALLBACK email 1, then today’s email 2.

   The LP email 1 stays ready as the stand-in if the angle emails wait. Alternative: the LP email 1 for everyone else, since it leads with the page’s product (/pages/broq, /nmn-99, /spmd-99).

6. Test address first: PACKET-LP publishes to the real pages and then tests; here the form opens to *opus_test=go* first.
7. Variants by nested splits: 21 pages (with stack-results-or-refund) are over a split’s 19 paths plus everyone else. X5yXrA’s lead split already nests splits that meet again at one step.
8. welcome_lead: single-product angle paths set it to the angle’s product (BLOODTEST by page), as W’s entry 5 recommends, so each skip removes the angle’s own email, unless a newer cart or views switch the lead first, as today. TOPLIST, LABTESTED and FALLBACK leave it to the lead split, as today.
9. The guard’s email is a copy of the new site email 1 (the guidebook version), not of TuGvED: step 119237529 carries the new one from the site swap on.
10. Open: TOPLIST and LABTESTED have no all-three lead in emails 3 and 5, so they keep the lead split’s pick (SPMD-99 unless a cart or views say otherwise).
11. Open: email 6 (TPNM6z) still carries the OPUS15 lines for ad page sign-ups, against the ad angle question’s “reworded so none mentions a code the person never got”. The 15% code question decides; an ad page copy without those lines needs a new subject from a writer.
12. BLOODTEST held (FALLBACK meanwhile) until its emails match the reworked pages.
13. contains /pages/stack also catches stack-v2 (meant) and any future stack- page until it gets its own line. stack-results-or-refund has its own FALLBACK line ahead of it in both splits, so it never gets TOPLIST’s emails.
14. Answers that change words or add a form come after the go-live above: the question on what leads welcome email 1 A (new subject and headline), the pop-up timing question C or the phone slide-in in A (a second form), the 15% code question A to C (a code block or an expiring coupon).