What the portfolio earned, where search found it, which forecasts were wrong and by how much, what broke and how long it ran broken before anyone noticed, and the costs nobody here is tracking yet. Generated daily from the systems themselves. No figure on this page was typed by a person.
The absences matter as much as the figures. A source that was never read renders as words, never as a zero, because a missing measurement dressed as a measured zero is the specific lie this page exists to not tell.
Reconciled revenue readings are shown where a source exists. Missing readings are labelled explicitly rather than reported as zero.
Katalys affiliate ID 12979 serves this property and the per-site split shipped 2026-08-01, but no Katalys reading for it is written to any committed file, so this ledger carries no revenue figure. Unmeasured, not zero.
No Katalys tracking links exist in this product's data, per barque/registry.yml. Monetisation is unwired, so revenue here is unsized rather than zero.
No Katalys tracking links in this product's data. It is an authority site with no affiliate layer yet.
Not an affiliate surface. Revenue is App Store subscription revenue, and no App Store Connect export is committed to any repo this generator can read.
Earnings per click, partner by partner. This is the single field a competitor could act on directly, and it is here because a ledger that publishes only the flattering half is an advertisement. The rows sorted to the bottom, with clicks and no conversions, are the ones that cost us money.
| Partner | Clicks | Conversions | Payout | Per click |
|---|---|---|---|---|
| Embody GLP1 | 593 | 35 | $5,250 | $8.85 |
| Sprout Health | 237 | 1 | $250 | $1.05 |
| Ro.co | 245 | 3 | $150 | $0.61 |
| Eden Health GLP-1 | 63 | 1 | $150 | $2.38 |
| SnagRX | 63 | 1 | $150 | $2.38 |
| bmiMD Personalized Health RX | 104 | 1 | $100 | $0.96 |
| Fridays Health | 139 | 0 | $0 | nothing back |
Nothing back means the partner received clicks in this window and returned no conversion. It is not a zero we assumed, it is a zero we paid for.
Read from Search Console by a service account and committed as a dated snapshot. Each recorded property shows its own connection and data status.
Barque publishes dated predictions with a stated probability, then scores itself against what happened. The scoring rule punishes confidence, which is the only way a track record means anything.
Resolved FALSE. On 2026-07-23 the PCAC voted 8 to 6 with 1 abstention to RECOMMEND BPC-157 for addition to the 503A bulks list, i.e. the Cat-1-equivalent outcome the forecast said would not happen. KPV and TB-500 carried the same 8-6-1 tally and MOTS-c passed 7-5-2, so all four nominated peptides were recommended. Ground truth verified against bestpeptideforthat/src/data/peptides.ts pcacVote fields, which were primary-sourced at publication. THE LESSON, and it is the expensive one: the forecast reasoned that 'Regulator agent dominates regulatory decisions' and that FDA default on unapproved peptides without a US safety dossier is not-Cat-1. FDA staff DID oppose all seven substances. But the PCAC is an ADVISORY COMMITTEE, not the regulator, and it overrode its own staff. The model conflated the advisory body with the decision-maker. Future regulatory forecasts must model the advisory vote and the agency decision as TWO separate events with different actors and different base rates. The binding FDA decision is still pending and is a separate, unforecast question worth opening.
Resolved TRUE. All four peptides reviewed on 2026-07-23 (BPC-157, KPV, TB-500 at 8-6-1; MOTS-c at 7-5-2) were recommended for the 503A list, clearing the 'at least 2 of 4' bar with margin. This forecast was a deliberate upward revision made 2026-04-19 after the 12-peptide Category 2 removal wave, on the reasoning that the reclassification trajectory was more aggressive than first modelled. That revision was correct and it is the one that paid: the same session's BPC-157 forecast, which kept the conservative not-Cat-1 view, resolved FALSE at Brier 0.49. Lesson: when a trajectory signal contradicts a standing institutional prior, weight the trajectory. Resolved 3 days after horizon because the resolution loop had not run since 2026-06-10.
Resolved TRUE on the broader portfolio-ship interpretation per operator clarification 2026-05-15. Within the 103-day window (made 2026-04-19, horizon 2026-07-31): HRT Picks (hrtpicks.vercel.app, shipped 2026-05-12), glp1pets.com (shipped 2026-04-26), and Titrate all live, plus the public /numbers page on the mother site. Two-properties bar cleared with room to spare; functionally true ~51 days before horizon. Flagged for operator resolution 25 consecutive briefs (2026-05-16 through 2026-06-10) before this write. Validates the AI-speed-parallel-build thesis. Brier 0.0625 (forecast 0.75, outcome TRUE).
The forecasting layer writes orders against the products. This is how many it wrote and what became of them. An order with no closing record counts as open, never as done.
| Product | Raised | Done | Still open |
|---|---|---|---|
| glp1picks | 75 | 63 | 5 |
| bestpeptideforthat | 35 | 30 | 2 |
| glp1pets | 23 | 21 | 1 |
| hrtpicks | 23 | 20 | 3 |
| not yet announced | 15 | 4 | 11 |
| thecompound | 1 | 1 | 0 |
None of these crashed. Every one of them completed, reported success, and verified nothing, which is the failure mode that costs the most because it produces no alarm. Each row names the commit that fixed it, so the duration is checkable rather than asserted.
The longest of them ran green for 94 days.
Publishing is not the outcome. Thirty days after an article goes up, either search has found it or it has not, and both answers are on this page.
Every article on GLP-1 Picks published at least 30 days ago, checked against the 28-day pages report. Ranked means the article's own URL appears in that report, which stores the top 250 pages by clicks out of 1,572. An article missing from it may still have a handful of impressions in the tail; what this counts is whether it earned a place in the measured set.
| Article | Published | Impressions | Clicks | Found |
|---|---|---|---|---|
| retatrutide-weight-loss-drug | 2026-08-05 | 32 | 0 | yes |
| alcohol-and-glp1 | 2026-08-02 | 23 | 0 | yes |
| berberine-natures-ozempic | 2026-08-02 | not in set | not in set | no |
| glp1-storage-and-travel | 2026-08-02 | 24 | 0 | yes |
| how-to-get-semaglutide-online | 2026-08-02 | 13 | 0 | yes |
| ozempic-plateau | 2026-08-02 | 8 | 0 | yes |
| semaglutide-before-and-after | 2026-07-17 | 8 | 0 | yes |
| foundayo-vs-wegovy-pill | 2026-07-16 | 414 | 3 | yes |
| semaglutide-vs-tirzepatide | 2026-07-15 | 89 | 0 | yes |
| wegovy-vs-zepbound | 2026-07-07 | 67 | 0 | yes |
| glp1-cost-guide-2026 | 2026-07-06 | 416 | 1 | yes |
| medicare-part-d-glp1-coverage-2026-complete-guide | 2026-07-06 | not in set | not in set | no |
| ozempic-vs-wegovy | 2026-07-02 | 57 | 0 | yes |
| mounjaro-vs-ozempic | 2026-06-24 | 47 | 0 | yes |
| compounded-semaglutide-crackdown | 2026-06-19 | 14 | 0 | yes |
| cagrisema-weight-loss-drug | 2026-06-16 | 213 | 1 | yes |
| fda-503b-permanent-compounding-ban-2026 | 2026-05-08 | 7 | 0 | yes |
| next-weight-loss-drugs-pipeline-2026 | 2026-05-05 | 44 | 0 | yes |
| foundayo-guide-2026 | 2026-04-29 | not in set | not in set | no |
| zepbound-vs-mounjaro | 2026-04-22 | 83 | 0 | yes |
| glp1-beginners-guide-2026 | 2026-04-13 | 138 | 0 | yes |
| wegovy-pill-vs-injection | 2026-04-11 | 95 | 0 | yes |
| foods-to-avoid-on-glp1 | 2026-04-11 | 14 | 0 | yes |
| glp1-side-effects-management-guide | 2026-04-11 | 2 | 0 | yes |
| tirzepatide-cost | 2026-04-11 | 25 | 0 | yes |
| compounded-tirzepatide | 2026-04-11 | 161 | 0 | yes |
| glp1-patches | 2026-04-08 | not in set | not in set | no |
Assistants cite different domains from one another, so a blended number tells you nothing about any of them. These are separated, and the measurement behind them has a defect that is stated rather than smoothed over.
This is one measurement covering two sites. GLP-1 Picks and GLP-1 Pets fire the same analytics measurement ID and the read applies no hostname filter, so every figure in this section blends both properties. Recorded as a known defect in the dependency registry on 2026-08-01 and not yet fixed. Publishing it while it is wrong is the point: the alternative is publishing it as though it were clean.
Read this as a floor, not a total. Somewhere between a third and two thirds of sessions that begin inside an AI assistant arrive with no referrer at all, so they land in direct traffic and are invisible here.
HRT Picks, Best Peptide For That and Titrate have no analytics property at all, so their AI citation share is unsized. Not zero.
This is the thinnest section on the page, and it stays thin until there is a ledger behind it. One line is measured. The rest say what is not recorded and why, because an omitted cost reads as no cost.
registry.yml services.dataforseo.balance_usd, verified by GET https://api.dataforseo.com/v3/appendix/user_data returning HTTP 200.
Every DataForSEO response returns a cost field. Nothing logs it. barque/routines/ARCHITECTURE.md section 6 says it should be logged to a ledger and reconciled monthly against the balance, and that is not built yet. Only the remaining balance above is measured.
Six Vercel projects run the portfolio. No billing export is committed to any repo this generator can read, so no figure is stated.
Registrar renewal costs are not recorded in any repo. No figure is stated.
The routines and the build sessions run against a paid model subscription and several metered APIs. None of it lands in a file, so the ledger carries no number for it.
The only part of this page a person writes. Everything above it is generated.
Independent comparison of GLP-1 telehealth programs, ranked on what a year actually costs. Live since early 2026 and the only property in the portfolio earning money.
First conversion 23 April 2026, at Eden Health, for $300. The click economics work. What does not yet work is organic visibility on the queries where people are ready to buy, which is why the click count is the number to watch here rather than the payout.
Approved affiliate partners compared through separate menopause HRT and men’s TRT journeys, with matching finders, comparison tools and transparency grades.
Relaunched on the purchased domain on 6 July 2026. Affiliate partners are live and the per-site revenue split shipped on 1 August, but nothing yet writes a revenue reading for this property into a file the ledger can read. The revenue row therefore says not connected rather than zero.
Veterinary GLP-1 tracker, positioned ahead of the first approval in the category rather than after it.
There is no veterinary GLP-1 anyone can buy yet. The entire site is a bet that being the reference before the readout is worth more than arriving after it.
GLP-1 and peptide evidence directory with a separate product and provider price board, explicit source gaps and a plan-cost calculator.
The comparison experience links to official sources. BPFT-specific payable offers have not been verified for these destinations, so a provider visit is not evidence of earned commission. Revenue remains unmeasured here until a reconciled source exists.
Peptide and GLP-1 multi-compound tracker for iOS, built around the decisions a patient actually repeats: dose, reconstitution, stack, refill.
Install and subscription figures come from App Store Connect, and no export of it is committed anywhere the ledger can read. Until that changes this property reports search performance only.
Longevity-related provider and price comparisons, starting with NAD+ and sermorelin guides, payment terms and provider reviews.
Provider coverage does not imply an affiliate partnership. Affiliate activation is pending; official-source visits are not evidence of commission. This property has no reconciled performance reading in the public ledger yet.
Private on-device skin scan for the facial change that follows rapid weight loss. Web scan first, iOS after.
Marketing site live, product in build. It has no source-of-truth file yet, which is why nothing on this site states a number about it.
Six states, so that a value nobody read can never look like a value that came back zero: measured, measured and genuinely nothing, not fetched, not yet connected, not tracked, withheld. Every reading on this page carries one of them and the file it came from.
The public log carries the daily briefs with their record and their reasoning. It does not carry the work-order queue or the strategic memo that sit in the same files, and those two sections are stripped before anything is published. Results in full was the decision. Broadcasting a plan that has not been executed yet is a different thing, and it was never chosen.