Cheap leads, no cards, and the cap again
- Delivery has stopped. Spend is $199.73 of the $200 cap. $150 went out on Sep 7 (Los Angeles day), the rest on Sep 8; the last impressions served around 19:00 LA on Sep 8.
- Put up with it (13s) is the new upstream winner. $0.47 a click, $1.46 a lead, 11 checkout taps on $35. Better than Says Who's $1.42–2.12 lifetime CPL, on a third of the watch time. It got 11 impressions in the purchase set, so it has never been tested where sales happen.
- Walking selfie (81s) took 63% of the money and has the worst unit economics. $126 across both sets, $4.67 a lead, $15.77 a checkout. The purchase set put 77% of its spend on it at $175 CPM.
- Chair workout, the only creative that has ever sold, got 8 impressions. Advantage+ inside the Identity set starved it, and Belly before/after (14 impressions), in favour of Walking selfie.
- Zero sales. 23 people opened a Stripe checkout; none entered a card. 19 came from the lead objective, now 0 for 43 lifetime; 4 from the purchase objective. No confirm failures, so the checkout is not broken: 10 closed the sheet within a minute, 13 left the page.
- Wallets are off on fiftystrong.app. Stripe live has payment-method domains for 50queen.com and checkout.stripe.com only, so Apple Pay, Google Pay and Link cannot render. Every checkout since Aug 30 logged express checkout unavailable. All 23 were in the Facebook or Instagram in-app browser, where Apple Pay is unavailable regardless, so this is a drag, not the cause.
- The Sep 2 trial's card was declined. The Chair · Identity monthly trial tried to bill $19.99 at 01:19Z on Sep 9 and got
insufficient_funds. The subscription is past due; Stripe retries on Sep 16.
$150 on day one, then the cap
| Identity · purchase | Leads v2 · lead | Total | |
|---|---|---|---|
| Spend | $94.41 | $105.36 | $199.77 |
| Impressions | 517 | 3,192 | 3,709 |
| CPM | $183 | $33 | $54 |
| Link clicks | 41 | 179 | 220 |
| CTR | 7.9% | 5.6% | 5.9% |
| CPC | $2.30 | $0.59 | $0.91 |
| Entrants (human) | 39 | 178 | 217 |
| Progressed (≥2 screens) | 21 · 54% | 77 · 43% | 98 · 45% |
| Leads | 17 | 45 | 62 |
| Lead rate | 43.6% | 25.3% | 28.6% |
| Cost per lead | $5.55 | $2.34 | $3.22 |
| Checkouts | 4 | 19 | 23 |
| Checkout rate | 10.3% | 10.7% | 10.6% |
| Cost per checkout | $23.60 | $5.55 | $8.69 |
| Sales (trials + paid) | 0 | 0 | 0 |
| Sales rate | 0% | 0% | 0% |
The purchase objective again bought a higher-intent entrant (44% went to email against 25%) but at four times the cost per click, so checkouts per entrant came out identical at 10% and the cost per checkout four times worse. On Sep 4–5 the same set, then serving only Chair, produced two trials on $117.
Both sets combined
| Creative | Spend | Impr | Clicks | CTR | CPC | Entrants | Prog. | Leads | Lead rate | CPL | Checkouts | Checkout rate | $ / checkout | Sales | Sales rate | Full watch |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Put up with it · 13s | $38.07 | 1,438 | 75 | 5.2% | $0.51 | 75 | 36 | 24 | 32.0% | $1.59 | 11 | 14.7% | $3.46 | 0 | 0% | 18.5% |
| Nothing special · Walking selfie · 81s | $126.15 | 1,576 | 105 | 6.7% | $1.20 | 102 | 46 | 27 | 26.5% | $4.67 | 8 | 7.8% | $15.77 | 0 | 0% | 2.8% |
| Says Who · 15s · validation cut | $33.38 | 671 | 39 | 5.8% | $0.86 | 38 | 15 | 11 | 28.9% | $3.03 | 4 | 10.5% | $8.35 | 0 | 0% | 15.5% |
| Belly · Same outfit before/after · 11s | $1.47 | 16 | 1 | 6.3% | $1.47 | 2 | 1 | 0 | 0% | – | 0 | 0% | – | 0 | 0% | 6.3% |
| I'm 55 · Chair workout · 32s | $0.70 | 8 | 0 | 0% | – | 0 | 0 | 0 | – | – | 0 | – | – | 0 | – | 0% |
CTR is link clicks over impressions. Leads are PostHog email captures; Meta's pixel counted 57 to PostHog's 62. Full watch is 100% video plays over impressions. Six entrants arrived without Meta's ad-set tag (shared or copied links) and are folded into their creative.
Put up with it holds to the end: 264 of 1,427 impressions watched all 13 seconds, and 15% of its entrants tapped a plan. Walking selfie averages 12 seconds of its 81 and 2.8% reach the end. It wins the auction on the hook, then does not close.
Where Meta put the money
| Ad set | Creative | Spend | Impr | Clicks | CTR | CPC | CPM | Entrants | Prog. | Leads | Lead rate | CPL | Checkouts | Checkout rate | Sales | Sales rate |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Leads v2 | Put up with it | $34.94 | 1,427 | 74 | 5.2% | $0.47 | $24 | 74 | 35 | 24 | 32.4% | $1.46 | 11 | 14.9% | 0 | 0% |
| Leads v2 | Walking selfie | $53.90 | 1,162 | 71 | 6.1% | $0.76 | $46 | 71 | 30 | 13 | 18.3% | $4.15 | 4 | 5.6% | 0 | 0% |
| Leads v2 | Says Who validation | $16.48 | 601 | 34 | 5.7% | $0.48 | $27 | 33 | 12 | 8 | 24.2% | $2.06 | 4 | 12.1% | 0 | 0% |
| Leads v2 | Belly before/after | $0.04 | 2 | 0 | 0% | – | – | 0 | 0 | 0 | – | – | 0 | – | 0 | – |
| Leads v2 | Chair workout | $0.00 | 0 | 0 | – | – | – | 0 | 0 | 0 | – | – | 0 | – | 0 | – |
| Identity | Walking selfie | $72.25 | 414 | 34 | 8.2% | $2.12 | $175 | 31 | 16 | 14 | 45.2% | $5.16 | 4 | 12.9% | 0 | 0% |
| Identity | Says Who validation | $16.90 | 70 | 5 | 7.1% | $3.38 | $241 | 5 | 3 | 3 | 60.0% | $5.63 | 0 | 0% | 0 | 0% |
| Identity | Put up with it | $3.13 | 11 | 1 | 9.1% | $3.13 | $285 | 1 | 1 | 0 | 0% | – | 0 | 0% | 0 | 0% |
| Identity | Belly before/after | $1.43 | 14 | 1 | 7.1% | $1.43 | $102 | 1 | 1 | 0 | 0% | – | 0 | 0% | 0 | 0% |
| Identity | Chair workout | $0.70 | 8 | 0 | 0% | – | $88 | 0 | 0 | 0 | – | – | 0 | – | 0 | – |
Both sets are single Advantage+ ad sets holding five ads, so Meta picks a winner after a few dozen impressions. In Leads v2 it split three ways. In Identity it gave 77% to Walking selfie and left Chair, Belly and Put up with it 33 impressions between them. The creatives Eugene most wanted read never got a read.
Twenty-three sheets, no cards
| Day (UTC) | Checkout started | Sheet viewed | Wallets available | Wallets unavailable | Sheet closed | Confirm failed | Sales | Sales rate of checkouts |
|---|---|---|---|---|---|---|---|---|
| Sep 4 | 3 | 3 | 0 | 3 | 3 | 0 | 0 | 0% |
| Sep 5 | 14 | 17 | 0 | 17 | 6 | 1 | 2 | 14.3% |
| Sep 7 | 10 | 11 | 0 | 11 | 4 | 0 | 0 | 0% |
| Sep 8 | 12 | 12 | 0 | 12 | 5 | 0 | 0 | 0% |
| Sep 9 to 03:00Z | 1 | 1 | 0 | 1 | 1 | 0 | 0 | 0% |
| Lifetime, lead objective | 43 | 0 | 0% | |||||
| Lifetime, purchase objective | 13 | 6 | 46.2% | |||||
- Every one of the 23 reached the embedded Stripe sheet and no confirm failure fired, so nobody typed a card and was refused. The single failure since Sep 4 was a Sep 5 CVC typo that succeeded on retry.
- Ten closed the sheet themselves after 9, 20, 11, 26, 26, 2, 22, 10, 36 and 57 seconds. Thirteen left the page.
- Plans opened: belly 12-week $59 ×11, belly monthly $19.99 ×8, control monthly ×2, control trial59 ×2. Every session came from the Facebook or Instagram in-app browser.
- Wallets last rendered on Aug 29–30, on 50queen.com. Stripe live registers 50queen.com (Apple Pay, Google Pay, Link, PayPal active) and checkout.stripe.com; fiftystrong.app is not registered, so since the Sep 4 move back nothing but the card form can show. In-app browsers block Apple Pay anyway; Google Pay and Link would still be on the table.
- Lifetime by objective: lead objective 0 completed of 43 checkouts; purchase objective 6 of 13.
At Sep 5's rate (2 of 14) the chance of 0 in 23 is about 2%; at the purchase objective's lifetime rate it is vanishingly small. This run's checkouts are 83% lead-objective traffic, which has never converted, so the zero is mostly the audience. The four Identity checkouts are too few to say anything.
Still no purchase read
| Arm | Paywall views | Checkouts | Checkout rate (of paywall) | Sales | Sales rate (of paywall) |
|---|---|---|---|---|---|
| belly_12week · $59 | 24 | 11 | 45.8% | 0 | 0% |
| belly_monthly · $19.99 | 21 | 8 | 38.1% | 0 | 0% |
| control · no flag | 16 | 4 | 25.0% | 0 | 0% |
The arm flag fires at the paywall, so paywall views are the denominator here, and there is no CTR or lead rate per arm. Both belly arms tap into checkout more than control and neither has produced a trial. The decision metric is purchase rate; keep it running.
British impressions, Canadian and Australian intent
| Country | Entrants | Leads | Lead rate | Checkouts | Checkout rate | Sales | Sales rate |
|---|---|---|---|---|---|---|---|
| United Kingdom | 71 | 15 | 21.1% | 3 | 4.2% | 0 | 0% |
| United States | 68 | 16 | 23.5% | 5 | 7.4% | 0 | 0% |
| Canada | 52 | 17 | 32.7% | 9 | 17.3% | 0 | 0% |
| Australia | 24 | 14 | 58.3% | 6 | 25.0% | 0 | 0% |
Advantage+ across US, CA, GB and AU bought the cheapest impressions, which are British. Meta does not break impressions out by country in this pull, so there is no CTR per country. Historical purchases carry no geo (server-side events), so this cannot yet be tied to money.
Stripe live
| Started | Creative · campaign | Plan | State |
|---|---|---|---|
| Sep 2 00:16Z | Chair · Identity (fiftystrong arm) | monthly $19.99 | Past due. $19.99 declined Sep 9 01:19Z (insufficient funds); retry Sep 16 |
| Sep 2 01:34Z | Overhead guided workouts · purchases-launch | trial59 $59 | Cancelled Sep 3 |
| Sep 5 08:26Z | Chair · Identity (fs.app) | belly monthly $19.99 | Trialing, ends Sep 12 |
| Sep 5 08:30Z | Chair · Identity (fs.app) | belly 12-week $59 | Trialing, ends Sep 12 |
Paid to date is unchanged: $14.99 four-weekly (Aug 30, active, ends Nov 22) and $49 twelve-week (Sep 1). Nothing from this run.
Against every earlier run
| Batch (objective, dates) | Spend | Impr | Clicks | CTR | Entrants | Leads | Lead rate | CPL | Checkouts | Checkout rate | $ / checkout | Sales | Sales rate |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Quiz Leads validation (lead, Aug 7–10) * | $195.30 | 6,485 | 539 | 8.3% | 539 * | 209 * | 38.8% * | $0.93 | 12 * | 2.2% * | $16.28 | 0 | 0% |
| Quiz Leads launch (lead, Aug 21–24) | $325.42 | 13,334 | 1,031 | 7.7% | 987 | 229 | 23.2% | $1.42 | 3 | 0.3% | $108.47 | 0 | 0% |
| Purchases launch (purchase, Aug 28–30) | $128.57 | 1,846 | 68 | 3.7% | 83 | 13 | 15.7% | $9.89 | 1 | 1.2% | $128.57 | 1 † | 1.2% |
| Identity test, Chair (purchase, Aug 30–Sep 5) | $331.00 | 2,523 | 91 | 3.6% | 84 | 14 | 16.7% | $23.64 | 8 | 9.5% | $41.38 | 5 | 6.0% |
| Leads v2, Says Who + Chair (lead, Sep 1–5) | $114.53 | 4,104 | 241 | 5.9% | 212 | 51 | 24.1% | $2.25 | 21 | 9.9% | $5.45 | 0 | 0% |
| This run, Leads v2 (lead, Sep 7–8) | $105.44 | 3,198 | 179 | 5.6% | 172 | 44 | 25.6% | $2.40 | 19 | 11.0% | $5.55 | 0 | 0% |
| This run, Identity (purchase, Sep 7–8) | $94.41 | 517 | 41 | 7.9% | 37 | 17 | 45.9% | $5.55 | 4 | 10.8% | $23.60 | 0 | 0% |
| This run, both sets | $199.85 | 3,715 | 220 | 5.9% | 209 | 61 | 29.2% | $3.28 | 23 | 11.0% | $8.69 | 0 | 0% |
* Validation predates the PostHog ingest fix: leads and checkouts are Meta pixel counts and the rates use link clicks as the denominator. † The purchases-launch trial was cancelled inside a day. Entrant counts in this table are per campaign over all time, so a few returning quiz ids sit in a different row than in the sections above (209 here against 217).
On leads this run is level with last week's Leads v2 and two to three times dearer than August's Says Who campaigns. On checkouts the lead half matches last week, about 18 per $100, and is twenty times August; the purchase half is ahead of Chair per dollar. On sales it trails the only batch that ever sold: Chair's 8 checkouts became 3 trials and 2 payments, and lead-objective checkouts are 0 for 43 lifetime, so the checkout column flatters this batch.
Lead objective against purchase objective, all ads combined
| Objective · period | Spend | Impr | Clicks | CTR | Entrants | Leads | Lead rate | Checkouts | Checkout rate | Sales | Sales rate | CPL | $ / checkout | $ / sale |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Lead · this run (Leads v2, Sep 7–8) | $105.44 | 3,198 | 179 | 5.6% | 172 | 44 | 25.6% | 19 | 11.0% | 0 | 0% | $2.40 | $5.55 | – |
| Purchase · this run (Identity, Sep 7–8) | $94.41 | 517 | 41 | 7.9% | 37 | 17 | 45.9% | 4 | 10.8% | 0 | 0% | $5.55 | $23.60 | – |
| Lead · lifetime (Aug 21 → Sep 8, three campaigns) | $545.39 | 20,636 | 1,451 | 7.0% | 1,371 | 324 | 23.6% | 43 | 3.1% | 0 | 0% | $1.68 | $12.68 | – |
| Purchase · lifetime (Aug 28 → Sep 8, three campaigns) | $553.98 | 4,886 | 200 | 4.1% | 204 | 44 | 21.6% | 13 | 6.4% | 6 | 2.9% | $12.59 | $42.61 | $92 |
| Objective · period | Entrants → lead | Lead → checkout | Checkout → sale | Per $100: leads | Per $100: checkouts | Per $100: sales |
|---|---|---|---|---|---|---|
| Lead · this run | 25.6% | 43.2% | 0% | 41.7 | 18.0 | 0 |
| Purchase · this run | 45.9% | 23.5% | 0% | 18.0 | 4.2 | 0 |
| Lead · lifetime | 23.6% | 13.3% | 0% | 59.4 | 7.9 | 0 |
| Purchase · lifetime | 21.6% | 29.5% | 46.2% | 7.9 | 2.3 | 1.1 |
Lifetime lead objective is Quiz Leads launch, Leads v2 week one and this run; the Aug 7–10 validation campaign ($195, 209 pixel leads) predates PostHog and is left out so every rate has the same base. Lifetime purchase objective is Purchases launch, the Chair identity run and this run. The 6 sales are 3 trials, 2 payments and the cancelled Overhead trial; on the 5 kept it is $111 a sale, and $277 per payment so far with one trial past due and two ending Sep 12.
Which objective to target
Purchase. It is the only objective that has ever produced a card: 6 sales from 13 checkouts against 0 from 43. The lead objective wins every upstream number, CTR 7.0% against 4.1%, leads at $1.68 against $12.59, eight times the leads per $100, and then loses all of it at the card form. The purchase objective's entrants go to email at the same rate, but 30% of its leads open a checkout against 13%, and 46% of its checkouts pay against none.
The confound is creative: the purchase objective has mostly run Chair, the lead objective mostly Says Who. Chair on the lead objective still produced 0 sales from 7 checkouts, which points at the objective rather than the ad. This run is the first time the purchase objective went to other creatives and it produced 0 from 4, too few to read.
The cost of the answer is a purchase-objective lead at $12.59 and a checkout at $43, so it only works if a sale is worth more than about $110. Run the lead objective only for the email list.
What the numbers say
Put up with it is the creative to promote
Cheapest click and lead in the account this week, 18.5% full watch on a 13-second cut, and 15% of its entrants tapped a plan against 8% for Walking selfie and 11% for Says Who. It has had no purchase-objective delivery, which is where the account's only sales came from.
Walking selfie is a hook without a close
Highest CTR, 12-second average watch on an 81-second video, worst cost per lead and per checkout. In the Identity set it also crowded out Chair.
Says Who is doing what it always does
Cheap clicks, leads at $2, checkouts that do not complete. Now 0 for 21 checkouts lifetime.
The purchase objective is not a sales machine by itself
This run it paid $175–241 CPM for an audience that went to email at twice the rate but checked out at the same rate, and Meta spent it on the wrong creative. The two Sep 5 trials came from Chair when Chair was the only ad in the set.
Five ads in one Advantage+ set is not a test
Meta chose a winner on hook after about 100 impressions. Anything that must be read needs its own ad set, or a set with fewer, matched ads.
Six moves
- Reset the cap and raise it. Two resets in four days; at $100 a day of budget a $200 cap stops delivery every second day and each stop reshuffles Meta's learning. Reset with
spend_cap_action=reset, thenspend_cap=60000covers a week. The amount is Eugene's call. - Register fiftystrong.app as a Stripe payment-method domain. One live command, reversible, restores Google Pay and Link inside the in-app browser and Apple Pay for anyone who opens the link in Safari. Not the cause of 0 for 23, but there is no reason to run paid traffic without wallets.
stripe payment_method_domains create --live -d domain_name=fiftystrong.app - Identity set: pause Walking selfie and Says Who, run Chair and Put up with it only. The only creative that has sold against the best upstream creative, on the objective that has sold. Success is a trial inside the first $150.
- Leads v2: pause Walking selfie. $4.15 a lead against $1.46 and $2.06. Keep Put up with it and Says Who at $50 a day only if a $2 email is worth having; the lead objective has still never produced a card.
- Belly before/after stays untested. 16 impressions across both sets. If it matters, it needs its own set or a turn with the others paused. Read nothing into $1.47.
- Read again on Sep 12 and Sep 16. Both Sep 5 trials bill or cancel on the 12th; the declined card retries on the 16th. Chair's paid count is 2, with one past due and two pending.
How this was read
- Meta
- Account act_1761459111652871, ad-level insights Sep 6–9 with daily increments, actions, link clicks and video fields; account-level hourly breakdown; creative asset feed spec for each ad's landing URL. Raw JSON saved beside this readout.
- PostHog
- Prod project 559701 via the CLI's execute-sql. One CTE per quiz id: first non-empty campaign, creative and ad-set tag, entry URL, screens viewed, email, paywall, checkout, trial and purchase flags, belly arm from the flag call, geo. Humans only: Meta data-centre cities and lpv-test excluded, first seen after Sep 7 12:00Z. Queries saved as sql/*.sql.
- Stripe
- Live mode: checkout sessions, subscriptions by status, charges, the past-due invoice, payment-method domains. Joined to PostHog on quizPublicId.
- Prior read
- Creative performance, 2026-09-07: lifetime numbers for Says Who, Chair and the purchases-launch tail.