Why cost per landing-page view went from about $0.70 to $4.75
FitX Meta account · Aug 15 – Sep 21, 2026 · read-only data pull, prepared Sep 21 · download the full daily data (CSV)
Short answer: the $0.70 number and the $4.75 number come from two different kinds of campaigns. The cheap one bought cheap views (Leads goal, ~$15–20 CPM). The expensive one asked Meta to find buyers (Purchase goal), which costs 3–6× more per 1,000 people — and on Saturday we also spent $1,884 in a few hours across 16 brand-new ad sets that were all still in Meta's learning phase. The ads themselves got more clicks per view, not fewer.
The eras, side by side
Cost per landing-page view = CPM ÷ 1000 ÷ (click rate × click-to-page rate). Every number below is in the CSV.
| Era | Dates | Meta goal | CPM | Click rate | Click → page | Cost / page view |
|---|
| Leads campaigns (the ~$0.70 baseline) | Aug 15 – Sep 8 | Leads | $14–21 | ~2.4% | ~70–78% | $0.62–1.36 |
| Hook test | Aug 21 – 31 | Traffic-style | $26–37 | ~4.2% | ~65% | $0.93–1.55 |
| Problem Test (first Purchase-goal campaign) | Sep 2 – 8 | Purchase | $39–59 | ~4.4% | ~55–65% | $1.57–3.25 |
| Deposit Purchase campaigns | Sep 9 – 17 | Purchase | $47 → $120 | ~4% | ~50–70% | $2.41–5.93 |
| Saturday: 16 concepts, $150/set | Sep 19 | Purchase | $104.92 | 3.93% | 56.2% | $4.75 |
What actually changed
| August (leads era) | Saturday Sep 19 | Effect on cost |
|---|
| CPM (cost per 1,000 views) | ~$17 | $104.92 | 6× worse — the main driver |
| Click rate | ~2.4% | 3.93% | 1.6× better — creatives improved |
| Click → page rate | ~70% | 56% | 1.25× worse — page/tracking question |
Three findings
- The goal switch, not the ads, caused most of it. On Sep 2 both kinds ran the same day: the Leads campaign paid a $14.73 CPM while the new Purchase campaign paid $46.91 — 3× on day one, before any creative changed.
- CPM kept climbing ($47 → $105–120) because every launch restarted learning. Ad frequency stayed at 1.1–1.3, so people were not seeing the ads repeatedly — this is not audience fatigue. 313 ads went live Sep 17, then 80 more Sep 19; Saturday pushed $1,884 through 16 brand-new ad sets in roughly 3.5 hours, all in learning phase at aggressive pacing.
- Fewer clicks turned into page views (~70% → 56%) — and PostHog says Saturday's page views were real. PostHog logged more Facebook sessions than Meta counted landing-page views (469 vs 397), so Meta is not overstating Saturday. The drop from August cannot be checked in PostHog (tracking only began Sep 18) and most likely reflects the stricter Purchase-goal audience and the new $299 page, not a broken pixel.
PostHog cross-check (site-side data)
Our own site analytics (PostHog) started recording Sep 18, 15:00 UTC — it can verify Saturday, but not the August $0.70 era. Full day-by-day file: meta_vs_posthog.csv.
| Saturday Sep 19 | Meta says | PostHog says | Read |
|---|
| People landing from Facebook | 397 landing-page views | 469 sessions (468 with clean ad tags) | Traffic was real; Meta counts conservatively (1.37× more sessions on site than Meta credits) |
| Main landing page | /pages/size-is-a-length — 505 of 653 page views | The ads landed where intended |
| On-page funnel | 543 sessions → 68 CTA clicks (12.5%) → 29 checkout starts → 14 purchase events | People did engage past the landing page |
Purchase counts disagree (PostHog 14 events vs Meta 1 attributed purchase) — different definitions, plus Sep 18 contains internal test traffic (will-fb-test / luke-ig-test tags, 4 "purchases" on 38 sessions). Treat purchase rows as unreconciled until deduped.
Data caveats before deeper conclusions
- The $0.70 baseline is from Leads-goal campaigns; it is not a fair benchmark for Purchase-goal campaigns.
- "Clicks" here counts all clicks, not just link clicks; page views come from the Meta pixel with its default attribution window.
- Meta and PostHog count differently by design: Meta "landing-page view" requires the pixel to fire after an ad click within its attribution window; PostHog counts every session. On the one fully overlapping day they differ by a stable 1.37× — consistent, not broken.
- PostHog coverage starts Sep 18, 15:00 UTC. The August ($0.70) era is Meta-only and cannot be independently verified.
- PostHog purchase/checkout events include internal test traffic from Sep 18 and are not yet deduped against Meta attribution.
- The Sep 18 launch evening shows under Sep 19 in Meta's daily rows; Sep 20–21 are zero because everything is paused.
No recommendations here on purpose — facts only. All campaigns are currently paused; no new ads have been launched. Source: Meta Ads API, account act_1468784101549109, pulled read-only on Sep 21.