Hard paywall from read 1: first-weekend analysis and forecast

The hard wall (zero free reads, 7-day card-upfront trial) went live Friday, July 31 at 13:00 Madrid. This report reads the PostHog export and Stripe webhook ledger through Sunday, August 2 (21:00 Madrid), compares the funnel before and after the wall, projects each day of next week plus the three following weeks, and gives a verdict on whether the wall is earning its keep.

Data: posthog.public.events (prod export) · billing.stripe_events · identity.subscriptions · engagement.article_reads. All days are Europe/Madrid. Generated 2026-08-02.

Visitors, last 7 days

6,491

+16% vs prior week

Visitors hitting the wall

90%

of weekend visitors (602 of 671)

Checkout starts per day

15–21

was 1–5 per day pre-wall

Trials started

canceled 5 minutes in

Paid conversions

first possible on Aug 7

Weekend traffic vs prior weekend

−2%

671 vs 685 visitors: no damage yet

Daily visitors, actual and forecast

Traffic has grown roughly 6× in five weeks (weekly visitors: 965 → 2,270 → 2,555 → 5,586 → 6,491), almost entirely Google organic. The two Tuesday spikes (Jul 21, Jul 28) were one-off SERP surges. The forecast band carries the base case with an August-seasonality decay of about 5% per week; the wide upper range exists because spike days are real but unpredictable.

Unique visitors per day (Jun 21 – Aug 2 actual, Aug 3 – 9 forecast)

Actual Forecast (base) Forecast range Wall live (Jul 31, 13:00)

Checkout starts: the wall works as an intent amplifier

Unique people starting a Stripe checkout ran 1–5 per day for the whole soft-wall era. In the first two full wall days (a weekend, the slowest traffic of the week) it hit 19 and 15. Same-weekend comparison: 1 checkout starter on Jul 25–26 vs 34 on Aug 1–2. Whatever else the wall does, it multiplies checkout intent by an order of magnitude.

Unique checkout starters per day (Jul 15 – Aug 2 actual, Aug 3 – 9 forecast)

Actual Forecast (base, whisker = range)

The post-wall funnel, and where it dies

Aug 1–2 (the two full wall days). The wall itself performs: 90% of visitors see it, and 5.6% of the walled click through to Stripe. The collapse is on the Stripe page: the card-upfront form converts 3% of the people who reach it. 97% of the intent the wall generates dies at the card field.

Forecast: next week, day by day (Aug 3 – 9)

Base case with low–high ranges. Model: visitors from same-weekday history (spike-trimmed, minus ~5% August seasonality); pageviews = 1.42 × visitors and walled uniques = 87% of visitors (post-wall observed); checkout starters = 5.2% of walled (range 3–6.5%); completed checkouts (trial starts) = 2.5% of starters (range 1–6%, observed n=1/34). Paid conversions next week are structurally ~zero: the only cohort whose trial ends inside the week is Jul 31 – Aug 2, which holds exactly one trial, already canceled.

Day	Visitors	Pageviews	Sessions	Bounce	Hit the wall	Checkout starts	Completed (trials)	Paid
Mon Aug 3	750500–1,200	1,050	765	~84%	650	3420–49	0.9	0
Tue Aug 4	800500–2,300	1,150	815	~84%	695	3621–52	0.9	0
Wed Aug 5	700450–1,100	1,000	715	~84%	610	3218–46	0.8	0
Thu Aug 6	800550–1,050	1,150	815	~84%	695	3621–52	0.9	0
Fri Aug 7	550400–700	800	560	~84%	480	2514–36	0.6	0
Sat Aug 8	330240–430	470	335	~82%	285	159–21	0.4	0
Sun Aug 9	280200–360	400	285	~82%	245	137–18	0.3	0
Week total	4,2102,840–7,140	6,020	4,290	~84%	3,660	191110–274	≈51–11	00–1

Forecast: the following weeks (through the Aug 28 checkpoint)

Paid conversions in each week are the prior week's trials times an assumed 20% trial-to-paid rate (range 5–40%). That rate is the single most uncertain number in this report: the only observed trial canceled within 5 minutes, which is evidence for the low end. August seasonality (Spanish holidays, thinner gazette publication) trims traffic ~5% per week. A possible SEO penalty from wall-induced pogo-sticking is NOT in the base case; it would show in Search Console with a 2–6 week lag.

Week	Visitors	Pageviews	Checkout starts	Completed (trials)	Paid conversions	Added MRR (cumulative)
Aug 3 – 9	4,2102,840–7,140	6,020	191110–274	≈51–11	00–1	€0
Aug 10 – 16	4,0002,600–6,600	5,700	180100–260	≈4.51–10	≈10–4	≈€3
Aug 17 – 23	3,8002,400–6,200	5,400	17295–250	≈41–10	≈10–4	≈€6
Aug 24 – 30	3,6502,200–6,000	5,200	16590–240	≈41–10	≈10–4	≈€9
Test cumulative	~15,700	~22,300	~710	≈185–40	≈30–10	≈€9€0–30

Before vs after, honestly controlled

The two full wall days are a weekend, so the fair comparison is the same weekend one week earlier, not the weekday average. On that basis traffic and bounce are unchanged, checkout intent exploded, and the engagement layer went dark.

Metric	Weekend Jul 25 – 26 (soft wall)	Weekend Aug 1 – 2 (hard wall)	Read
Visitors	685	671	flat (−2%)
Bounce rate	85.1%	81.9%	flat, maybe slightly better
Unique wall impressions	38	602	16× by construction
Unique checkout starters	1	34	34×: the wall creates intent
Completed checkouts	0	1 (trial, canceled in 5 min)	the Stripe card page is the choke point
Article reads recorded	985	0	by design: read-tracking is quota-gated, "Lo más leído" starves
Digest signups	1	2	weekend-typical either way; the real test is weekdays (was 6–10/day Jul 28–30)

Verdict: is the hard wall from read 1 worth it?

Not yet proven worth it, and the base case says it won't be, but the deciding number arrives Aug 10 – 14. Hold the test to the checkpoint.

What two full days establish, and what they cannot:

The wall converts attention into intent better than anything tried so far. Checkout starts went from 1–5 per day to 15–21 per day, a same-weekend jump of 34×. The audience (largely public employees reading their own gazette) will click "Seguir leyendo gratis". That finding survives even if the trial flops.
No visible audience damage yet. Weekend traffic −2%, bounce unchanged. The SEO guardrail (Search Console clicks) lags weeks and must still be watched.

The card-upfront trial is where 97% of the created intent dies. 45 checkout starts produced exactly 1 trial (2–3%), and that trial canceled 5 minutes in, using the trial as a free read with extra friction. The wall is not the weak link; the Stripe card form is.
The base-case payoff is single-digit euros. ~18 trials and ~3 paid conversions (~€9 MRR) by the Aug 28 checkpoint, against a traffic asset compounding at double-digit weekly growth and a read-engagement layer (article reads, "Lo más leído") that the wall switches off entirely.
The wall taxes the compounding assets to feed the weakest asset. Every walled read is a person who might have become a repeat organic visitor or digest subscriber. If weekday digest signups do not recover to ~5 per day this week, the wall is eating the email list to feed a 3% card form.

Decision rule for the checkpoint (agreed measure: paid conversions, not trial starts):

Keep the wall if trial-to-paid lands ≥ 30% (≥ 2 of the ~5 trials expected from the Aug 3 – 9 cohort pay on Aug 10 – 16). That pace (~3–4 payers/month) beats the soft-wall era (1 payer in 2 months) by an order of magnitude and justifies the engagement tax.
Revert early (do not wait for Aug 28) if any of: cumulative paid conversions < 3 by Aug 21; Search Console organic clicks fall > 20% week-over-week for two consecutive weeks; weekday digest signups stay under 2 per day through Aug 8.
Either way, the next experiment is the mid-funnel, not the wall. The measured cliff is the card field, so the highest-leverage variant is a softer capture between wall and card (for example a no-card trial or an email-first step), not a return to the old 1-free-read status quo, which produced measurably near-zero intent.

Table view: full daily series (Jun 21 – Aug 2)

Day	Pageviews	Visitors	Sessions	Bounce %	Wall shown	Wall uniq	Checkout starters	Digest signups