A data-grounded read of the paywall funnel plus a four-perspective roundtable, framed around your two options: bundle everything into one site-wide subscription, or hold the current per-gazette model and wait.
The leak is upstream of pricing: almost nobody who hits the wall even reaches the point where packaging matters. And you cannot trust the numbers you would use to judge a change, because one event never fires and 13 of your 14 "active" subscribers are you. Fix the measurement, then change the ask, not the bundle.
Stripe shows 14 "active" subscriptions. That number is a mirage. Stripped of your own test purchases, here is what is actually there.
Your ~99% number is correct, and the true picture is starker: real external MRR today is €3. Every "active members" dashboard you look at is inflated by your own subscriptions.
All-time, since the analytics export began on 31 May. This is the whole story on one screen.
The dominant cliff is the first one: the wall appears and people vanish before pricing is ever on the table. Repackaging a price card only touches the ~2% who make it past this point.
Last 30 days. Traffic is real, healthy, and spread across 17 gazettes (16,444 pageviews). The question was whether visits and checkouts line up.
| Gazette | Pageviews | Visitors | Paywall | Checkout started | Real paid |
|---|---|---|---|---|---|
| Madrid | 2,803 | 1,893 | 257 | 2 | 0 |
| Valencia | 1,640 | 1,200 | 197 | 2 | 0 |
| Andalucía | 1,431 | 1,083 | 157 | 0 | 0 |
| Gipuzkoa | 1,331 | 947 | 167 | 0 | 0 |
| Bizkaia | 1,327 | 844 | 181 | 5 | 0 |
| País Vasco | 1,289 | 846 | 175 | 0 | 0 |
| Barcelona | 1,031 | 811 | 102 | 4 | 0 |
| Málaga | 1,014 | 711 | 82 | 2 | 1 |
Top 8 of 17 shown. Full catalog tails off (Sevilla, Castilla y León, Cataluña, Galicia, Murcia, Alicante, Canarias, Álava, Extremadura).
Paywall views track pageviews almost linearly (it is quota-driven). But checkout_started is 0–5 per gazette — that is statistical noise, not a demand curve. No gazette converts better than another.
No, and that is the finding. Málaga is not a top-traffic gazette and did not have your highest paywall count. It is one random cell in the grid. n = 1 cannot validate or invalidate any pricing change.
Before any pricing decision, these have to be fixed, or you will be measuring fiction.
checkout_completed has fired 0 times, ever. It is meant to fire after the Stripe redirect and is never captured, so you are blind at the exact step you are trying to diagnose.
13 of 14 "active" members are you. Any "active members", MRR, or conversion panel that does not exclude accounts@boletinbien.com is reporting a vanity number.
All four rejected your A-vs-B as a false binary — independently. That agreement is itself a signal.
The leak is uniform across the whole catalog — the fingerprint of a funnel/intent problem, not a packaging one. Do not rebundle on n=1; but do not passively wait either. Fix instrumentation and attack the paywall cliff.
A subscription is the wrong shape for one-shot Google intent. Why would someone reading one notice pay monthly? Stop pricing the subscription. Test monetizing the daily digest or a one-time unlock instead.
Option A is days of risky, semi-irreversible migration (17 Stripe products, per-gazette entitlements, moving the one real customer). Observability is the only P0. Instrument first, then decide. Per-gazette is only a real liability if "one browse-first reader" is the true direction.
She came to resolve one anxiety, not to shop. Do not hard-gate the article she Googled. Paywall the depth — alerts, "watch this oposición", the digest. Sell the future ("get told the day your results publish"), not the page.
Once measurement is honest, the genuine choice is about the offer's shape, not its bundle size.
Do not do Option A (bundle) now. It spends days of engineering to answer a question your data cannot yet even ask. Revisit it only if you deliberately commit to the "one browse-first regional reader" direction. Do not do a passive Option B either — waiting two months only helps if you fix the funnel while you wait.
checkout_completed from the Stripe webhook, and exclude your own account from every dashboard. Roughly an afternoon. Nothing else means anything until this is done.