About 1,370 readers a day reached the end-of-preview paywall over the launch weekend — in line with the weeks before, so the new design did not cost the paywall any reach. Server reading records confirm the same: preview-end traffic was flat straight across launch.
One honest caveat on rates: the in-app counter that tallies paywall views had a bug during this window and logged only about 1 in 5 of them. So a reliable "of everyone who saw the paywall, X% subscribed" figure isn't available for the launch weekend yet — the captured base is too small to trust. A fix is in progress; the rate becomes measurable again once it ships. The 16 subscriptions below are confirmed from payment records and are unaffected by the counter bug.
What we can still say about behaviour: a signed-in reader subscribes in one step, while a guest has to create an account first — and that sign-up step is where most guests drop off. Nearly every subscription came from readers who were already signed in.
Two very different journeys. A signed-in reader subscribes in one step and converts at 2.15%. Guests are the bulk of paywall views — about 25× as many — but have to create an account first, and only 0.13% subscribe: that sign-up wall is where almost all of them fall away. So a signed-in reader converts roughly 16× better than a guest. Measured per reader and per session straight from payment and reading records — not the broken counter that produced the old, inflated 3.4%.
Of the readers who actually reached the preview-end paywall, the share who subscribed within two days was essentially unchanged — 2.16% before launch, 2.15% after. The redesign held conversion steady: it neither lifted nor hurt it.
This is measured per reader straight from payment and reading records — not the in-app counter — with the same two-day window on both sides, so it sidesteps the tracking gap above and the noisy week-to-week swings in raw sign-up totals. It's the comparison that holds up to scrutiny: steady, not a step change in either direction.
Active means a live subscription being billed: 5 monthly plus 4 yearly (past their free trial, or on a plan with no trial). In free trial is a yearly sign-up still inside the 7-day free trial, not charged yet. Canceled means a reader subscribed and then canceled, all 3 during the free trial, so nothing was charged (no revenue lost).
Yearly dominates sign-ups (11 of 16), but it carries all the risk: every in-trial and every cancellation is yearly. The 5 monthly are all active and paying.