[eandreeva-twr]

Kind: stale finding — two statements in the Adapty Mail sending-domain docs stopped being true.

What changed in the product

1. Which domain sends an email. For a project with several sending domains, Adapty Mail no longer prefers the newest domain. For each email it compares how the recipient's mailbox provider group opened each domain's emails on the last full day. Domains with at least 80% of the best open rate share the emails, weighted by the room left in their daily limits; the other domains send only when every domain ahead of them is over its limit. A domain with too little data (under 200 delivered to the group that day) counts among the best, so new domains still warm up. In the Other group, which is not judged on opens, domains share by room left alone. The sticky domain per recipient per campaign is unchanged.
2. A sharp fall in opens now acts on the tier. For Google, Yahoo and iCloud, the daily review also compares the last full day with the three days before it, counting only opens that arrive on the day of delivery. Under 80% of the three-day rate, the group stays on its tier instead of moving up; under 50%, it moves down two tiers at once (to Tier 1 at the lowest). The comparison runs only when each side has at least 200 delivered emails and the three earlier days reached the move-up open rate. Microsoft and Other are not compared this way.

Doc text that became wrong

• mail-sending-domain → Multiple domains: "Newest domains take priority. A domain added later is preferred over an older one." and "the next email goes out from the newest domain with room left."
• mail-sending-domain → How tiers work: "For a move down, the review leaves out the latest day" — still true for the one-tier move, not for the new two-tier one, so it now says "a move down of one tier" and a new paragraph describes the day-over-day comparison.
• mail-domain-health → troubleshooting row "A group doesn't move to the next tier": adds the fall-in-opens condition and keeps linking to How tiers work for the explanation.

Assumptions recorded

1. "The last full day" is the day before the review, and the comparison counts only same-day opens — stated as such.
2. Left out one detail: when a domain has too little data, the other domains must also reach at least the group's move-up open rate to stay in the leading set. It changes the outcome only in edge cases and would need a third sentence in the bullet.

No deliverability outcome is stated or implied; the text describes the mechanism only.