The send itself
Time of day, language, interruption level, Android channel, and whether the copy promised a parcel, a sale, or a bank alert. A number without that context is a poster with the type stripped off.
Kuala Lumpur
Push notification reviews
UTILITY
GARDENHUB
Lock-screen counts, read without the monthly slide language.
We sit with the send log, the copy calendar, and the opt-out file. Then we tell you what those figures actually describe — delivered is not opened, opened is not read, and a festive blast is not the same week as a quiet Tuesday.
01
What we actually examine
App analytics for push notification engagement is a narrow subject. We stay inside it: the messages that were sent, the counts the operating systems return, and the human decisions that produced both.
The returned counts
Delivered, displayed, opened, dismissed, and permission withdrawn. We keep those columns apart. Mixing them into one “engagement” figure is the most common error we are asked to unwind.
The week around it
Hari Raya bursts, payday reminders, and a quiet mid-month lull do not belong on the same average. We read clusters, not a single heroic percentage.
◆ ◆ ◆
Push notification engagement audit
A structured reading of how an app’s alerts are sent, how iOS and Android return their counts, and what those figures can honestly be said to mean.
Written findings, a copy sample, and a sitting with the people who press send. Remote or at the Kuala Lumpur desk. Typical span: three to four weeks once logs are in hand.
02
Related work on the same desk
If a full audit is more than you need, the shorter reviews still start from real sends rather than from a generic scorecard.
Cadence and copy review
A shorter pass over a fortnight of alerts: when they went out, what they said, and whether the opt-outs clustered on particular mornings.
Team briefing on the counts
A half-day sitting for the people who have to explain opens, dismissals, and permission loss in a Monday meeting — using your own recent sends as the examples.
How a review runs, week by week
Intake, log sample, copy sample, a conversation with the senders, then a written note. No dashboard walkthrough dressed up as a meeting.
03
From a grocery app in Petaling Jaya
We do not publish star ratings. The useful evidence is what someone asked us to look at, and what changed in the send calendar afterwards.
“They refused to treat ‘delivered’ as a compliment. Once the Android channel report was split from the iOS opens, we could see the Saturday morning fruit-promo was doing the damage. The written note arrived a week later than promised because our export was a mess — that delay was on us, and they said so plainly.”
04
Field notes
Short pieces on counts, copy, and Malaysian send weeks — written from the same work, not from general business advice.
Sample seven days of copy before you change the programme
Print the alerts in the order they arrived. Margin notes beat a new template library that nobody will send.
Quiet hours that match how households use the phone after dinner
A 22:00 cap copied from a European playbook will not describe a Klang Valley evening. Look at dismissals by hour before you write a rule.
After the festive blast: reading a week of opt-outs in Malaysia
Hari Raya, Deepavali, and Chinese New Year weeks do not belong inside a monthly average. The opt-out file is often the only honest narrator.