Ayutthaya · Push analytics lab

Read the send, not just the tap.

Most dashboards stop at open rate. Operators who keep users still need delivery truth, quiet-hour leakage, token decay, and whether a sequence actually moved a next action. Apieventgrid is a small school for that craft.

Open the Signal to Send syllabus

Analytics charts on a laptop screen
Figure 01 — delivery latency by local hour, Bangkok sample used in class.
2,180 notification sequences reviewed with operators since 2023
17 min median time from first ingest to a usable decision in the lab
28 seats per Signal to Send cohort — we do not run lecture halls

From the desk

Operators, not fan letters.

These notes came after real modules, not after a survey prompt. One of them still argues with us.

“Module 2 forced us to split ‘accepted by FCM’ from ‘shown on lock screen’. That single cut explained why our Friday grocery blast looked healthy and still failed to move baskets.”

Waranya P. · CRM lead, Bangkok grocery app · Signal to Send

“Quiet Hours, Loud Data was useful, though the ICT examples assume you can actually change send windows in the ESP. We could not, so we used the worksheet as a briefing for engineering instead.”

Client in ride-hailing · Chiang Mai

Programmes

Four ways into the same problem.

Every course stays inside push notification performance. We do not dilute the week with generic “growth hacking” or email journeys.

Person reviewing a mobile app

Flagship · 12 weeks

Signal to Send

Event taxonomy, delivery versus influence, token health, and incrementality you can run without a dedicated analyst.

Syllabus and seats

Laptop with performance charts

Workshop · 3 days

The Delivery Funnel

Map accepted, delivered, rendered, and influenced on one live campaign. Leave with a funnel sketch your ESP can actually fill.

See all programmes

Pedagogy

What we refuse to count as “performance”.

01

Open rate without a clock

A tap at 02:10 ICT is not the same signal as a tap during a commute. We teach time-to-open against local quiet hours before we celebrate a percentage.

02

Reach that includes dead tokens

Audience size that still contains expired device tokens inflates every downstream metric. Token hygiene is a performance topic, not an engineering footnote.

03

Sequences judged in isolation

A “win-back” series that steals conversions from an already-scheduled receipt message is not lift. Incrementality sits in the flagship, not in an appendix.

04

Dashboards nobody can argue with

If product, CRM, and finance cannot share a definition of influence, the chart is decoration. We write definitions before we build views.

Operators collaborating around a table

The room

A cohort small enough to bring your live app.

Apieventgrid runs from a desk in Phra Nakhon Si Ayutthaya and a studio day in Bangkok when a cohort needs whiteboards. You work on your own notification catalogue, not a fictional coffee brand.

Nalinee Srisuk leads Signal to Send. She spent nine years reading CRM exports before she started teaching the same worksheets to operators across Thailand.

How the lab is run

Journal

Field notes we actually assign.

Open rate is not the story

Why a rising tap percentage can hide a collapsing render rate, and what to put on the first slide instead.

All journal pieces

Next intake

Bring one live campaign. Leave with a definition you can defend.

Seats are limited because every operator arrives with a real catalogue. If you are unsure which programme fits, write to the desk with a screenshot of your current push report.

Write to the desk