Upinion · Pilot — Channel → wa.me → Flow · End-to-end test

Test the architecture on 10 real users, before we point it at Algeria.

We validated the wa.me Channel pattern with Gerben's own phone (Day 2 Istanbul). Now we validate it with real, non-technical members of an existing Upinion community — to find the rough edges before we expose the pattern to a vulnerable population where breakage costs trust.

Owners: Rik & Julian (build) · Noor (community outreach) · Gerben (review) · Timeline: ~10 working days

Goal

Validate the Channel → wa.me → Flow architecture end-to-end on real WhatsApp users, with no surprises. Not Algeria yet — a friendly internal community we already trust.

1Why now — what this test answers

2Prerequisites — what must be true before Day 1

WhatOwnerStatusNotes
WhatsApp Business account (test number) Rik / Julian todo Cloud API tier. Either via Turn.io trial (recommended — Flows ready) or direct Meta sandbox.
Flow JSON deployed to Meta Flow Builder Rik / Julian ready Use upinion-demos/whatsapp/flow-specs/anwar-safety-survey-3q.json. Adapt question copy if needed for this audience.
Bot listener: keyword → Flow trigger Rik / Julian todo When user sends `START_SURVEY` in chat, server responds with an interactive Flow message. ~1 hour with Turn.io.
WhatsApp Channel created Noor (community manager) todo Manual creation (no API). 5 min in WhatsApp Business app. Name: 'Upinion Pilot' or similar — these are existing members, no masking needed.
List of 10 consenting community members Noor todo Pick existing engaged members from a non-Algerian community (e.g., Lebanon, Greece, Bangladesh). Mix of iOS + Android. Mix of engagement levels.
Topup integration — DEFERRED skip First pilot is functional only. Topup tests come after. Removes Reloadly as a blocker.
Pseudonym / microsite — DEFERRED skip Internal pilot with known members. Brand as 'Upinion'. Pseudonym tests come with the Algeria pilot.
Multi-language Flow — DEFERRED skip English only for this test. Multi-lang variants once the base loop is proven.

3The flow — step by step (what each user does, what we measure)

#WhoWhat happensWhat we measure
1 User Follows the Upinion Pilot Channel (Noor sends them the join-link via existing DM) Joined ✓ / not joined
2 Upinion Posts a single Channel message: "Hi — quick 3Q survey to test our new flow. Tap below 👉 https://wa.me/<bot>?text=START_SURVEY" Sent ✓
3 User Sees the channel post (push notification + Updates tab) Notification received ✓ / opened the post
4 User Taps the wa.me link Tap registered (server log)
5 WhatsApp Opens the bot chat with `START_SURVEY` pre-typed in the input Chat opened ✓ (server sees no message yet)
6 User Taps send Inbound message received by bot ✓
7 Bot Replies with an interactive Flow trigger message Flow trigger sent ✓ (Meta API ack)
8 User Taps the Flow CTA → Flow opens with Q1 Flow opened ✓ (Flow API webhook)
9 User Completes Q1 → Q2 → Q3, submits Each question answered (granular)
10 Bot Receives completion payload, sends thank-you message Completion recorded ✓

4Reach-out script (community manager — drop in your DMs)

Hi [name] — we're testing a new way to send surveys, and we'd love your help. It takes 2 minutes: 1. Tap to follow our pilot Channel: [link] 2. When our message arrives, tap the link inside it 3. Tap "send" when WhatsApp opens 4. Fill in the 3 questions, hit submit That's it. If anything breaks or is confusing, just reply here — that's the whole point of the test. Thanks for being on our test team. — Noor

Tone: warm, low-stakes, explicit "this is a test". The goal is to make breakage safe to report.

5What success looks like — the metrics that matter

MetricTargetWhat it tells us if missed
Channel follow rate ≥ 8 / 10 If <8, the join flow is the blocker.
wa.me link tap rate (of followers) ≥ 7 / 10 If <7, the CTA copy isn't compelling enough.
Keyword send rate (of tappers) ≥ 6 / 10 Two-tap UX. If users don't send, the pre-typed keyword is confusing.
Flow open rate (of keyword senders) ≥ 6 / 10 If <6, Flow trigger render is broken on some devices.
Flow completion rate (of openers) ≥ 5 / 10 If <5, the Flow itself has friction (question count, copy, UX).
End-to-end completion (channel → submit) ≥ 5 / 10 Hard floor for going to Algeria pilot. Below this, fix before scaling.
Median time-to-complete < 3 min If much longer, the Flow is too heavy.

6Known unknowns — watch for these

7Timeline

WhenOwnerDeliverable
This week (now-Fri) Rik / Julian Bot listener + Flow JSON live on a test number. Internal end-to-end test passes.
Next week (Mon-Wed) Noor Channel created. 10 members invited via DM. Reach-out script sent.
Next week (Thu-Fri) Pilot users Run the flow. Server logs every step. Noor follows up with each user.
Following Monday Whole team Review metrics, qualitative feedback, decide: go/no-go on Algeria pilot scope.

8Quick links — what you need to start