Savane Info
Landing-hub maximalist
Trust via credibility
Polished editorial portal builds trust. Migrants engage because the site feels established.
⚠ Reads as a publisher, not a place to be heard.
Open demo →Upinion · Web App 2.0 · Brainstorm
Five bets on how the next Upinion respondent surface should feel.
We're rebuilding the answering platform for "Digital voices for dignity" — a remote, anonymous engagement project for sub-Saharan migrants and Touareg communities in Algeria.
Today's job: walk the five demos, argue what's good and what's painful, and decide which bets (or hybrids) move into the production build.
1Why this matters — the Algeria reality
16,000+
migrants expelled Apr–Jun 2025 alone
Algeria → Niger 'Point Zero'
5–6 mo
average time in Algeria before deportation
Marwan, 2026-05-13
0
physical Upinion presence on the ground
fully remote project
6
languages required at launch
Tamashek · Hausa · Bambara · Arabic · English · French
2The shift — old flow → Web App 2.0
Today
- Facebook ad
- Facebook Messenger
- Onboarding (mandatory before any answer)
- Q → A → Q → A (bi-directional)
- Information message back
- (end)
Bi-directional but narrow. Q&A only — no testimonies, no peer content, no service requests, no hub. Mandatory Messenger onboarding gates everything. Hard to return after closing the chat.
Web App 2.0 — what we're testing
- Facebook ad
- Web App (language first)
- Hub: testimonies · news · support requests
- Fork: Onboard OR answer-once
- Conversation (voice + text)
- Topup incentive · return to hub
User decides depth. We capture data either way. Trust built BEFORE any account commitment. Hub gives the community a reason to come back.
3The five demos — each tests one bet
SOUT
Minimal funnel — brutalist
Completion via simplicity
Radical minimalism. One screen, one ask. Highest completion rate.
⚠ Too sterile. Low ambient trust.
Open demo →Amawal
Touareg / Upinionator
Trust via cultural specificity
Design from the audience outward. Visually speaks to Touareg communities.
⚠ Doesn't scale to other geographies without redesign.
Open demo →Connexa
App-first PWA
Retention via permanence
Installable shell. App icon on phone = stickier than browser history.
⚠ Install friction. Data cost is a luxury (Marwan).
Open demo →Connexa
Omni-channel hub
Conversion via familiarity
Meet them where they already are — WhatsApp, Messenger, Telegram, web.
⚠ Per-channel engineering cost. WhatsApp can't be advertised to.
Open demo →4Non-negotiable must-haves
🎤
Voice + text everywhere
Literacy varies. Some dialects aren't written.
🌐
Language picker first
Tamashek, Hausa, Bambara, Arabic, English. NOT French-first.
📡
Low bandwidth
Internet credit is a luxury. Test before launch.
🎭
Pseudonymous brand
No Upinion. No NL Embassy. No partner names anywhere.
🔒
No PII collected
Private mode. Close-windows warning after session.
⭐
Star-rate + delete answer
Existing features. Non-negotiable.
💸
Topup incentive
Reloadly integration. Tied to answering, not signup.
🧩
Single + multi + text + conditional
Minimum question types. Trim everything else.
5Open questions for today
Web app · native PWA · or WhatsApp / social messaging?
- Web app — lowest friction at top-of-funnel. Just click the ad. But hard to return: browser history is unreliable.
- PWA — installable shell, app icon on the home screen, works offline. But install is a barrier, and data cost is a luxury (Marwan).
- WhatsApp / social messaging — meets them where they already are. Familiar surface, push notifications work. But: WhatsApp can't be advertised to directly; we'd need Upinionators or Facebook to drive in. And running over a third-party network = more surveillance surface.
- Hybrid: web app for first touch (zero friction), nudge to PWA install after first conversation completed, WhatsApp/Telegram routing for active community members?
Tested in: Demo 1/2/3 (web) · Demo 4 (PWA) · Demo 5 (omni-channel routing)
Drop-out — should we measure it, and how do we define it?
- Today we lose people silently — we don't know where in the funnel they fall off. We can't optimise what we can't measure.
- Suggestion — instrument 6 funnel checkpoints (anonymous, no PII): (1) ad-click reached landing (2) language picked (3) first hub action (testimony read / support request / start convo) (4) first conversation completed (5) returning session within 7 days (6) onboarded into community.
- All counters server-side, aggregated, anonymous. Each demo emits the same 6 events so we can compare across designs.
- Plus qualitative signal: a 1-question 'why did you leave?' optional prompt on the close-windows / private-mode screen.
- Define 'drop-out' as session-level (one ad → no answer = drop-out at step 3) AND cohort-level (% of step-1 users who reach step 4 within 24h).
Tested in: Every demo — instrument identically so the data is comparable
How do we give back? What's in it for the community?
- Topups (Reloadly) — tied to answering conversations, NOT to signup. Already validated as feasible in Algeria pending phone-provider test.
- Verified information — anti-trafficking awareness, safe-route updates, hotline numbers, counter-narratives to disinformation circulating in local groups.
- Testimonies / peer stories — moderated repository where migrants can read what others have been through. Increases trust + engagement, builds 'this place is for us'.
- Routed support — light-touch psychosocial / legal support via RTE (real-time engagement) with providers based in the Netherlands or remote. Not a service we deliver — a service we connect them to.
- Voice to power — RTEs with embassies and policy stakeholders. We promise this in the proposal. The respondent's voice is heard, not just stored.
- News for migrants — curated updates on routes, border closures, asylum policy changes that affect their journey decisions.
Tested in: Hub feature on every demo — what's IN the hub is the bet
6Decisions we need before we leave the room
- Which demo (or hybrid) takes the team's confidence into a production build?
- Web · PWA · WhatsApp — pick one as the spearhead, or commit to a hybrid?
- Sign-off on the 6-checkpoint drop-out instrumentation across all demos.
- Pick the launch hub: which 'give back' features go in the MVP — and which wait?
- What's the first thing they see after they click the ad? (Animation? Language picker? Hub?)
7Hard constraints (apply to every demo)
- Mobile-onlyMax 480px viewport. Desktop optional.
- Anonymous networkVerifiable 'black box' — no trace back to Upinion or NL Embassy.
- Reset?reset=1 clears localStorage.
- Private-mode exitClose-windows warning after every session.
- Restricted areasNo outreach near Assamaka / Point Zero / detention centers.