Centeem in numbers: what we learned during the private beta
A transparent review: users, volumes processed, bugs encountered, tester feedback, feature requests. What these 3 months taught us about the Algerian market and what we're going to change.
When you launch a fintech, the worst thing you can do is lock yourself away for 12 months building in a corner, then open to the public with your fingers crossed. The private beta exists to remove that risk — to confront what we've built with real people and their real usage.
Our private beta lasted 3 months (February → April 2026) with 200 invited testers. Here are the raw figures, what we understood, and what's going to change for the summer public beta.
The raw figures
Tester profiles: 78% individuals, 15% pro/professionals, 7% tourists (visitors invited at tech fairs). Wilayas represented: 14 out of the 69 (concentration in Algiers / Oran / Constantine, normal for an invite-only private beta).
Top 5 observed use cases
- P2P between Centeemers (52% of tx) — Families, flatmates, friends reimbursing a shared expense. Usage explodes as soon as the 2nd Centeemer in the close circle signs up (classic network effect).
- BaridiMob top-up (24%) — Dominant among individuals vs CCP (young people don't like the post office). BaridiMob = top up from home, no trip needed.
- Professional QR payment (11%) — Across the 30 beta professionals, top usage among grocers and barbers. PayLink used less than expected (lack of a clear use case in beta).
- BaridiMob withdrawal (8%) — Regular usage but limited by the beta cap. Confirms the importance of this channel for the general public.
- Savings account (5%) — More testers than expected created a goal (holiday, wedding, car). A positive surprise — we're going to prioritize this feature.
Top 5 bugs encountered (full transparency)
We won't lie, we had plenty of bugs. Here are the 5 most impactful ones, and what we did:
BaridiMob withdrawal → "Server error"
22% of withdrawals were failing for 2 weeks with a generic message. Root cause: Prisma transaction timeout due to a fee op outside the transaction. Fix deployed, no cases since. See changelog.
CIN reading crashing on Xiaomi
8% of Xiaomi testers had contactless reading that wouldn't read the CIN. Specific hardware. Workaround: fallback photo mode (front+back). Permanent solution in progress.
CCP receipt wrongly rejected
The OCR was rejecting perfectly legible receipts with a "photo blurry". Cause: quality threshold too strict, while Gemini was extracting correctly. Fix: we delegate the decision to Gemini, no more blocking on the quality side.
Bio button 1-tap doesn't work
On Samsung in particular, the 1st tap on FaceID/fingerprint didn't trigger the prompt. Cause: known expo-local-authentication bug. Auto-retry workaround, resolved.
Excessive lazy load
Several screens (Withdraw, Top-up, Send) showed a spinner on every return, even though the data hadn't changed. React Query migration + MMKV cache → screens now open in under 100 ms.
🔧 Transparency policy: all our bugs are listed in our public changelog. We prefer to own up + show what we do to fix them, rather than pretend everything is perfect.
What testers asked us for
We collected feature requests via: (1) an in-app "Suggest an improvement" button, (2) a private beta tester Slack, (3) an individual 30-min call with 30 testers of their choice.
Top 5 requests:
- Virtual cards for online shopping (Amazon, AliExpress, etc.) — Top request by far. We have an active waitlist, planned post-licence.
- Incoming international transfers (receiving from abroad via SWIFT/IBAN) — A real expectation from the diaspora, we're working with our partner bank.
- Multi-person payment request (1 bill split between 4 friends) — Implemented in April, already available.
- "Dark" mode (dark theme) in the app — Implemented in March, we made it follow the system default automatically.
- Notifications on Apple Watch / Wear OS — V2 roadmap, public beta first.
What surprised us
3 insights we hadn't anticipated:
Savings, used more than expected
We thought the savings account would be a nice-to-have. Wrong. 32% of testers created a goal, and several told us it was THE feature that makes them recommend Centeem to their close ones (vs just "one more payment app"). We're going to double the product investment on this part.
Professionals want to "share" the QR
Several small merchants asked us to be able to send their QR Pay by WhatsApp to a customer who isn't physically at the shop. We discovered it was almost the same use case as PayLink, but in a different mental form. We're going to clarify it in the UI.
The diaspora wants to "send to family"
More than 15 testers spontaneously asked us: "how do I send €200 from France to my father in Algeria?". Bingo. The SEPA → DZD bridge is our biggest project for 2026-2027. Several European banks are already watching us.
What we're changing for the public beta
- Rewritten KYC onboarding with better visual instructions (where to place the CIN, selfie posture, etc.)
- In-app tutorial on first login (4 illustrated screens explaining the key uses)
- Follow-up notifications that are proactive (e.g. "Your KYC is validated, you can get started!")
- Agent network x10: we go from 30 beta professionals to 3,000 planned for the public launch
- 24/7 support in FR / AR / EN (was 9am-6pm in beta)
- Savings account v2 with shared goals (family / couple)
🙏 Thank you to the 200 testers
Without you, Centeem would not be what it is today. Every bug reported, every feature suggested, every 30-minute call moved us forward. You all keep your beta account with a special "Early Adopter" badge for life. We'll keep you informed first about upcoming features.
Public beta: summer 2026. Download the app now to reserve your spot and be notified as soon as it opens.