Legal
Support and uptime
What we commit to, what we merely aim for, and what we do when we miss. Written to be checkable rather than impressive.
Last updated 14 August 2026
1. An honest starting point
This is a target, not a warranty
We would rather publish a number we hit than a number that sells. A marketing 99.99% from a team of our size, on the infrastructure we run today, would be a fiction — and you would find that out during your first incident rather than on this page.
2. Our availability target
- Target: 99.5% of each calendar month, measured on the UbiLearn API and web application.
- What that permits: about 3 hours and 39 minutes of unavailability in a 30-day month, including planned maintenance.
- How we measure it: an external uptime monitor polls our health endpoints from outside our network at short intervals. “Unavailable” means the service returns errors or fails to respond to those checks, not that a single request was slow.
- What we do internally: alerts fire on error rate, latency, queue depth, failed payments and failed webhooks, and reach a human rather than a dashboard nobody is watching.
3. Other things we measure
| What | Target |
|---|---|
| API response time, 95th percentile | Under 300 ms |
| Time for a video lesson to start playing | Under 3 seconds |
| One-time-code email delivered | Under 60 seconds |
| Largest contentful paint on a learner page (4G) | Under 2.5 seconds |
| Database backup | Nightly, with restores tested |
These are engineering targets we hold ourselves to, on the same footing as the availability number: measured, reviewed, and not contractual promises.
4. What the target excludes
- Planned maintenance announced in advance, as described below.
- Failures of third parties we depend on and cannot control — our cloud provider, the payment gateway, email delivery networks and the public internet between you and us.
- Problems caused by your own network, device, browser or firewall.
- Suspension of a workspace for non-payment or for a breach of our acceptable-use terms.
- Degradation caused by something you did — an unusual bulk import, an automated script hammering the API, or content that breaches our upload limits.
- Force majeure events beyond our reasonable control.
5. How we handle incidents
- Alerts page a human directly. Someone starts investigating rather than waiting for a customer email.
- For an incident that makes the service unusable for multiple customers, we email workspace owners with what is happening and what we know, and we update them as it changes rather than going quiet.
- We do not close an incident by declaring it fixed; we close it when the underlying cause is understood.
- For any incident that caused a significant outage, we write a plain-language post-incident summary — what broke, why, what we changed — and send it to affected workspace owners within five working days.
- Rollback is always available: we deploy versioned images and can return to the previous release quickly, and database changes are designed to be compatible with the release before them.
6. Planned maintenance
- Most changes ship with no downtime and no announcement, several times a week.
- Where maintenance will be disruptive, we give at least 48 hours' notice by email to workspace owners.
- We schedule disruptive work between 11 pm and 5 am IST on weekends, when Indian learners are least likely to be studying.
- Emergency maintenance to fix a security issue may happen without notice. We will tell you afterwards what we did and why.
7. Support channels and response times
Support is by email. There is no ticket maze and no tiered call centre — messages reach the people who build the product.
| Plan | First response target | Included |
|---|---|---|
| Starter | Next business day | Email support and full documentation |
| Growth | Within 8 working hours | Email support plus a 30-minute guided onboarding session |
| Business | Within 4 working hours | Priority email support, a named contact, and a 90-minute onboarding session |
- Support hours are Monday to Friday, 9:30 am – 6:30 pm IST. Messages sent outside them are answered the next working morning.
- Anything that has taken a workspace offline is treated as urgent on every plan, including Starter, whatever the hour.
- Write to support@ubilearn.com from the email address on your workspace, and include your subdomain — it saves an entire round trip.
8. How we prioritise
| Severity | What it means | What we do |
|---|---|---|
| Critical | The workspace is down, learners cannot sign in, or payments are failing for everyone. | Worked continuously until resolved, with updates as they happen. |
| High | A core function is broken — video will not play, quizzes will not submit, certificates will not issue — with no workaround. | Worked in business hours until resolved, with a daily update. |
| Normal | Something is wrong but there is a workaround, or a question about how to do something. | Answered within the plan response target and scheduled into the next release. |
| Low | Cosmetic issues, feature requests and suggestions. | Acknowledged, logged, and weighed against everything else people are asking for. |
9. Backups and recovery
- The database is backed up nightly to encrypted storage, with 30 days of retention.
- Restores are tested — a backup nobody has restored is not a backup.
- Our current recovery point objective is 24 hours, which means a worst-case restore could lose up to a day of data. We are honest that this is a starting position and it improves as we grow.
- Uploaded media is stored redundantly by our cloud provider.
- Keep your own master copies of course videos and documents. Any sensible platform will tell you that, and we would rather say it here than in an apology email.
10. Reporting a security issue
If you believe you have found a vulnerability, email support@ubilearn.com with “Security” in the subject line and enough detail to reproduce it. We acknowledge within one working day and keep you updated until it is resolved.
- Please give us reasonable time to fix an issue before disclosing it publicly.
- Please do not access, modify or delete data belonging to another customer while testing, and do not run tests that degrade the service for others.
- We will not pursue legal action against researchers who follow these two rules and report in good faith.
11. Contact
Support: support@ubilearn.com · Sales and contractual SLAs: sales@ubilearn.com ·
This page sits alongside our terms of service, which set out the contractual position, and our privacy policy. Where this page and the terms differ, the terms govern.