What you can expect when something breaks
This statement says what we commit to, how we will tell you when the service is not working, where you can check it yourself, and — because it matters more than the rest — what we deliberately do not promise.
What we commit to
We do not offer a contractual uptime guarantee, and we are not going to publish a number we cannot stand behind. A percentage on a page is worth nothing without the measurement behind it and a remedy when it is missed, and we have neither yet. What follows is what we actually do.
We aim for the application to be available during UK business hours, Monday to Friday, and in practice it runs continuously. It is a single web process and a single background worker on managed infrastructure in Ireland, with the database and uploaded documents in the European Economic Area and the United Kingdom respectively.
We are a single-developer organisation. There is no out-of-hours rota and no second engineer. An incident raised at nine on a Tuesday morning will be seen quickly; one raised at midnight on a Saturday may not be seen until the morning. We would rather say that than let you infer a support desk that does not exist.
Planned changes
We deploy during the working day rather than overnight, because a deploy nobody is awake to watch is the one that goes wrong quietly. A deploy normally causes no interruption: the new version starts before the old one stops. Where a change needs the database altered in a way that could interrupt you, we will tell you beforehand by email and carry it out outside business hours.
How we tell you when something is wrong
We do not run a status page. For a service of this size a status page tends to be the last thing updated during an incident, and a stale green tick is worse than no tick at all.
Instead:
- If the problem affects your data or your records, we email you directly. Not a banner you might not see — an email to the address on your account.
- If the problem is security-related, we follow the process in our incident response procedure: the Information Commissioner’s Office is notified within 72 hours where the breach is reportable, and the Information Commissioner's Office within 72 hours where personal data is involved.
- For anything else, the service itself will tell you. Where an action cannot complete we show you what failed and what to do about it, rather than a generic error.
Checking for yourself
The application publishes a health endpoint that answers 200 when it is running and able to serve requests, and an error status when it is not:
https://accounts.dii.ltd/up
It is deliberately simple. It tells you whether the application is up; it does not tell you whether HMRC is up, whether your bank's provider is responding, or whether a particular filing succeeded. For those, the relevant screen inside the application is the honest answer, because it reports what actually happened to your data rather than what a monitor thinks.
Reporting a problem
- Something not working: support@dii.ltd
- A security risk or incident: legal@dii.ltd, or the machine-readable contact at
https://accounts.dii.ltd/.well-known/security.txt
Both reach the same person, who is the responsible individual named in our regulatory filings. There is no triage queue in between.
What we do not promise
Recorded here rather than left for you to discover:
- No uptime guarantee and no service credits. See above.
- No 24/7 support. One person, UK hours.
- No guaranteed response time to a non-security report, though in practice it is the same day.
- No control over the services we depend on. Hosting, payments, document email and bank data all sit with other providers, and when one of them is down no amount of engineering on our side changes that. We will show you HMRC's own answer rather than hiding it behind ours.
- No independent audit of these arrangements. Nobody outside this organisation has reviewed them.
If any of this is not enough for your circumstances, tell us before you rely on the service rather than after.