Our website keeps falling over.

You know the message by heart now. Is the website down for anyone else? Sometimes it is a colleague, sometimes a customer, occasionally, worst of all, your boss, and it always arrives before any alert does, because there is no alert. You open the site on your phone with that particular held breath, and either it loads and you exhale, or it does not and the afternoon reorganises itself around a crisis you have lived through before.

The pattern is what wears you down. If the site failed once, catastrophically, someone would fix the cause. Instead it fails a little, often. Down for ten minutes on a Tuesday. Achingly slow whenever the newsletter goes out. A checkout error that three customers mention and nobody can reproduce. Each incident is small enough to survive and vague enough to close without a real answer, and the file of unexplained restarts quietly thickens.

Asking for help produces the loop you have also memorised. The hosting company checks and reports the server is running, which is technically true and completely useless, since running is not the same as serving your website. Whoever last touched the code suspects the hosting. The hosting suspects the code. You sit in the middle of a supplier stand-off, holding a problem that is officially nobody's, translating between two parties who each stop at the exact boundary where your outage actually lives. Meanwhile you have started flinching at your own marketing: every campaign, every busy season, every burst of the traffic you work hard to attract now carries the private question of whether the site will survive its own success.

Here is what your experience has already taught you, even if nobody has said it plainly: intermittent failures are not bad luck, they are a symptom with a cause, and the cause is findable. The reason yours has not been found is not that your site is mysterious. It is that nobody with the skills to read the whole system, application and infrastructure together, has ever been made responsible for it end to end.

How we stop the cycle

Diagnosis first, properly. Our audit and performance service examines the site the way the stand-off never does: code, database, hosting and configuration as one system. Intermittent failure has a short list of usual suspects, resource exhaustion under load, greedy database queries, memory leaks that build until a restart, misconfigured caching, an underpowered or oversold server, and the evidence for each is right there in logs and metrics once someone actually correlates them with the incident times. You get the findings in plain English, ranked by impact, with before-and-after measurements as fixes land. If slowness is part of your pattern, our FAQ on slow Umbraco sites shows how consistently these causes repeat.

Then the structural fix, which is ownership. The reason your outages last as long as they do is the gap between suppliers, so we close it: on our managed hosting, from £99 a month, the same accountable team runs the infrastructure and understands the application. There is nobody to hand the problem to, which is precisely the point. Monitoring means we typically know the site is struggling before you do, and the message you have memorised gets replaced by one from us, saying what happened and what we did about it. Backups are verified with real test restores, security patches go on within 48 hours of release, and the full picture of what a failure looks like when it is handled properly is in what happens if our website goes down.

For sites where the website is genuinely business-critical, a retainer adds reserved specialist time each month, from £500, with the typical first fix inside 18 hours; fragile sites usually carry a backlog of small hardening work, and the retainer is how it steadily gets done rather than perpetually deferred. None of this requires a rebuild, and we will not pitch one unless the evidence demands it, which in our experience it rarely does. Most falling-over websites are sound sites with one or two specific faults and no owner.

The cycle ends when someone is accountable for the whole system and the cause gets named. Tell us what the failures look like, when they cluster, what the host said, what has already been tried, and we will tell you what we would investigate first. You have already done your tour of duty refreshing the site on your phone; that job is better held by monitoring, and by us.