Every outage investigation should start with the same question, and it is not “what is broken?” It is: broken for whom? A website can be perfectly healthy for ten million people and dead on your phone, or genuinely down worldwide while your neighbour’s cached copy still loads. The checks below take about a minute, in order, and each one rules something out.
The sixty-second check, in order
1. Try one other website. If nothing loads at all, your connection is the story, not the website. If everything else works, move on.
2. Switch networks. Turn off Wi-Fi and load the site on mobile data, or the reverse. Home broadband faults, office firewalls and captive hotel portals all block specific sites in ways that look exactly like an outage. A site that loads on one network and not the other is not “down” — one of the two paths to it is.
3. Ask a status page. Check the service here, or on any outage tracker. Real outages produce a visible spike of reports within minutes because thousands of people hit the same wall at the same time. A flat line of reports while the site fails for you is strong evidence the problem is local.
4. Run an independent check. Use a website checker tool (we have one) that fetches the site from a server somewhere else. If that server also cannot reach it, the problem is at the site’s end, and nobody’s settings can fix it.
5. Try a second device or a private window. A private (incognito) window has no cache, no cookies and usually no extensions. If the site loads there, your normal browser is carrying something stale or broken — and you have found the problem without touching the website at all.
What “down” actually means
“Down” is not one failure. A site can refuse connections entirely (server off, network cut), answer with a 5xx error (server alive, application broken), load its home page while payments fail (one subsystem down), or work nationally except in your state or on your provider (a routing or CDN fault). This is why two people can honestly disagree about whether a service is down: they are testing different parts of it. When you report a problem, the useful detail is which action failed — login, payment, upload — not just “it’s down”.
The problems that look like outages but are at your end
Stale cache and cookies. Sites change their login systems and keep your old session. A private window test exposes this instantly; clearing cookies for that one site fixes it.
VPN and private DNS. Many Indian banking and government sites block or break for traffic that appears to come from another country, and private-DNS apps (the ones that promise ad blocking) quietly break banking domains. Switching them off is the single most common “fix” for a site that is supposedly down.
Wrong device clock. Login systems and OTP screens depend on accurate time. A phone clock set manually and drifting by a few minutes can make secure sites refuse you.
An old app version. Apps lose server support quietly. The site is fine; your six-month-old app is knocking on a door that moved.
Your provider’s DNS. If sites load by mobile data but not on your broadband, and several unrelated sites fail, suspect the broadband provider’s DNS and try the connection again after restarting the router.
The payment trap
One warning deserves its own section: when a payment fails, do not immediately pay again. A payment that shows as failed may have left your account and be travelling; paying twice creates two debits and a much slower refund. Check the verdict on the service’s status page, check your transaction history, and give the automatic reversal system its window first. Our guide on failed UPI payments explains the timelines in detail.
A quick decoder for what the site shows you
Sometimes the site is not down or up but talking in codes, and the code narrows the fault considerably. A 404 means the server is alive but that exact page is gone — check the address before blaming an outage. A 403 means the server refused the request, which for Indian banks and government sites usually means a bot filter, a VPN, or a region block: the site is up, it just will not talk to you this way. A run of 500-series errors is a genuine fault inside the service. A page that loads but shows an endlessly spinning login is often a broken third component — the part that checks your password — which is why the home page can look healthy while nobody can sign in. And a warning about an expired or invalid security certificate is your browser protecting you: never click through it on a banking or payment site, however urgent the task feels. Note which of these you saw, because “it shows a 403 on my broadband but works on mobile data” is a diagnosis, while “it doesn’t work” is only a symptom.
When it really is down
Sometimes the answer is simply that the service is broken for everyone. Then the routine changes: stop retrying (repeated attempts can lock accounts and duplicate payments), note any reference numbers you already have, check the status page occasionally rather than constantly, and use the official helpline on the service’s helpline page here if your matter is urgent. Genuine large outages in India usually trend visibly within minutes and clear within a few hours; the exceptions make the news.
The minute of checks in this guide will not fix a single server. What it fixes is the hour people otherwise spend restarting routers and reinstalling apps during an outage that was never theirs — or blaming a website for a problem sitting in their own pocket.