The first fifteen minutes

Website Down? The First 15 Minutes, Before Anyone Touches Anything

A customer texts a screenshot of your home page, except it is not your home page. Every instinct says log in and start clicking. Sit on your hands for a minute. Most outages have dull causes, and the order you check them decides whether this costs an hour or a week.

November 18, 2026 · 11 min read

acmeflowershop.com · emergency job log
One morning outage, worked in order
Down at 9:12, the cause was a plugin update
  1. 09:12 owner reports a white page; outside check fails too
  2. 09:18 domain renews in March; hosting invoice paid
  3. 09:21 host status page: no incident; no suspension email
  4. 09:26 full backup of files and database, broken as found
  5. 09:34 error log: fatal error in a gallery plugin updated 02:00
  6. 09:41 plugin rolled back; home page returns 200
  7. 09:55 tested checkout, contact form, phone link and login
Illustrative: sample emergency job on a fictional florist's website

The short answer

When a website goes down, first confirm it is down for everyone with an outside check, such as a free online status checker or a phone on cellular data. Then check the cheap causes in order: domain and hosting renewals, the security certificate, the host's status page and anything changed in the last day. Write down the exact error before anyone changes anything.

On this page
  1. Down for you, or down for everyone
  2. A fifteen-minute order of work for a dead website
  3. Reading the three-digit number on the error page
  4. What Google does while your pages are dark
  5. Six panic moves that stretch a two-hour outage into two days
  6. Jobs to handle yourself, and the point where you should call
  7. When you call us: response, price and what we test
  8. Hearing about the next outage before a customer does

Down for you, or down for everyone

Your office computer is the least reliable witness in the building. It caches old answers, it sits behind your router, and it may be the one machine a security rule has blocked. So the first test happens somewhere else. Put the address into our free is my site down checker, which tries the site from our server and reports why it failed. Then load it on a phone with Wi-Fi switched off. If both see a working site, your customers probably do too, and the problem is local.

If the outside check fails, look closely at what the screen shows, because each kind of failure points somewhere different. A parking page full of ads is a different emergency from a white page, and both differ from a padlock warning. The table below is a triage sheet, not a repair manual. Each row links the page on this site that walks through that one failure in detail, so you can skip straight to the one that matches.

What the broken page looks like, and where to start
What it usually points toRead next
A registrar parking page or adsThe domain registration lapsedDomain and DNS emergencies and domain expired
A notice from the hosting companyThe host switched the account offSuspended by the host
The browser cannot find the serverDNS records are wrong or missingDNS not resolving
A full-screen certificate warningThe SSL certificate expired or does not matchExpired SSL certificate
A blank white page on WordPressA fatal PHP error, often a plugin or themeWhite screen of death
A database connection messageThe site cannot reach its databaseDatabase connection error
A hosting platform's deployment errorA build failed or the domain points at a missing buildNext.js and React sites down

A fifteen-minute order of work for a dead website

This is the sequence we follow on an emergency call, written so an owner can do the first six steps alone. None of it needs a developer, and none of it can make things worse. It is ordered from cheapest to most expensive, because the cheap causes are quick to rule out and embarrassing to miss after paying someone to dig through server files.

  1. 1

    Check from outside the building

    Run the address through the outside checker, then try a phone on cellular data. If it loads out there, clear your browser cache, try another browser and suspect your own network before the website.

  2. 2

    Write down exactly what you see

    Take a screenshot. Note the time, the full address, the error text and any number on the page. That note saves the person fixing it the first ten minutes, and it shows whether the failure changes later.

  3. 3

    Check the three renewals

    Log in to the registrar and read the domain's expiration date. Open the hosting account's billing page. Click the warning or padlock to see the certificate's dates. A lapsed bill is a fix you can make yourself, today, for the price of the bill.

  4. 4

    Read the host's status page and your inbox

    Search your email for messages from the host and registrar: invoices, suspension notices, abuse reports. Then check the host's public status page. If it shows an incident, subscribe to updates and wait; nobody on your side can fix their data center. The same page often carries security notices when nothing is down, and how to read one during a patch week takes about ten minutes to learn.

  5. 5

    Ask what changed in the last day

    A plugin update, a theme edit, a deploy, a DNS change, a new host, a quick tweak by a well-meaning relative. If anything changed shortly before the failure, that change is the first suspect, and the person who made it is the first call.

  6. 6

    Copy the broken site before anyone repairs it

    If you have hosting access, ask the host for a snapshot or start a full backup of files and database as they are now. The broken state holds the evidence. A restore or reinstall on top of it erases the clue to what went wrong.

  7. 7

    Decide who fixes it

    If a renewal was the cause, pay it and give DNS and caches time to catch up. If not, send your note from step two to whoever will repair it. Our emergency website repair page lists the logins we ask for.

Desk at night under a red lamp, a laptop with a dark screen, its cable trailing off the desk, a coffee mug and a lined notepad
The first fifteen minutes are for looking and writing things down, not for clicking.

Reading the three-digit number on the error page

A number in the 500s means the server knows it failed. The Internet Engineering Task Force's HTTP standard, RFC 9110 as read in November 2026, says the 5xx class means the server "is aware that it has erred or is incapable of performing the requested method." That is useful in itself: the problem lives on the server or in front of it, not in your visitor's browser. Each code narrows it further, and each has a page on this site with the full walkthrough.

Server error codes as RFC 9110 defines them, with what they tend to mean on a small business site
CodeWhat the standard saysWhat it tends to mean in practiceWalkthrough
500The server "encountered an unexpected condition"Code or configuration crashed: a plugin, a PHP version change, a bad edit500 error
502A gateway or proxy "received an invalid response" from the server behind itThe CDN or proxy is fine; the application behind it is not answering properly502 error
503"temporary overload or scheduled maintenance"Stuck maintenance mode, an overloaded server or a host limit503 error
504A gateway or proxy "did not receive a timely response"Something behind the proxy is too slow: often a database query or a heavy page504 error
403The server "understood the request but refuses to fulfill it"Permissions, a firewall rule, or a host restriction after a suspension403 error

The Mozilla Developer Network's status code reference, as of November 2026, calls a 500 "generic," meaning the server could not find a more specific code. Treat a 500 as a sign to read the server's error log, not as a diagnosis. And if the number changes while you watch, from 502 to 504 and back, write that down too. A changing code usually means something is starting and crashing, and that pattern is worth telling whoever picks up the job.

What Google does while your pages are dark

Owners worry about rankings within the first hour, so here is what Google itself says. Its guide to HTTP and network errors, as of November 2026, says 5xx errors make its crawlers "temporarily slow down with crawling," in proportion to how many URLs fail. Pages already indexed "are preserved in the index, but eventually dropped" if the errors persist, and the page gives no number of days. Once the server answers normally again, Google "gradually increases the crawl rate."

In plain terms: an afternoon outage costs you sales and calls, not your search presence. Days are different. Google's page on reducing crawl rate warns that if Googlebot sees these codes on the same URL "for multiple days, the URL may be dropped from Google's index." If you must take the site down on purpose, say because it was hacked or is charging wrong prices, Google's guide to pausing a business online says to return an informational page with a 503 for 1 to 2 days at most, as of November 2026, and adds: "Don't return a 503 HTTP response status code for the robots.txt file."

Six panic moves that stretch a two-hour outage into two days

Every one of these feels like progress at the time. Each destroys information, adds a second problem on top of the first, or hands control to someone you have not vetted. We see the aftermath more often than the original fault.

  • Reinstalling the platform over the top. Reinstalling WordPress or a theme may bring the site back, but it overwrites the files that showed why it broke, and sometimes custom code nobody saved.
  • Restoring last night's backup without saving today. Orders, form entries and edits made since the backup disappear. And a backup nobody has ever restored is a hope, not a plan.
  • Changing nameservers or DNS records in a hurry. Cloudflare's TTL documentation, as of November 2026, says longer TTLs mean "updates to your records take longer to go into effect." A wrong change lingers, and undoing it takes just as long.
  • Deleting logs to free space or start clean. The error log is the only witness that saw the crash. Clearing it is the IT version of mopping a crime scene.
  • Pressing Update All to see if it helps. Ten updates at once turn one unknown into eleven.
  • Handing the admin password to whoever answers a social post first. A stranger with your hosting login is a second emergency. Give access to people you can identify, then change the password afterward.
The fastest repair starts with the slowest five minutes: reading the screen before touching it.

Jobs to handle yourself, and the point where you should call

We would rather you not pay us for a renewal you can click through yourself. Here is the honest split. If your situation sits in the left column, follow the steps above and save the money. If it sits in the right column, the clock matters more than the invoice.

Handle it yourself

  • The domain expired and you can log in to the registrar to renew it
  • The host's invoice went unpaid and the notice says so
  • The site fails only on your office network or one browser
  • The host's status page shows a known incident; subscribe and wait
  • The certificate lapsed and your host renews it with one button

Call someone now

  • Renewals are paid and the site still shows a server error
  • A white screen or database error on WordPress, and no tested backup
  • Redirects, spam pages or a browser warning: see the hacked website page
  • Pages load but checkout fails: see checkout and payments down
  • A custom application rather than a website: see web app down

When you call us: response, price and what we test

We respond the same business day; our hours are on the contact page. For WordPress, the fastest route is a Fix it now repair, paid online at a fixed price: site down or white screen is $249, on pages such as the white screen walkthrough and the 500 error page. We start within one business day of payment, with a full refund if we can't fix it. More WordPress help is on the WordPress repair page.

Every other platform, from a custom build to a hosted builder, is from $575, depending on complexity, with a written estimate before work starts; the emergency pricing page shows what moves the number. Before we call a site fixed, we test the paths that make money: checkout, forms, the phone link and login. The broader service is on fixing broken websites, and the section overview lists every kind of emergency we take.

Hearing about the next outage before a customer does

Most owners learn their site is down from a customer, which is the most expensive alarm there is. An uptime check is cheaper. Google Cloud's uptime check documentation, read in November 2026, describes the idea well: requests sent on a schedule "from multiple locations around the world," with an alert when they fail. Two details are worth copying in any tool. A check can require certain content on the page, which catches the white screen that still returns a normal status. And it can warn before the certificate expires.

Our hosting and maintenance plans start at $55 a month and, as that page states, include uptime and security monitoring, tested updates and off-server backups. For a free look at what might break next, run the website health check. DataCram is a San Antonio web studio, building for the web since 1999 and working with businesses across the US by phone, video and a shared project page. See our work or call (210) 346-0848.

Questions owners ask in an emergency

Probably not for customers. A phone on cellular data reaches the site through a different network and different DNS resolvers. If it loads there and on an outside checker, the fault is local: a browser cache, your office router, or a security rule that blocked your office address. Clear the cache, try another browser, then restart the router before calling anyone.

Google's documentation, as of November 2026, says server errors make it slow its crawling and that indexed URLs are kept but eventually dropped if errors persist. It warns that seeing errors on the same URL for multiple days may get it dropped. A few hours costs you visitors that day; the risk to search grows with days, so fix it quickly.

Not first. Take a copy of the broken site, then find out why it broke. A restore can bring back the same fault, erase orders and form entries made since the backup, and destroy the evidence of a hack. If the backup has never been tested, you may also find out mid-emergency that it is incomplete.

If the host's status page shows an incident, or you see a suspension notice, call the host. If renewals are paid, the host reports no problem and the site shows a server error or a white screen, call a developer. Send whoever you call a screenshot, the time it started and anything that changed recently.

For a WordPress site that is down or showing a white screen, our Fix it now repair is $249, paid online, with a full refund if we can't fix it. Other platforms and custom builds are from $575, depending on complexity, with a written estimate first. If the cause is an unpaid renewal, you only need to pay the renewal.

Yes, and Google describes how. Its guide to pausing a business online, as of November 2026, says to return an informational page with a 503 status for 1 to 2 days at most, and not to return a 503 for the robots.txt file. A holding page served with a normal status can end up indexed in place of your home page.

Keep reading

Free · No obligation

Tell us what the broken page says

Send the address, a screenshot and anything that changed in the last day. We reply the same business day with what we would check first and what it would cost.

Which of these sounds like you? (pick any)

Rather talk now? Call (210) 346-0848.

CallGet a quote