← Back to blog

7 Step Playbook for SMBs: Map Website Downtime Causes to Fixes

September 28, 2026
7 Step Playbook for SMBs: Map Website Downtime Causes to Fixes

Most website downtime traces back to five categories: human error, infrastructure or power failures, DNS and certificate problems, software or deployment issues, and attacks. When your site goes down, start with the fastest checks first: confirm the outage is global rather than local, check your host and CDN status pages, verify DNS resolution and certificate expiry, and correlate the timing with any recent deploy. Most causes resolve in minutes to a few hours once identified correctly.


TL;DR:

  • Most outages are caused by human error, misconfigurations, or recent changes, making processes like CI/CD gates and postmortems critical for prevention.
  • Power, network, and hardware failures account for about 45% of impactful outages, often external to your control and requiring reliance on provider status pages and SLAs.
  • DNS, TLS certificates, and domain issues can mimic a complete site outage; regular checks and dual providers help minimize this risk.
  • Server overloads due to resource limits or shared hosting issues frequently cause timeouts, which can often be prevented with proper capacity management and caching.
  • Implementing monitoring tools, automated backups, and considering managed hosting plans like inSave Hosting can significantly reduce the risk of downtime and streamline recovery.

inSave Hosting
Keep Your Website Online
inSave Hosting provides reliable, security-focused hosting with WordPress tools, free migration, and performance features for growing websites.
Explore hosting solutions

Table of Contents

Human error and change management failures

Misconfigurations, bad deploys, mistaken DNS edits, and credential misuse cause a large share of outages that never involve a hardware failure at all. Operators surveyed by the Uptime Institute report that better management and processes could have prevented many of their most recent incidents, which points to process gaps rather than equipment failures as the real culprit in a large share of downtime.

If your site went down right after someone touched the codebase or the DNS panel, that timing is your first clue. Check deploy logs and DNS change history before you look anywhere else.

  • Bad deploys: a code push introduces a config error or breaks a dependency.
  • Mistaken DNS edits: a typo in a record sends traffic nowhere or to the wrong server.
  • Credential misuse: shared logins or overly broad permissions let one mistake take down production.
  • Manual server changes: an untested change to a live configuration file breaks the web server.

Reducing this category means gating changes rather than trusting individual judgment: CI/CD pipelines that require review, pre-deploy checks against a staging copy, written runbooks for common changes, and permission limits so junior staff cannot edit production DNS or server configs directly. Postmortems after every incident, even small ones, close the loop.

Pro Tip: Keep every deploy reversible in under five minutes: tag releases, keep the previous build ready to redeploy, and never treat a rollback as a last resort.

Infrastructure and third-party outages: power, hardware, and network failures

Power and network failures tend to take down everything at once rather than one page or feature, which is the fastest way to tell them apart from a software bug. A failed uninterruptible power supply, a generator that does not kick in, or a failed disk array in a data center produces a sitewide outage with no application errors in the logs, because the server itself is unreachable.

Network and fiber issues look similar from the outside but originate outside the building: a carrier outage, a cut fiber line, or a bad route update can knock out an entire region. Signs include a mismatch between your own monitoring and your host's public status page, unusual traceroute results that die at a specific hop, and other sites on unrelated networks reporting similar issues in your area.

The Uptime Institute's annual analysis finds that power-related incidents remain among the leading causes of impactful outages, accounting for roughly 45% of impactful outages in 2025, with network failures close behind.

About 45% of impactful data center outages in 2025 were power-related, according to the Uptime Institute's outage analysis, which means a large share of the worst outages have nothing to do with your code.

When the problem sits with your host, carrier, or upstream provider, you are working within their service level agreement rather than fixing anything yourself. That reality shapes what you can actually do:

  • Check third-party status pages first: your host, CDN, and DNS provider all publish incident pages that update faster than support tickets.
  • Read the SLA before you need it: know what compensation and response time you are actually owed.
  • Keep a runbook for provider coordination: who to contact, what information to have ready, and how to communicate the outage internally while you wait.

For anything mission critical, multi-region hosting and redundant network carriers reduce the blast radius of a single failure, though that level of redundancy is usually reserved for larger operations rather than a single small business site.

DNS, domain, and TLS certificate problems that mimic total outages

An expired domain, an expired TLS certificate, or a broken DNS record can make a perfectly healthy server look completely dead to every visitor. These are also some of the fastest problems to diagnose once you know where to look, and Cloudflare specifically flags expired domains and expired certificates as common, easy-to-fix causes of full-site outages worth checking first.

A certificate problem shows up as a browser security warning rather than a blank page, while a DNS problem usually shows as "server not found" even though the origin server is running fine. Because DNS sits in front of everything else, a single misconfigured or expired DNS setup becomes a single point of failure for the whole site, which is why Cloudflare recommends dual authoritative DNS across two providers so one outage does not take resolution down entirely.

  1. Run dig yourdomain.com to confirm the domain resolves to the expected IP address.
  2. Run whois yourdomain.com to check registration and expiry dates.
  3. Run openssl s_client -connect yourdomain.com:443 to check certificate validity and expiry.
  4. If the domain or certificate has expired, renew immediately through your registrar or certificate provider and allow time for propagation.

A quick certificate check with these same commands can rule this cause out in under two minutes.

Pro Tip: Set low TTLs before any planned DNS change so a mistake resolves in minutes instead of a full day, and turn on auto-renew for every certificate you manage.

DNS routing and TLS certificate flow

Hosting limits, server overload, and resource constraints

A site that was fine yesterday and is timing out today has often simply outgrown its hosting plan. Shared hosting means sharing CPU and memory with other accounts on the same server, so a noisy neighbor's traffic spike can slow your site down even though nothing on your end changed. A single server with no redundancy is also a single point of failure by design: if that one machine hits its resource ceiling, everything on it queues or fails.

  • Watch for 503 Service Unavailable errors, which usually signal the server is overloaded rather than broken, since 500-level codes point to server-side problems and 503 specifically points to capacity or maintenance.
  • Check your host's dashboard for CPU, memory, and bandwidth usage against your plan's limits before assuming it is a code problem.
  • Add caching or a CDN if the bottleneck is repeated requests for the same content rather than genuine growth.
  • Move to a VPS or dedicated resources when traffic has outgrown shared hosting permanently rather than during occasional spikes.
  • Rule out administrative causes: a suspended account from a missed payment produces the exact same symptoms as a technical overload.

Application bugs, bad deploys, and CMS or plugin failures

Error codes are your fastest clue here. A 500 error usually points to application code or configuration, a 502 or 504 suggests the gateway or CDN cannot reach your origin server, and both patterns are documented in detail in standard HTTP error references. If you run into a 502 Bad Gateway specifically, the fix usually starts at the origin server rather than the CDN in front of it.

  1. Check whether the outage started right after a deploy, and if so, look at the deploy timestamp against your error logs first.
  2. Reproduce the error in staging before touching production again.
  3. For CMS sites, disable the most recently updated plugin or theme, or reset a broken .htaccess file, to isolate the cause.
  4. Roll back to the last known good version rather than trying to patch forward under pressure.

Smoke tests after every deploy, canary releases to a small slice of traffic first, and feature flags that let you disable a broken feature without a full rollback all shrink the time between a bad deploy and a full recovery.

Security incidents and volumetric attacks

A sudden traffic spike from unfamiliar geographic regions, combined with the site slowing to a crawl or falling over entirely, is the classic signature of a DDoS attack. A hijack looks different: unexpected changes at your registrar or DNS provider that you did not make, sometimes alongside a sudden drop in legitimate traffic as visitors get redirected elsewhere.

  • Enable WAF and CDN protections immediately if you have them available, since these absorb malicious traffic before it reaches your origin server.
  • Contact your registrar or ISP the moment you suspect a hijack rather than trying to fix DNS records yourself while access may be compromised.
  • Throttle or rate-limit suspicious traffic patterns while the attack is active.

A basic DDoS protection setup covers most of what a small site needs without enterprise-level cost. Registrar locks, two-factor authentication on every hosting and domain account, and a written incident response playbook are the cheapest insurance against this category.

Pro Tip: Lock your domain registrar account and enable two-factor authentication everywhere today, before an attack forces you to do it under pressure.

Monitoring and detection: shortening time to detect and time to recover

You cannot fix what you have not noticed, and the gap between an outage starting and someone noticing is often the largest chunk of total downtime. Monitoring practices such as synthetic uptime checks, real-user metrics, error tracking, and infrastructure alerting all help teams detect issues earlier and recover faster, which directly shortens both mean time to detect and mean time to recover.

  • Synthetic uptime checks ping your site from outside your network on a schedule, catching outages before a customer reports them.
  • Error and exception tracking surfaces application bugs as they happen rather than after a support ticket arrives.
  • A public status page cuts down on repeated support requests during an active incident.
  • A written runbook for your top outage scenarios turns a panicked scramble into a checklist.

For a small team without dedicated ops staff, a managed option such as inSave Hosting's monitoring tools covers these basics without building custom tooling. Larger teams with in-house engineering may prefer to build their own stack once traffic and complexity justify the investment.

A prioritized checklist for triage and prevention

Work through this in order rather than jumping straight to the most dramatic fix.

  1. Confirm the outage is real and global, not a local network or browser issue.
  2. Check host, CDN, and DNS provider status pages for open incidents.
  3. Verify domain registration and certificate expiry dates.
  4. Correlate the outage start time with your most recent deploy or DNS change.
  5. Roll back the last change, bypass the CDN, or fail over DNS if a fix is not immediately obvious.
  6. Restore from backup only after ruling out a simpler fix.
  7. Schedule a postmortem, and use it to plan redundancy or process changes.
ActionTypeTimeframe
Check status pages and DNS/cert expiryQuick winMinutes
Roll back last deployQuick winMinutes to an hour
Restore from backupQuick winUnder an hour
Add redundant DNS or multi-region hostingStrategic investmentWeeks
Staff training and written runbooksStrategic investmentOngoing

Why process beats hardware for most small teams

Most of the outages I have seen traced back to a missed renewal, an untested deploy, or nobody noticing for two hours, not to a server catching fire. A cheap uptime monitor, a written rollback runbook, and a habit of checking certificate expiry dates catch more downtime than any hardware upgrade. Read the linked resources here before you spend money on infrastructure you may not need yet.

— Ihor

inSave Hosting: managed hosting that removes the easy mistakes

Most of the causes above come down to something slipping through the cracks: a certificate nobody renewed, a backup that did not exist, a server nobody was watching. inSave Hosting builds those safeguards into the plan itself rather than leaving them to you.

inSave Hosting

  • Free SSL with automatic renewal removes the expired-certificate outage entirely.
  • Daily backups and free migration mean a bad deploy or a corrupted install is a restore away from fixed, not a crisis.
  • 99.9% uptime and monitored infrastructure cover the sitewide failures that are outside your own code.
  • Free one-year domain registration on qualifying plans keeps the registration-lapse scenario off your list of worries.

If your outages keep coming from administrative lapses or an overloaded shared server rather than your own code, take a look at shared hosting plans or, for WordPress sites specifically, the WordPress hosting plans built around automated backups and one-click management.

Sources

FAQ

Why do websites randomly go down?

A site rarely goes down for no reason: the usual culprits are an expired domain or certificate, a DNS misconfiguration, a host or network outage, or a recent deploy gone wrong. Checking status pages, DNS, and certificate expiry first usually finds the cause within minutes.

How to prevent website downtime?

Prevention comes down to redundancy and process: dual DNS providers, automated certificate renewal, tested deploys with an easy rollback, and monitoring that alerts you before customers notice. Written runbooks for your most likely outage scenarios cut recovery time significantly.

How to fix a website being down?

Start by confirming the outage is global rather than a local issue, then check your host, CDN, and DNS provider status pages, followed by domain and certificate expiry. If a recent deploy caused it, roll back immediately rather than trying to patch the live site.

What causes a website to time out?

A timeout usually means the server is overloaded, misconfigured, or unreachable due to a network or DNS problem rather than a coding error in the page itself. 500-level errors like 502 and 504 specifically point to the gateway failing to get a response from the origin server in time.