Fluxx Tech
Most businesses find out their website is down when a customer calls. By then it has usually been down for hours, and if it happened overnight, since the previous evening.
Monitoring costs very little and turns a half-day outage into a fifteen-minute one.
What to monitor
Check | Why |
Homepage availability | The basic “is it up” check |
Key pages | Product, checkout, contact, login |
Response time | Slow is a warning before down |
Content check | Page loads but is empty or shows an error message |
Transaction check | A simulated search, login or add-to-cart |
SSL certificate expiry | An expired certificate blocks every visitor |
Domain expiry | The most embarrassing and most avoidable outage |
DNS resolution | Misconfiguration or registrar problems |
Server resources | CPU, memory, disk; disk full is a common silent killer |
Database availability | The site can be up while the database is not |
Email deliverability | Enquiry emails silently failing |
Backup success | A backup that stopped running three months ago |
Content checks matter more than ping checks. A site returning a 200 status with a blank page or a database error passes a basic uptime check and is completely broken for customers.
Check from more than one place
A site may be reachable from your office and not from another city or network. Monitor from multiple locations, including at least one outside India if you have international customers.
Set alerts that get acted on
• Alert by the channel people actually read: WhatsApp, SMS or a phone call, not only email
• Alert at least two people, so a single unavailable person is not a single point of failure
• Escalate if unacknowledged after a set time
• Avoid alerting on every momentary blip; require two consecutive failed checks
• Separate severity: homepage down is urgent, a slow page is not
• Review alerts monthly and remove noisy ones
Alert fatigue is real. A team that receives twenty false alarms a week stops reading the twenty-first, which is the real one.
The two expiry checks that save embarrassment
SSL certificate expiry. When it lapses, browsers show a security warning and most visitors leave immediately. Alert at 30, 14 and 7 days. Use auto-renewal where available, and still monitor it, because auto-renewal fails sometimes.
Domain expiry. The site, email and everything else stop. Recovery can be slow and occasionally expensive. Keep auto-renew on, keep the registrar contact email current and monitored, and set a calendar reminder independent of the registrar.
Both of these cause complete outages, both are entirely preventable, and both happen constantly.
Write an incident response plan
One page, agreed before anything breaks:
1. Who is on call? Named people, with contact numbers and a backup.
2. First five minutes:
• Confirm it is real; check from another network or phone
• Check whether it is the whole site or one page
• Check hosting provider status page
• Check recent deployments; a change in the last hour is the most likely cause
3. Common causes, in order of likelihood:
Cause | Check |
Recent code or plugin change | Deployment log; roll back |
Disk full | Server storage |
Database down or at connection limit | Database status |
Expired SSL or domain | Certificate and registrar |
Traffic spike or bot attack | Traffic logs |
Hosting provider outage | Provider status page |
DNS change | Recent DNS edits |
Payment or third-party service failure | Provider status |
4. Decide: fix forward or roll back. Under pressure, rolling back to the last working state is almost always right. Investigate afterwards.
5. Communicate. Tell customers before they tell you. A WhatsApp status, a social post or a holding page saying “we are aware and working on it” prevents most of the frustration.
6. Afterwards: write down what happened, why, and what will prevent it. A short note, not a document nobody reads.
Keep a credentials and access list
An outage at 11 pm with the one person who has hosting access unreachable is a preventable disaster.
• Hosting, domain registrar, DNS, CDN, database, email: who has access
• Stored in a password manager the business controls, not one person’s laptop
• At least two people with access to each
• Support contact numbers for every provider, with your account numbers
See our data backup and disaster recovery guide.
Set realistic expectations
“100% uptime” is not achievable and not worth promising. Decide what matters:
• How long can the site be down before it genuinely costs you money?
• Which pages must never be down: checkout, booking, login?
• What is your target time to detect, and time to restore?
For most small business sites, detecting within five minutes and restoring within an hour is a reasonable and affordable standard.
Frequently asked questions
How much does monitoring cost? Basic uptime monitoring is often free or very cheap. Transaction and content checks cost more and are worth it for ecommerce.
Who should receive alerts? At least two people, including whoever can actually fix it, not only the business owner.
Does monitoring prevent downtime? No. It shortens it, which is usually where the cost is.
