A site that "went down" overnight can lose customers for weeks before the owner finds out. Availability monitoring solves this problem: the system checks the site every few minutes by itself and notifies you when something goes wrong. Below: what to measure, how to set up checks and alerts, and how to act during an outage.
What uptime is and what the percentages mean
Uptime is measured as the percentage of time a site was available, and the difference between the "nines" is much bigger than it seems. 99% availability means about seven hours of downtime per month, 99.5% about three and a half hours, 99.9% about 43 minutes and 99.99% only a few minutes. A business card site can tolerate less; for a store with round-the-clock orders aim for 99.9%. These values are usually written into a support contract as an SLA; more on this in the material on a website support contract and SLA.
What website downtime costs
The direct loss is the orders and leads that did not arrive during the outage. For an online store it is easy to estimate: average revenue per hour multiplied by the duration of the outage. Indirect losses are harder: customers who did not come back, lower trust and the effect on search visibility. If unavailability repeats, search engines may crawl the site less often and rank it lower. A separate risk is downtime during ad campaigns, when you pay for clicks that lead to an unavailable page. Slow performance also cuts conversion even when the site is formally available: see the material on why a site loads slowly.
What exactly to monitor
A simple check of "does the home page respond" catches only the grossest failures. For real protection of the business, check several layers.
- Page availability: home page, catalog, cart, checkout page.
- Response correctness: a 200 status and the expected text on the page, not just the fact that it opens.
- Server response time: a sharp slowdown often precedes an outage; see also why a site loads slowly.
- SSL certificate: expiry date; see the material on why an SSL certificate is needed.
- Domain expiry: losing it shuts the site down completely; this is covered in the article on an expired domain.
- Key scenarios: the lead form, checkout, login to a personal account.
Tools and monitoring methods
For most small and medium businesses an external service that checks the site from several locations is enough. There are free and paid options (for example UptimeRobot, Better Stack, Pingdom); free plans usually have a check interval of about five minutes and paid ones down to one minute, so always confirm current terms on the service's website. More complex projects additionally use server resource monitoring (CPU, memory, disk), including systems such as Zabbix or Prometheus. Besides availability, watch for errors inside the application itself: we wrote about it in the article on website error monitoring. It is important that the check runs not from the same server where the site works, otherwise it will go silent together with the server during an outage.
How to set up alerts that help rather than annoy
Monitoring without sensible alerts is useless: either nobody learns about an outage or people stop reacting because of false alarms. Set up several channels, for example email and a messenger, and do not rely on only one. Require confirmation of unavailability from several locations before raising an alarm, to filter out isolated network glitches. Decide who receives alerts and who acts at night and on weekends. And keep a simple incident log: what happened, how long it lasted, what was fixed. Such a history reveals recurring problems that are worth eliminating systematically. If you do not have the resources to run this yourself, monitoring and response can be handed over to website technical support.
What to do when the site is down: the order of actions
During an outage it is important to follow a plan known in advance rather than improvise.
- Make sure the problem is real: check from another network and another device.
- Check the simplest causes: domain and SSL validity, hosting status, updates just made.
- See whether it is related to an application error: see the material on error 500.
- If the cause is an attack or overload, see the article on protection from DDoS attacks.
- Restore from a backup if other methods did not help: more in the article on website backups.
- After recovery, record the cause and the measures so the outage does not repeat.
Checklist: minimum monitoring for a business
If you do not have monitoring yet, start with the minimum: it takes less than an hour and immediately reduces the risk of missing an outage. If you want to outsource it, our website technical support team sets up and runs monitoring as part of maintenance.
- A check of the home page and key pages every 1–5 minutes.
- Control of the response code and the presence of expected text.
- Monitoring of SSL and domain expiry with an advance alert.
- Alerts in at least two channels for two responsible people.
- A defined outage action plan and the hosting provider's contacts.
- A monthly review of the incident log and the uptime figure.
