Keeping a website healthy after launch: what to monitor, back up and fix
Short answer: A website maintenance plan is a short routine that keeps a launched site online, safe, fast and able to be restored. It has four parts: monitoring (a service that checks your site every few minutes and alerts you), backups you have tested by restoring, updates, and a one-page list of who owns every login. Free monitoring tools cover most small sites. The step most people skip is practising a restore, because a backup you have never restored is only a guess.
Your site is live and nobody is watching it, and you want a plain routine that catches problems before a customer does.
I'm Ranjit, and I run Blur Studio: 15+ years building software, 120 apps taken live, and clients who have gone on to raise $10M+. I build AI systems and processes that help businesses do a lot more with a lot less time. I also work differently from most of what else is out there, more on that further down this page. Say hello.
Short version: set up alerts so you hear about an outage before your customers do, keep two kinds of backup (the site and its data), restore one copy to prove it works, and put the renewals and logins on one page. The rest of this article shows how, with real tools and limits.
What to monitor
Uptime monitoring means a service visits your site on a schedule and alerts you if it fails. A basic "is it up" check returns a status code of 200, which says the server answered. That is not enough. A site can answer 200 and still show a blank page, a broken form or an expired certificate. Check these:
- Home page, with a keyword check. Ask the monitor to confirm a phrase appears, such as your business name. An error page will not contain it.
- Your most important page. The quote or contact page, since that earns money.
- The form endpoint. Submit a test form on a schedule or watch for a regular "heartbeat" message (a ping your own system sends so the monitor can tell it is alive). See forms that stop sending leads if one is quiet.
- SSL certificate expiry. SSL is the padlock that encrypts traffic. An expired certificate scares visitors away.
- Domain expiry. If the domain lapses, the site and your email can vanish.
- Speed. Check Google's page speed report monthly. See page speed problems on marketing sites.
Monitoring tools and free tiers (November 2026)
| Tool | Free plan, as listed on its pricing page | Paid, as listed |
|---|---|---|
| UptimeRobot, an uptime checker with status pages | 50 monitors, 5-minute checks, 1 status page, 3 months of history, and a keyword monitor. Its page says it is good for personal and commercial projects. It is inconsistent on whether SSL checks are free, so confirm that before relying on it | Solo $9 a month billed annually ($10 monthly), with 60-second checks |
| StatusCake, a monitoring service | 10 uptime monitors at 5 minutes, 1 page speed monitor, 1 domain monitor, 1 SSL monitor | Superior listed at $20.41 or $24.49 a month (the page does not label which is annual), 100 uptime monitors at 1 minute |
| Better Stack, a monitoring and status page service | 10 monitors and heartbeats combined, 1 status page | Extra packs of 50 monitors at $21 a month billed yearly ($25 monthly). Its monitoring page lists SSL and domain expiration among the things it checks, without detail on scope |
| Pingdom, an older monitoring service | Check its pricing page; the amount is not shown without choosing a plan | Check its pricing page |
Limits and prices change, and free plans can be reduced, so confirm on each page before you rely on one. Two monitors from different providers is a cheap way to avoid a single tool failing silently. Send alerts to at least two people, by text or an app, not only email on the server that might be the one that is down.
Backups: what to back up depends on your site type
A backup is a copy you can restore from. What you must copy differs by how the site is built:
| Site type | What to back up | Notes from the platform docs |
|---|---|---|
| WordPress | The database and the files (core, themes, plugins, uploads, wp-config.php and .htaccess) | WordPress.org says a small site about weekly and a busy one daily, always before an upgrade or move, and to keep 3 to 5 recent copies in different places |
| Managed WordPress host | Use the host's backups and also keep your own | WP Engine makes daily checkpoints kept for 30 days, and says a restore overwrites the current content |
| Static or git-based site (Netlify, Vercel, Cloudflare Pages) | Your code repository, plus any forms, CMS content and data held elsewhere | Rollback keeps old versions only for a limited time (below) |
| AI website platform | Check how it keeps versions | Ploy keeps each publish as a saved version you can restore: see Ploy hosting, speed and uptime |
| Site or app with a database | The database and any uploaded files | Supabase Free lists no daily backups, Pro keeps 7 days and Team 14. Uploaded files in its storage are not part of those backups (its docs, November 2026) |
| Forms and data in a third-party tool | An export of the records | Schedule a monthly export to a place you control |
WordPress.org also suggests occasionally making a manual backup to check that your automatic ones work. That is the right instinct: a backup that was never opened is unproven. For apps with data, the library page on AI-built app backups goes deeper.
Rollback is not a backup
Many hosts let you roll back to an earlier version of the site, which is quick and useful. It is not the same as restoring your data. Know the limits:
- Vercel, a hosting platform, says an instant rollback does not change environment variables (the private settings your site reads at run time) and that the configuration of the old deployment may be out of date. It also says to expect that external APIs, databases and content systems may behave differently after a rollback. On its Hobby plan you can only go back to the immediately previous deployment. After a rollback, new pushes no longer go live automatically until you promote a deployment again (its docs, last updated July 2026).
- Netlify, a hosting platform, calls rollbacks instantaneous, but if automatic publishing from Git is on, a new Git-triggered production deploy overwrites the version you rolled back to unless you lock publishing. By default it deletes deploys after 30 days, or 90 on paid plans, apart from the live one and a few recent successful ones.
- Cloudflare Pages, a hosting service, removes deployments older than 30 days except the 100 most recent and the current one. Its page does not say what happens to your data or settings on a rollback, so do not assume they come back.
The simple rule: a rollback restores pages, not your database and not your secrets. Back those up separately.
Practise a restore
- Pick a copy of your backup and restore it to a test address, not live.
- Open the home page, a service page, the contact form and, if you have one, a login.
- Submit a test form and check that it arrives.
- Write down how long it took and what you needed. That is your real recovery time.
- Repeat after any big change, and at least twice a year.
Pass means the restored site works end to end. Fail means something is missing: fix the backup, not the test.
Updates and security
Software that is out of date is the usual way small sites get hacked. On WordPress, back up first, check that plugins and themes are compatible, then update. WordPress.org says minor and security updates now apply in the background on most sites, while major releases need you to click update. A failed update can leave a file named .maintenance in the site folder that keeps the site showing a maintenance message; deleting it is the common fix, but check the WordPress documentation first.
For any site, also: use a password manager and two-step login on the host, domain registrar and email; remove users who have left; delete plugins you do not use; and never share one login among several people. If you let AI tools edit your site, take a restorable copy first. See updating websites by chatting with Claude or ChatGPT. For the monitoring side on one host, see how to know your site is slow or down before customers do.
Certificates and domains: renewals that bite
Let's Encrypt, a free certificate authority, says its default certificates last 90 days and should be renewed every 60 days, so automatic renewal is normal. Certificate sellers such as DigiCert and Sectigo describe a schedule from the CA/Browser Forum, the body that sets rules for certificates, that cuts the maximum life of public certificates to 200 days from March 2026, 100 days from March 2027 and 47 days from 2029. Check the Forum's own ballot text before you plan around exact dates. Either way, yearly manual renewal is ending. Set your site to renew automatically and use a monitor that watches expiry. See custom domain, DNS and SSL setup and fixing "not secure".
For domains, turn on auto-renew, keep the payment card current, and make sure the registrar's emails go to a shared address, not to a person who might leave. Check your registrar for what happens if a domain lapses: the grace period varies by registrar and domain type, so ask for the exact period and do not rely on it.
A monthly routine you can copy
GoDaddy, a hosting company, lists a similar monthly, quarterly and yearly split in its maintenance guide (April 2026). Here is a version for a small site:
- Monthly (30 minutes): check monitor reports and alerts; confirm last backup date and size; apply updates; test the main form; check page speed on the home and a key page; look for broken links; read the search engine's coverage report for errors.
- Quarterly: review content for old prices and staff; test checkout or booking end to end; remove unused plugins and accounts; restore a backup to test.
- Yearly: confirm domain and hosting renewals and the payment card; review hosting costs and plan; review legal pages and cookie notice; review who has access to what.
The one-page runbook
When something breaks you do not want to hunt for logins. Make one page, kept somewhere that is not on the site itself:
- Domain registrar, and who owns the account.
- Where DNS (the address book that points your name to your host) is managed.
- Hosting or platform, and the support contact.
- Where the backups live and how to restore.
- Email, analytics, search console, form tool, CRM.
- Who is the first, second and third person to call.
- Renewal dates for each.
If the site goes down: an incident checklist
- Confirm it. Open it on mobile data and ask someone elsewhere. Check your monitor.
- Check your host's status page for a known outage.
- Look at what changed. Update, plugin, edit, DNS change, expired certificate or domain?
- Undo the last change. Roll back the site version, or disable the last plugin.
- Restore from backup if undoing does not work, to a test address first if there is time.
- Tell people. Put a note on your phone line, email signature or social page. A short status page helps.
- Check forms and leads for the period you were down, and follow up on any missed.
- Write down what happened and one change that prevents a repeat.
Call for help if the problem is in a domain, DNS or database you do not control, if you suspect a hack, or if the first restore fails.
What care plans cost, and what to ask
GoDaddy's guide gives ranges that suit a seller of hosting: $50 to $500 a month for a small to medium business, $300 to $1,000 for ecommerce, and $1,000 to $5,000 or more for large corporate or custom sites (April 2026). Treat any range as indicative, since sellers set it. A good plan names what it covers: hosting and domain handling, security monitoring and backups, updates and bug fixes, speed work, and an hour allowance for changes. Ask how often backups run, where they are kept, whether they have been restored, how fast they respond to an outage, and whether you own the logins. More detail in what website care plans cover.
A worked example, with made-up numbers
A small dental practice has a WordPress site with a booking form that gets about 15 bookings a week. Suppose an update breaks the form on a Friday night. With no monitoring, they find out Monday: three days of lost bookings, perhaps 6 or 7. With a keyword and form check every 5 minutes and an alert by text, they know by Friday night, roll the plugin back, and lose none. The tools cost nothing on a free plan. The numbers are invented, but the cost of finding out late is real.
How I'd approach this
I would not buy a care plan on day one. I would do the free version first, learn what breaks, then decide whether to pay someone to do the monthly routine.
First-week plan: day one create the runbook page with logins and renewal dates. Day two set up two free monitors with alerts to two people, with a keyword check on the home page. Day three confirm automatic backups exist for both the site and the data, and where they are stored. Day four restore one to a test address and time it. Day five put the monthly routine in the calendar with a named owner. Decision rule: if your site takes bookings or payments, or nobody on your team will do the monthly routine, pay for a care plan. If it is a brochure site and someone will do it, keep it in house.
If you would rather hand the monitoring, backups and fixes to someone, Blur Studio builds the custom version of this, including ongoing care for sites and apps: see what that includes.
Questions people ask
What does a website maintenance plan include?
Monitoring and alerts, backups you have tested, updates, security checks, speed checks, renewal dates for domain and certificate, and a short list of who owns each login. Some plans add content changes by the hour. See what care plans cover.
How much does website maintenance cost?
GoDaddy's guide (April 2026) gives $50 to $500 a month for a small to medium business site, more for ecommerce and custom sites. That range comes from a seller, so use it as a guide and compare written quotes. Doing it yourself costs time and often nothing for the monitoring tools.
Is free uptime monitoring good enough?
For many small sites, yes. UptimeRobot lists 50 monitors at 5-minute checks on its free plan, and StatusCake and Better Stack have free tiers too. Check the current limits, and use two alert contacts. Pay when you need faster checks or more monitors.
How often should I back up my website?
WordPress.org suggests about weekly for a small site and daily for a busy one, always before an upgrade or move, and keeping 3 to 5 copies in different places. A site that takes orders or bookings should be backed up daily or better. Test a restore at least twice a year.
Does rolling back restore my database?
No. A rollback puts back an earlier version of the site pages. Vercel's documentation says it does not roll back environment variables, and warns that databases may behave differently. Back up your data separately.
How do I know if my SSL certificate is about to expire?
Use a monitor that checks certificate expiry, such as the SSL monitors on StatusCake or Better Stack. Turn on automatic renewal at your host. See custom domain, DNS and SSL setup.
What should I do first when my site goes down?
Confirm it from another network, check your host's status page, then work out what changed last and undo it. Restore from backup if that fails, tell your customers, and review any missed leads afterward. The checklist above lists the order.
Do I need maintenance for a simple static or AI-built site?
Less than WordPress, but not none. You still need monitoring, renewals, a copy of your content and data, and a way to undo a bad change. See Ploy hosting, speed and uptime for how one platform handles versions.
More in this topic: Websites: custom builds, portals, lead capture and site care (9 guides) →