Website maintenance is four jobs: keeping a site running, fast, accountable and able to survive change. Six ways in, and where to start.

Website maintenance isn't one task but four jobs: keeping the site running, keeping it fast, making someone accountable for it and making it survive change. Each has its own article in this section. This page repeats none of them — it tells you where to start.
How to use this section. If you have no specific problem, start with the three questions just below. If you do, find your sentence under "Six ways into this section" and go straight to the article that answers it. The calendar at the end tells you what to check, and how often, once everything is in order.
Answer three questions. Each takes a minute, and if you can’t answer one of them, that tells you which article to start with. These aren't technical questions — a business owner should be able to answer all three without calling the provider, because each concerns something that will be needed within a quarter of an hour on the day of an outage.
Website maintenance is four jobs
Digital Vantage
"We're signing a website maintenance contract — what should we watch for?"
Offers describe tasks and say nothing about three things: what you get when the provider breaks a promise, who owns what they build, and who holds the keys when you part. An SLA of 99.9% allows 44 minutes of downtime a month, and the compensation is often a partial fee refund, not your loss. The article also helps when the site is looked after by an employee — it explains why one person cannot provide on-call cover — and before changing provider, because it ends with five steps for handing a site to another firm.
→ Website maintenance services
"PageSpeed shows a red score — is it urgent?"
The score you see in PageSpeed and what Google measures are not the same list. Three thresholds, the difference between a test and data from real users, and the order in which to spend money on speed: first a layout that doesn't jump, then the images at the top of the page, the server last. Useful to anyone who has received an offer to "optimise to 90+ in PageSpeed" and wants to know what to ask before accepting it — and to anyone whose site gets too few visits for Google to measure it at all.
"We hear about outages from our customers"
What to check beyond whether the server responds, what free monitoring gives you, who should get the alert, and why the length of an outage is decided by the response rather than by how often you check. It applies to every site that collects enquiries or sells, and to every company whose provider assures it that it is "keeping an eye on things" — because your own monitoring is the only independent measure of whether the contract is being kept.
"The site shows a 500, 502 or 503 error"
A server error code tells you which segment failed — and whether to call the hosting company or the developer. With a "code — place — who to call" table and WordPress's critical error message. It applies when the error is already there, or appears now and then and disappears before anyone investigates — in the first minutes, "who should fix it" matters more than "how".
→ 500, 502, 503 and 504 errors
"After a redesign, Google is reporting 404 errors"
Where they come from, when to redirect and when to leave the error, what 403, 401 and 400 mean — and a map of all response codes. Most often needed after a redesign, a menu change or a content clean-up, when addresses that used to work multiply. It also explains why our own site answers non-existent addresses with a 200 code — and why that is a deliberate choice.
→ 404 Not Found, 403, 401 and 400
"We're changing hosting, domain or platform"
Three different operations called by one name, each with a different risk. Domain transfer, DNS propagation and a 301 redirect map — and why Google keeps seeing the old address for weeks. Best read before the decision, not during it: if the changes come together, separate them in time, because after a drop following three simultaneous changes you can't tell which one caused it.
Three paths, depending on what brought you here.
Nothing has happened and you simply want to put things in order. Start with the three questions above, then monitoring — the only article in this section after which something starts working without your involvement. Finally, maintenance services, if you have a provider, so you know which items on this list they handle and which stay with you.
Something is already happening. The site is down or showing an error — go straight to server errors and the "who to call" table. Pages are disappearing from Google, or customers land on "not found" — go to 404 not found. Both articles start with what to check in the first minutes.
You're preparing for a change or a conversation with a provider. Before a move — migration. Before signing a contract or changing firms — maintenance services. Before accepting an offer to "speed up the site" — Core Web Vitals. Each of these three articles gives you a list of questions you can ask at the meeting.
It shows most clearly in speed, because that is the one thing measured publicly across the whole web.
Roughly half of all sites pass Core Web Vitals
HTTP Archive, Web Almanac 2025, Performance chapter (CrUX)
According to the Web Almanac 2025, a good result is achieved by 48% of sites on mobile and 56% on desktop, and home pages do worse than inner pages. The home page is usually the most frequently changed part of a site — and every change can spoil a result that used to be good. That is why maintenance is a rhythm, not a project. We break down what this means for your site in the article on Core Web Vitals.
How often | What | Where it is described |
|---|---|---|
continuously, automatically | whether the site responds and shows the right content | |
weekly | server errors and new 404s in Search Console | |
monthly | updates one item at a time, a report from the provider | |
quarterly | a test restore of a backup, Core Web Vitals data | |
yearly | domain, accesses, PHP version, contract terms | |
with every move | a redirect map and a check that it works |
Every row missing from your provider's offer is a job that stays with you — which is fine, as long as someone in the company knows it.
Four jobs: keeping the site running (monitoring and responding to errors), keeping it fast (Core Web Vitals), making someone accountable for it (the contract with a provider) and making it survive a change of hosting, domain or platform. Updates and backups are covered in the security section.
It depends on four layers of the bill: infrastructure, licences, labour and overage. We break them down in a separate article on website maintenance costs, and you can work them out for your own site in the maintenance cost calculator.
Partly: keeping an eye on the domain, basic monitoring, updates on a simple site. Responding to outages outside working hours and changes on the server are harder. Most companies hand the technical part to a provider, but knowing where everything is and whose name it is in should stay in the company.
With three questions: who the domain is registered to, how you will find out about an outage, and when someone last restored a backup. If you can’t answer one of them, that tells you which article in this section to start with.
In an audit we go through access, monitoring, the speed Google measures and your redirects — and tell you which of the four maintenance jobs is weakest in your case.
Website maintenance is four jobs: keeping a site running, fast, accountable and able to survive change. Six ways in, and where to start.
A 502 Bad Gateway, 500, 503 or 504 error tells you which part failed: the application, the link between servers or an overload. And who to call.
A 404 Not Found on your own site is usually a page removed without a redirect. What 4xx status codes mean, what Google does and why our 404 returns 200.
Core Web Vitals are not your PageSpeed score: its heaviest metric is one Google does not use for ranking. The three thresholds and what to do about them.
Website maintenance services are sold as tasks but signed as a contract. Response time, SLA, domain access and code ownership — check these before you sign.
Website monitoring: a 200 code does not mean the page works — ours returns it for addresses that do not exist. What to check, how often, and who gets the alert.
Website migration is three operations: new hosting, new domain, new addresses. What to tell Google, how to transfer a domain and how to set up 301 redirects.
Table of Contents · 6 sections · 6 minutes read
Rate this article
Back to the guide: Websites — a guide to the whole section

What each part of the PageSpeed Insights report means: 28 days of user data, the Lighthouse score, phone vs desktop, and why the score keeps changing.

An ecommerce SEO audit runs mostly on free Google reports: indexing, Core Web Vitals, rich results, duplicates and Merchant Center data.

SLA meaning: how much downtime fits in 99.9%, how AWS, Microsoft and Google SLAs compare, SLO, RPO, RTO and 10 things to check before signing.

Cloud computing by the NIST definition: five traits, IaaS, PaaS and SaaS, public, private and hybrid cloud, and how Polish businesses actually use the cloud.

Low code and no code explained: who a citizen developer is, what a low code platform suits, its price limits and what you can take with you when you leave.

WordPress themes are not chosen on looks: three fields in the directory tell you what a theme will cost you in a year, and what disappears when you switch.

A 502 Bad Gateway, 500, 503 or 504 error tells you which part failed: the application, the link between servers or an overload. And who to call.

A 404 Not Found on your own site is usually a page removed without a redirect. What 4xx status codes mean, what Google does and why our 404 returns 200.

A free site is a real option with a precise limit. Three routes, what each one gives you, what it withholds, and what it costs once a year has passed.