If you run a website, you already know the feeling: you press “update” and wait a second longer than usual before the page reloads. Most of the time everything is fine. When it is not, the difference between a quiet fix and a bad afternoon is whether you decided on a fallback beforehand. This FAQ answers the questions site owners ask most often about having a plan B for a content management system, in plain, friendly terms.
What does “plan B” mean for a website?
It means settling in advance how you will react if one specific part of your site stops working, and the sign that tells you to do it. The part might be a plugin, a theme update, a hosting account, a payment form or a contact form. The fallback might be restoring yesterday’s copy, switching a feature off, or showing a simple notice page while you sort things out.
Which parts of a CMS site fail most often?
The usual suspects are the places where two things meet. An update that changes how a plugin talks to the theme, a form that depends on an outside mail service, a checkout that relies on one payment provider, an expired certificate, a domain that renewed on a card that has since been replaced. Content itself rarely breaks. Connections do.
That is why a fallback list looks like a list of connections: hosting, domain, email, payments, third-party scripts and the people who hold the passwords.
Is a backup the same thing as a plan B?
No, and it is a useful distinction. A backup is a copy of your data. A plan B is a decision about your next move. Having a copy is the raw material, but if nobody knows who restores it, how long the restore takes, or when to choose it over fixing the live site, the copy sits unused while visitors wait.
Write one line for each backup: where it lives, who can restore it, and the situation in which you would use it. If you cannot finish the line, you have found a gap.
How do I decide when to switch to the fallback?
Pick a trigger you can observe without arguing. For an update, it could be “if the homepage or checkout shows an error after the update, and a quick look at the error log does not explain it within fifteen minutes, we roll back”. For a payment provider, it might be “if three attempts fail in an hour, display the alternative payment option”. The time limit is the clever part: it stops an hour of hopeful poking from becoming the plan.
What should I do before pressing “update”?
- Take a fresh copy of the files and database, and confirm you can open it.
- Try the update on a staging copy first, if your hosting allows one.
- Choose a quiet hour, not the busiest time of the week.
- Write down what you will check afterwards: the homepage, a key service page, a form submission and the checkout, if you have one.
- Decide your rollback trigger and who makes the call.
None of this takes long. The first time you do it feels fussy, and the second time it feels routine.
Do my most important pages need their own fallback?
Yes, in a modest way. If your contact form is the way customers reach you, a fallback might be a visible email address or a click-to-call link in the page itself. A site structured around a clear set of essential pages makes this easier, and the write-up covering the foundation pages a firm’s site ought to include helps when deciding which ones deserve protection first.
Should planning for fallbacks start before the site is built?
Ideally, yes. It is much cheaper to choose a host with easy restores, or to avoid depending on a single plugin for a critical job, during the design stage than after launch. If you are at that early stage, the step-by-step method for planning a business site before development begins pairs well with a short “what if this fails” column added to your plan.
What about the human side: passwords, logins and who is on call?
This is where many sites quietly rely on one person. If only one team member can log in to hosting or the domain account, that is a dependency like any other. The fallback is a second authorised contact, a password manager with shared access for the right people, and a written note saying who calls the host. Keep the note somewhere that does not depend on the website being online.
How do I test that a restore actually works?
Restore the copy onto a staging address or a spare folder, open the main pages, log in, submit a test form and look at a recent post. Then note how long the whole exercise took. That number matters, because it becomes part of your trigger: if a restore takes forty minutes, a rule that says “give the live fix an hour” leaves visitors staring at an error for an hour and forty minutes. Rehearsing the steps once also teaches you where the login details are kept, which is usually the part that surprises people.
Can I show visitors something while I fix things?
You can, and it is a good fallback in its own right. A simple maintenance page with a short message, an email address and a phone number is better than a white screen. Prepare it ahead of time, store it where you can switch it on in a minute, and keep the wording calm and honest. Visitors forgive a brief pause far more readily than a broken checkout.
Is there anything fun about all of this?
There is, honestly. Site owners who plan this way often find they enjoy launches more, because they stop bracing for the worst. If you run a site that promotes local activities and want to double-check what a clear category listing looks like, MyEventNow, an events-listing website that gives upcoming events, venues, a blog and a forum their own sections, makes a tidy example to study when you map how visitors travel through your own pages.
What if I am not technical?
You do not need to be. The method is the same for a one-person shop as for a larger team: list the connections, pick one fallback for each, write the trigger, and tell somebody. If you work with a developer or a hosting provider, ask them two questions: how long does it take to restore the site from a copy, and who decides when we do it? Their answers will tell you whether your plan B is real.
How often should I review the plan?
Whenever something significant changes: a new payment provider, a redesign, a change of host, a new team member with access. Otherwise, a quick read through every few months is plenty. Treat the document as a short note, not a project.
The short version
Decide what you will do if a connection fails, name the sign that tells you to do it, make sure someone else knows, and test the restore before you need it. That is a plan B, and it is far less work than recovering without one.