Every Friday in October, this blog takes one chapter of Strategic Web Management and argues its central idea, paired with the podcast episode and worksheet built from it. This week is Chapter 4, Hosting Without the Headaches, and the idea is the one the chapter spends its longest stretch defending: a backup is not a plan until someone has restored it.
That sounds like a slogan. It is actually a claim about where protection lives. Most organizations believe they are protected because a backup exists somewhere, and they believe it until the afternoon a restore fails. The chapter's position is that a backup's existence proves almost nothing, that its value is decided by four properties nobody checks at setup, and that the only honest evidence of protection is a log with dates and initials on it. Monday's post gave you the fifteen-minute test. Today is the case for why that test, and not the backup, is the thing you are actually managing.
The question that sets the budget
Chapter 4 opens the backup discussion with a question rather than a tool: how long can your website be down before it starts to cost you money? The book calls the answer your recovery time objective (RTO), the amount of downtime the organization can absorb before the loss becomes real. A brochure site for a consulting practice might tolerate two days. A site taking donations during a fall appeal might tolerate two hours.
The RTO sets everything else: whether a weekly backup is responsible or negligent, whether the host's included backup is enough, and whether managed hosting with one-click restores is an expense or a bargain. Skip the question and the backup schedule gets chosen by the host's default, which is a decision made by nobody about something that matters.
Four properties, not one
A backup, the chapter argues, has four properties, and a weakness in any one makes the others worthless.
Frequency is how often a copy is taken, and it must be shorter than the amount of work you could stand to lose. Retention is how far back the copies reach; a site compromised three weeks ago and backed up nightly has twenty-one infected backups and no clean one unless retention runs longer than the time it takes to notice a problem. Custody is who controls the copies and where they live, which is where most organizations fail without knowing it: the host's backups sit inside the same account as the site, so a suspension, a billing dispute, or the host's own outage takes both down together. The 3-2-1 rule answers custody directly: three copies, two kinds of storage, one somewhere the host does not control.
The fourth property is verification, and it is the one the chapter treats as the whole point. A backup nobody has restored is unverified. Organizations regularly discover mid-incident that months of "successful" backups captured the files and skipped the database, or captured a theme and skipped ten years of uploaded photos. The software reported success every night; nobody had opened one.
Why a signed log is the deliverable
The recommendation in Chapter 4 is a restore test every quarter, logged with the date, what was inside, where every copy lives, and the initials of the person who ran it. The initials are not ceremony. A name on a log changes how carefully the test gets run, and it answers the question a successor or a board member will eventually ask: when did someone last prove this works?
The log also outlives the person who kept it. Should you leave, be out sick during an incident, or simply be busy on the day the site disappears, the log tells the next person where the copies are and when they were last known to be good. Four dated lines a year is the entire discipline.
The mission tie-back
While every effort should be made to keep the site online, the honest planning question is not whether something will fail but how much of your organization's work you are willing to lose when it does. Chapter 4 argues that the answer belongs in writing, that the host's reassurance is not evidence, and that the quarterly restore test is the moment a backup stops being a hope and becomes a plan. Episode 5, Hosting without the Headaches of The Web Managers Podcast walks the same ground, scorecard included.
Get the chapter worksheet. The M4.1 Hosting Evaluation Scorecard and Backup Verification Log is the worksheet for Chapter 4 and Episode 5. It's free with a Web Managers account at https://webmanagers.com/register. The full chapter, with the four properties and the RTO conversation written out for a board, is in Strategic Web Management: First Steps for the Accidental Web Manager.