More backups fail from a full destination drive than from any software bug. The cure is not a bigger drive — it is a schedule with a memory of what to delete.
Every backup product in this comparison offers the same ingredients: full images, incremental images, schedules and retention. The difference between a setup that runs for years and one that silently dies in month three is how those ingredients are combined. This guide gives the pattern that survives contact with reality — in standalone Reflect on a home PC, and in Site Manager repositories watching an office fleet — plus the simple maths to size storage before you buy it.
Reflect also builds synthetic fulls — merging a full with its incrementals into a fresh full in the background — which keeps chains short without re-reading the whole disk. Whichever pattern you pick, the chain is only as strong as the storage rules beneath it.
For most PCs and small servers: one full image weekly, incrementals on the other days, run outside working hours or at lunch. The weekly full bounds the chain length — a restore never assembles more than a week of pieces — while the daily incrementals keep each night's work small and quick, typically a few percent of the full's size. Machines with heavy daily change (video work, big databases) justify a mid-week full; machines that barely change can stretch to a monthly full with differentials.
Retention answers: how many backup sets do we keep, and when do old ones get deleted? Set it by time or by count — keep four weekly sets, for example — and let the software purge old sets automatically as new ones complete. Both standalone Reflect and Site Manager repositories support this; the classic home and office failure is leaving retention at keep everything, watching the destination fill over months, and then having every new backup fail. A full destination is not a storage problem you notice — it is a backup failure you discover later.
Thirty seconds of arithmetic replaces the two usual purchases: the drive that is too small by Diwali, and the enormous one bought in panic afterwards.
Whatever schedule you choose, wire up the alarm: standalone Reflect's email notifications per machine, or Site Manager's daily summary for the fleet. The dangerous backup failure is never the dramatic one — it is the quiet Tuesday-night failure, repeated for a month, found during a restore. Retention, schedule and alerts are one system; set all three or the weakest one defines your real protection.
See both tools working on real machines — Reflect on a single PC, Site Manager running an office full of computers.
▶️ Watch Full VideoIt is simpler, not safer. A nightly full costs roughly seven times the storage of the weekly pattern for the same week of protection, and in practice that cost is what pushes people to shorten retention or let the drive fill. The weekly-full pattern with intact chains restores just as reliably.
Restores can use points up to the last good link before the damage — later increments in that chain depend on the missing one. This is why chains are kept short with weekly fulls, why verification matters, and why the second copy under the 3-2-1 rule exists.
Long enough to notice a problem late. Ransomware and slow corruption are often discovered days or weeks after they start, so most offices keep at least three to four weekly sets, with one older set kept offsite. Decide by asking how old a disaster you still want to be able to undo.