Cheaper is a rounding error next to unplanned downtime

The claim

When a business chooses a hosting provider or a support vendor primarily on monthly price, it is optimising the smallest term in the equation. The cost that dominates the total is unplanned downtime, and it is dominated by it by an order of magnitude for any business that transacts online. A provider that is $200 CAD a month cheaper but delivers one extra four-hour outage a year has, for most businesses, cost you money, not saved it.

Put your own downtime cost on paper

The number is not abstract; you can compute it from figures you already have. Take monthly online revenue, and convert it to a cost per hour of business time:

monthly online revenue / business days per month / business hours per day
= revenue exposed per hour of downtime

A business doing $80,000 CAD a month online, over 22 business days at 9 hours, has roughly $400 CAD of revenue exposed per business-hour. That figure is the honest input to every hosting decision, and most owners have never calculated it, which is precisely why price wins the argument by default.

The costs beyond the lost hour

Revenue exposed during the outage is only the visible part. The full cost of an incident includes several things the invoice comparison ignores:

• Deferred and lost sales. Some customers return after an outage; some buy elsewhere and do not. The split varies, but the lost fraction is real and permanent.
• Staff time consumed. An incident pulls your best technical people off planned work for its duration and often the rest of the day, and that work still has to happen later.
• Support load afterward. Every customer who hit an error generates a question, a chargeback risk, or a review, each of which costs time to handle.
• Reputation, which compounds. A customer who meets an error on their first visit may never return, and you never see them in the data to know it happened.

A conservative accounting typically puts the full cost of a single four-hour business-hours outage at two to four times the raw revenue-exposed figure. Using the example above, a four-hour outage is not $1,600 — it is realistically $3,000 to $6,000 CAD once the trailing effects are counted.

What the cheaper option is usually cutting

A provider is not cheaper by magic; the lower price comes from spending less somewhere, and it is worth knowing where. Common economies that produce more downtime: oversubscribed hardware, so your instance competes for CPU with noisy neighbours; no meaningful redundancy, so a single hardware failure is your outage; thin or offshore-only support, so the 2 a.m. incident waits for business hours; and no proactive monitoring, so you find out about problems from customers. None of these appear on the pricing page. All of them appear in the incident log.

The comparison that should drive the decision

The cheaper provider is $2,400 a year cheaper on the fee and $6,600 a year more expensive on the total. The fee is the number on the contract; the total is the number that leaves your bank account. Optimising the first at the expense of the second is the most common expensive mistake we see in this category.

What this does not argue

This is not a claim that expensive is always better — price is not a proxy for reliability any more than it is the opposite, and there are overpriced providers delivering mediocre uptime. The claim is narrower: reliability, not price, should be the primary axis of the decision, and price should be the tiebreaker between options that clear your reliability bar. Ask a prospective provider for their actual measured availability, their redundancy design, and their real support response times — and weigh the monthly fee only after those answers have narrowed the field.

How to decide

Compute your own revenue-exposed-per-hour figure this week; it is the single most clarifying number in any infrastructure conversation. Then, for any provider you are considering, estimate the annual downtime cost using that figure and compare totals, not fees. The decision usually inverts once the full arithmetic is on the page, and it inverts toward the option that was never the cheapest line item — because the cheapest line item was never the largest cost.