Downtime cost calculator. For the sale, not the year.
Most downtime calculators divide a year of revenue by 525,600 minutes, which describes no retailer that has ever run a promotion. This one asks about the sale you are actually worried about. Set it up in three taps and drag the slider.
Your sale
Those two are example figures, not your data. They sketch a big-ticket retailer on a peak sale day. Average order value swings enormously by sector, and order count is just revenue divided by it, so put your own numbers in before you quote the result to anyone.
£6,250 at risk
Revenue at risk
£6,250
30 minutes down in a 24 hour sale. About 25 orders. That is 97.92% uptime across this campaign, not across the year.
Every minute is worth
Orders per minute
Sale revenue
This spreads your orders evenly across the sale. Real sales are front loaded, so an outage in the launch hour costs more than this, and some interrupted shoppers do come back. Treat it as the order of magnitude, not the invoice.
And if it never goes down, just gets slow?
A site that stays up but crawls at checkout still loses the order. Hard down is a hosting problem. Slow under load is the one a performance test finds, and it is the more common of the two. Fair warning on the second slider: nobody publishes a measured abandonment rate for a site that is slow rather than down. We looked. So treat 20% as a placeholder to argue with, not a finding, and move it to whatever your own analytics say.
The number is easy. Knowing whether it happens is the hard part.
Any spreadsheet can tell you what a minute of your sale is worth. What it cannot tell you is whether your checkout survives the traffic you are about to send at it. That is a question with an actual answer, and you can have it weeks before the sale rather than fifteen minutes into it: run your real journeys in real browsers, at your forecast peak, and keep pushing until something bends.
How this calculator works
How is the cost of downtime calculated?
Your expected orders multiplied by your average order value gives the revenue for the whole sale. Divide that by the number of minutes the sale runs and you get revenue per minute. Multiply that by the minutes you are down and you have the revenue at risk. The model carries no industry coefficient of its own: once your figures are in, every number on the page is arithmetic on them. The values it opens with are examples, and the next answer says where they came from.
Where do the starting numbers come from?
They are examples, there so the page shows working arithmetic before you touch anything. They are not a claim about a typical shop. They deliberately sketch a big-ticket retailer on a peak sale day, because that is what most Adobe Commerce and Magento stores are: a high order value, and so fewer orders for the same revenue. There is no official average order value for the UK or the EU: neither the Office for National Statistics nor Eurostat publishes one, and the private panels that do disagree by roughly two to one, because each measures a different kind of merchant. Order count is simply revenue divided by average order value, and audited UK online retailers span roughly nine pounds to nearly three hundred on that figure, so no pair of defaults can fit more than a slice of the shops that land here. Put your own numbers in.
Why does it ask for a sale window instead of annual revenue?
Because spreading a year of revenue evenly across 525,600 minutes is not true of any retailer. Money arrives in bursts: evenings, campaign sends, peak weeks. Inside one sale the traffic is far flatter, so dividing by the minutes in that window is a fair approximation. Across a year it is not.
Why does it say revenue at risk rather than revenue lost?
Because not every interrupted shopper is gone for good. Some come back later, some were browsing rather than buying. Calling the figure a loss would overstate it. Treat the number as the order of magnitude of your exposure, not as an invoice.
What is campaign uptime, and why is it so much lower than 99.9%?
It is the share of your sale window that stayed up, not the share of your year. Thirty minutes of a 24 hour sale is 97.92% campaign uptime, which looks alarming next to an annual SLA figure. That contrast is the point: percentages that sound generous over a year describe a serious outage over a sale.
Does this only count a site being completely down?
The main figure does. The section underneath covers the more common case, where the site stays up but slows down badly enough that shoppers give up at checkout. You supply both the duration and the share who abandon, and that is deliberate: there is no published measurement of abandonment during degraded performance specifically, as opposed to general cart-abandonment rates or general page-speed studies. We went looking for one and it is not there. So the slider starts at a round twenty percent that you should overwrite with whatever your own analytics show.
How do I find out whether my site will actually go down at peak?
By testing it before the sale rather than during it. A performance test drives your real journeys in real browsers at the concurrency you forecast, and keeps going until something bends. That is what Evaluat does, and it is the only way to turn this estimate into a number you can plan against.