How Much Do Website Outages Truly Cost?
Downtime affects all websites at some point or another, and even the best prepared sites will encounter problems that will impact visitors. At the same time, any number of vendors remain ready to provide the latest and greatest infrastructure, claiming to eliminate every potential problem. Each level of additional complexity can cut out another fraction of potential downtime, but eventually sites reach the point of diminishing returns and the investment becomes worthless. So how do retailers decide which investments will provide a positive return? Would investing in $12,000 of redundant servers be worthwhile?
The best way to decide is to build a model of the financial impact of outages and use that to compare the costs with the investment needed to avoid them. How should an e-commerce site calculate the cost of downtime? The best approach will vary from site to site, but even a simple model should always include three fundamental components:
- direct sales loss from people that are unable to complete their transactions;
- indirect sales loss from browsers that are turned off from your site and never get to the purchasing phase; and
- the increased operational costs of responding to traditional and social media inquiries during and after the outage.
Direct Sales Loss
An outage creates a direct loss of revenue in a very simple way: the outage, whether it occurred due to malicious intent, high traffic load or poorly managed server updating, shuts down every visitor's ability to access the checkout stage of shopping. Most retailers sell items which can be purchased at a competing site, meaning any downtime results in not only a direct sales loss but also opens the door for the competition.
The cost of this type of outage can be estimated by the following formula: Cdirect = avg hourly sales volume x (outage duration in hours + avg purchase session length in hours).
The average purchase session length is critical because it addresses anyone already in the checkout flow prior to the outage. For example, in a $10 million per year business ($1,141/hour) where the average customer takes five minutes (.083 hours) to purchase, the direct cost of a hour-long outage equals $1,236.
Indirect Sales Loss
Outages affect browsing consumers similarly by denying them access to the products they're interested in and might purchase. Imagine shopping in a brick-and-mortar store when a sudden power outage not only turned off the lights but causes all of the merchandise to disappear as well.
You can estimate the cost impact from these shoppers using the following formula: Cindirect = avg hourly visitors x conversion rate x avg sale x (outage duration in hours + avg visit length in hours).
For example, if your website has 1,000 visitors per hour, who spend an average of 20 minutes (.333 hours) on the site and 5 percent eventually purchase a $30 item, the indirect cost of a hour outage equals $1,999.
Increased Operational Cost
Finally, outages trigger a number of complaints from visitors to service representatives via email, phone calls and social media channels like Twitter. In addition to the indirect (and difficult to measure) impact to the brand from the social media discussions, outages bring a direct cost of increased volume for call-center staff.
This operational cost can be estimated as follows: Ccomplaints = increased ticket volume x average ticket response cost.
For example, if 8 percent of your site visitors complain about the outage, and the average ticket cost is $10, then our hypothetical hour-long outage increases call-center costs by $800. Now add the direct sales loss, the indirect sales loss and the operational cost and you find the total cost for the outage equals $4,035.
While this model is based on a number of simplified assumptions, it should provide a starting point for developing a more accurate model to predict operations costs. Then, by taking into account factors such as seasonal fluctuations in traffic, sales, and call-center costs and processes, retailers can get a relatively accurate assessment of the true cost of downtime.
Once retailers have this model, they can more effectively evaluate whether to make certain infrastructure investments. If the $12,000 investment in servers is only likely to prevent a single hour-long outage, it's probably not worth it. But if it can eliminate a hour-long outage every month, then it pays for itself after three months and may be a wise investment.
Jason Abate is the CEO and founder of Panopta, a company that monitors websites, email servers and network infrastructure.