The Performance Beacon

The web performance, analytics, and optimization blog

Four lessons we can all learn from Tax Day outages

If you’re one of the 68 million Americans who have yet to file their taxes, there’s some good news: the IRS says you have three extra days.

But there are perils to waiting till the eleventh hour. And according to a Harris survey we commissioned earlier this month, many people are aware of these risks:

  • 46% of respondents are nervous about their tax websites having problems on Tax Day.
  • 31% are worried that their personal information is vulnerable. (This isn’t the only security-related risk. According to tax experts, the longer you wait to file, the more opportunity someone else has to file a return in your name, making you more vulnerable to tax fraud.)
  • 15% are worried that they could go to all the trouble of filing online, only to lose all their work at some point in the process.

At this time of year, a lot of eyeballs are pointed at tax sites, but there are important takeaways for any business:

1. Beware of load creep

Each year, more people file online than ever before. In the Harris survey, 72% of respondents said they either filed or planned to file their taxes online, more than any previous year.

What we can all learn: Online traffic is growing relentlessly. When you’re testing your site, it’s never safe to assume that the past is any predictor of the future. That’s why you need to test in production and augment your tests by monitoring and measuring your traffic.

READ: Why performance testing in production is not only a best practice — it’s a necessity

2. Traffic surges are unpredictable

One out of five (21%) of survey respondents said they plan to file at the last minute. This is even higher than the roughly one out of seven last-minute filers reported by the IRS last year.

Nation of procrastinators

That’s a lot of traffic — enough to put serious strain on a website’s infrastructure. Way back in 2007, the TurboTax outage that happened on Tax Day eve scared the dickens out of thousands of procrastinators. After that situation, Intuit threw a lot of hardware at the problem, so 2008 went well. But the service suffered again in February 2009, during a “mini peak” when an unexpected number of early filers logged on. (Disclosure: After this happened, Intuit became a SOASTA customer.)

What we can all learn: As above, past behavior isn’t an accurate prediction of future behavior. Your site needs to be ready, always, for unexpected traffic. Again, this is where monitoring and measurement complement your testing efforts.

READ: How to prepare your site for the holidays: 12 valuable lessons from

3. You’re only as fast and available as your slowest third party

You don’t just need to worry about the performance of your tax filing site. If you contribute to a 401K, have a lot of investments, or rely on other third-party financial services, then you need to worry about how each of the sites that host your accounts perform. That’s because your filing site makes calls to each of these to pull essential information. A slowdown — or worse, an outage — at any of those sites can bottleneck your filing process. When you consider that some people need to pull info from 20-30 different sites in order to file, that’s a lot of potential points of failure.

What we can all learn: You need to factor third-party performance into your dev, test, and monitoring activities. To illustrate: One year, during Nordstrom’s anniversary sale, the engineering team realized that their order management system had stopped processing orders any more. They discovered that, just hours before the sale started, the vendor they used for fraud checks had implemented a change to the service, which was blocking all orders.

READ: 10 pro tips for managing the performance of your third-party scripts

4. Things fail

Earlier this year, the IRS site experienced an outage that affected several services and tools. The downtime was reportedly due to unexpected hardware failure.

What we can all learn: Despite your best preventative measures, sometimes failure just happens. That’s why real-time monitoring, alerting, and diagnostics are your friend.

READ: A 7-part guide to sanity check the hysteria behind holiday website outages


If you want to mitigate these four issues, then you need to consider the following:

  • Continuous testing in production is crucial. Continuous testing lets you gauge performance and availability quickly and efficiently. It also lets you find and fix errors and bottlenecks that won’t show up in lab tests.
  • Testing in the cloud lets you simulate unprecedented loads — 200% or more of your anticipated peak load — even while people are using your site. This exposes component- and system-level problems that you won’t see otherwise.
  • User monitoring gives you real-time visibility into how people are using your site. Real user monitoring lets you see how 100% of your visitors are using your live site, where performance issues are happening, how those issues are affecting your user and business metrics, and how to triage and fix the problems.

Learn more about the SOASTA Digital Performance Management platform

Tammy Everts

About the Author

Tammy Everts

Tammy has spent the past two decades obsessed with the many factors that go into creating the best possible user experience. As senior researcher and evangelist at SOASTA, she explores the intersection between web performance, UX, and business metrics. Tammy is a frequent speaker at events including IRCE, Summit, Velocity, and Smashing Conference. She is the author of 'Time Is Money: The Business Value of Web Performance' (O'Reilly, 2016).

Follow @tameverts