Managing Web Performance in a Modern World

This article by Mehdi Daoudi was originally published on Internet Retailer. 

As the scope and functionality of websites and applications continue to grow, the complexity of online systems is rising as well. New infrastructure is often needed to deliver web content globally, and various third party technologies and tags are often incorporated into a site to boost its performance or add advanced functionality such as advertising, personalization, and analytics. But this also means that the amount of new elements that must be monitored has skyrocketed in the past few years.

However, recognizing that each of these new elements must be monitored to ensure your site’s overall web performance is just a first step. Because once you start your monitoring initiative, you might find yourself drowning in an endless wave of information. Now, rather than remaining in the dark as to whether or not you have a problem that’s affecting end users, you suddenly have to sift through mountains of data just to find the problem – if one even exists in the first place.

Keep First and Third Party Systems Separate

The reality is that enhancing user experience and improving speed with a site or application means outsourcing to a lot of external vendors in order to meet many of your IT objectives. That can include infrastructure designed to boost your page load times such as DNS and CDNs, as well as page elements which are there to grow revenue such as advertisements, tracking tags, social media sharing buttons, etc. Every single one of these assets is intended to create better user experiences and increase site monetization, but they also represent a potential pitfall for your site’s load time and availability if not monitored and managed properly. Poor performance from just one third party service can be all that’s needed to drag down the performance of an entire site.

Yet with all of that information to keep track of, confusion can arise as to the exact origin of the problem. Is it one of your own systems that’s failing, or does the issue lie with of one of your vendors?

By separating your different elements into different first and third party zones, you can ensure that you only get alerted for issues that are within your control.

Be Careful of Who You Let Into Your House

As a buyer of third party systems, one of the most effective ways to hold your vendors accountable is through Service Level Agreements (SLAs). That means that you will need to monitor them regularly and establish clear performance baselines for each one so that you’ll know if an SLA has been breached.

The chart on the left shows a performance issue suffered by an ad server, and on the right is the corresponding issue suffered by an eCommerce site hosting those ads.

However, you must also have minimum performance requirements in place before you can even consider incorporating a new third party element into your site. This is where conflicts between the IT and Business/Marketing teams can arise, as a new ad or tag could bring in additional if it performs properly, but if it doesn’t it can torpedo the entire site. Standards can vary on a case-by-case basis, but at the very least there should be an explicit understanding between the two sides that if a tag is failing to live up to expectations or is having a negative net effect on the company’s bottom line, it has to go.

Slice and Dice Raw Data with Comprehensive Analytics

Now, even if your monitoring tool allows for grouping of different first and third party zones, you can still run into problems with the amount of raw data that comes in. Remember that too much data is often just as useless as no data; in order to really take the metrics that you collect and put them to use, you need a powerful analytics engine that will help you interpret what all that raw data means.

That means looking at the metrics of various systems that make up the infrastructure of a page independently of one another. Doing so will allow you to establish benchmarks against which you can measure future performance metrics, determine the effectiveness of different optimization techniques, and identify historical trends.

In terms of looking at historical data, it’s also vital to be able to analyze data as far back as two or three years in order to properly understand the full performance picture. Identification of month-over-month and year-over-year trends, especially for events that happen only once a year like specific holiday sales, require a long-term retrospective look. This can show you how your site performance may or may not have changed since adding new third party tags, as well as how these tags themselves hold up during critical, peak periods.

This chart shows the degradation of performance over a period of 12 months, with each bar indicating a three-month period. The X-axis represents a range of webpage load times, with the Y-axis representing the number of tests that fell in that range. As you can see, the tests from six and 12 months ago had far more results in the first range of response times, indicating that the site’s performance was far more consistent performance than the “base” of the last three months. This could also be seen with the “base” having a broader spectrum of values when compared to the other periods.

Slow Site = Lost Revenue

When it comes to eCommerce websites, time really is money, as a slow site is liable to drive customers away and towards a competitor. Worse yet, studies have shown that when a site is slow or unavailable, a majority of online shoppers will no longer buy from that site. And as we all know by now, it is more expensive to replace a lost customer than it is to keep an existing one.

The unfortunate fact is that end users don’t care about the complexity of a site, or that a problem is with a third-party vendor rather than a company or brand’s first-party infrastructure. From their perspective, the end result is the same – poor site performance, frustration, and aggravation. The only way to prevent that kind of negative result is with a thorough, comprehensive website and application monitoring strategy combined with powerful analytics. This is the key to organizations managing the complexity presented by third-party services – truly maximizing their value while minimizing exposure to overall website performance risks.