The 2017 holiday season was notable due to its strong e-Commerce site performance (speed, availability), with clear signs showing e-Commerce retailers are improving their ability to perform under load.
- The top performers made incremental load time improvements this year compared to last. In holiday season 2017, the top-performing sites (both desktop and mobile) got faster under load compared to 2016. For example, the top three desktop sites in 2017 delivered performance ranging from 0.88 seconds to 1.29 seconds, compared to 1.25 to 1.48 seconds last year. On mobile, the top three sites ranged from 0.84 seconds to 1.3 seconds. Last year, those numbers ranged from one second to 1.3 seconds.
- While several leading e-Commerce retailers, including The Gap, H&M, Lowe’s and J. Crew experienced problems during Black Friday weekend, these issues were much shorter in duration than the prolonged outages we’ve seen in past years.
There are several dynamics at play here. Amazon set a new standard this past summer when its Prime Day performance beat out major competitors who weren’t even having sales that day. The company has set a new bar to clear (the “Amazon Effect”) and e-Commerce retailers, particularly the traditional top performers, are responding — taking every step to ensure shoppers get service seamlessly. In addition, the clear absence of a prolonged outage suggests that e-Commerce retailers are finally getting a better handle on sniffing out and fixing web performance problems early, before a wider swath of customers are impacted. This is being driven by other industrywide trends including:
Customer Centricity Is Taking Center Stage: According to Forrester, 40% of consumers have a high willingness and ability to shift spend, meaning they can easily reward or punish companies based on a single experience — a single moment in time.
According to one study in Forbes, 75% of companies said their number one objective was to improve customer experience — overtaking price and product as a key competitive differentiator. As this year progresses, expect IT organizations to continue shifting their mindset from internal system management and metrics (the domain of traditional application performance monitoring tools, or APM) to increasingly embed customer-centricity into their monitoring approaches.
Synthetic Monitoring Is Making A Comeback: Synthetic monitoring, also known as “active” monitoring, entails generating “dummy” traffic from the cloud to periodically access web sites from a wide variety of geographic locations. This gives IT teams a realistic view of a web site or application’s speed and availability for various end-user segments around the world.
Synthetic monitoring has been around for a while, but it is making a comeback in a big way, evolving to keep up with the times.Modern approaches eliminate extraneous Internet “noise,” giving IT teams the truest, clearest view of performance while filtering out false alerts. When combined with rich diagnostics, modern synthetic monitoring enables a painstaking degree of granularity of identifying the source of problems. Ultimately, this enables IT teams to quickly and accurately correlate performance to a wide degree of variables — internal infrastructure elements, third-party services, networks and APIs. Also, IT organizations are supplementing synthetic monitoring data with real user monitoring (RUM), or “passive” monitoring. This identifies what actions users take once they do access a site or application, thus helping IT teams prioritize certain conversion paths or landing pages for speed optimization.
AI Emphasizes “High Quality Data In”: Artificial intelligence is gaining major traction in performance monitoring, with tools using AI to discover patterns within very large data sets including log files, service desk and, increasingly, various monitoring practices. Modern synthetic monitoring itself is a predictive technology, using “robots” to simulate end users’ interactions, including the location and network from which they are accessing services. This enables IT teams to proactively identify and address issues before real end users are impacted. Because synthetic monitoring capabilities eliminate extra noise (the cloud, regional ISP networks, etc.) that can skew a true performance reading, this helps IT teams determine if a problem is truly theirs or not, thus reducing time spent war-rooming or finger-pointing unnecessarily. These “false alerts” tend to increase as web infrastructure grows more complex, so AI based on clean, noise-free data tends to yield the most accurate, actionable insights.
Addressing Elements Beyond The Firewall: Among the performance problems we tracked this holiday season, external third-party services buckling under load were often the culprit (which is consistent with past years). As more digital services move outside the firewall, traditional, inwards-focused APM becomes less sufficient. By monitoring the performance of third-party services and identifying when they are causing problems for a site overall, organizations can reduce their risk exposure, filling the gap left by more traditional APM solutions.
APIs Matter – And We’re Watching Them: In addition to third-party services, many web sites also depend on application programming interfaces, or APIs, to extend and create feature-rich functionality. For example, when an online shopper places an order on an e-Commerce site, the payment gateway uses APIs to verify the user’s credit card data. If the API that integrates the payment options on the site is broken, not only does this result in an abandoned cart, it also adds to frustration resulting in a negative user experience.
Like third-party services, APIs can suffer performance degradations (both under load as well as normal traffic periods) as a result of numerous factors, including API endpoints experiencing downtime; bugs in the integration causing errors and timeouts; API authentication taking longer than usual; the API service not being robust enough to handle spikes in incoming requests or outbound responses; and latency in API calls. Any of these can add up and create performance bottlenecks. Since API-dependent processes often support customer-facing, revenue-generating (and therefore mission-critical) applications, continually monitoring and testing APIs is an absolute must, and more e-Commerce retailers are paying attention to this critical area.
From a performance perspective, holiday season 2017 can be considered a success, but there’s always room for improvement, and e-Commerce retailers are making headway. Will we ever experience a perfect, crash-free, completely seamless holiday e-Commerce shopping season? No one knows for sure, but e-Commerce sites are clearly heading in the right direction and demonstrating a much higher degree of control. Customer-centricity; modernized synthetic monitoring; AI based on clean, high-quality data; third-party service and API monitoring are all factoring into this success and we expect these trends to continue gaining momentum in the months ahead.
Mehdi Daoudi is co-founder and CEO of Catchpoint, a digital performance analytics company. Before Catchpoint, Daoudi spent more than 10 years at DoubleClick and Google, where he was responsible for quality of services, buying, building, deploying and using various internal and external monitoring solutions to keep an eye on the DART infrastructure delivering billions of transactions a day using thousands of servers, routers. HP Openview, Sitescope, SMARTS, Adlex, Coradiant, Gomez and Keynote are some of the products that were used, and sparked his interest in building something in this space.