In the iconic Dr. Seuss children’s book, the surly Grinch does his best to ruin Christmas for the children of Whoville. In the digital commerce arena, another G-word is set to wreak havoc this Black Friday and Cyber Monday (BFCM): the Glitch. More accurately, retailers need to be worried about multiple, elusive yet costly micro-glitches.
BFCM and the Big Glitch
BFCM (officially from Friday, November 24 to Monday, November 27 this year) signals the start of holiday shopping hysteria. With the pre-hype starting already on November 1, BFCM is quite simply a bottom line-defining phenomenon. Last year, more Americans shopped online over BFCM than voted in the Presidential election. Retailers raked in well over $3 billion over the long 2016 Thanksgiving weekend, and this year spending is set to rise yet again by over 5%, to some $3.5 billion. According to the National Retail Foundation, overall online holiday season sales grew by nearly 13% last year—and there’s no economic or technological reason to expect this year’s growth to fall short.
Retailers take BFCM very seriously, and with good reason. According to companies I work with, every incident that negatively impacts sales during regular shopping days costs some $40,000 in lost revenues. During BFCM, that amount is closer to $250,000.
To eliminate revenue-sapping incidents, companies prepare web site infrastructure well in advance for the tremendous increase in traffic and sales volume. They tweak code and minutely review customer paths to purchase. They build bespoke BFCM situation rooms to closely monitor operations and handle problems. And they succeed, for the most part. With some notable exceptions, the era of large-scale site meltdowns—resulting from Glitches with a capital “G” —during BFCM is mostly behind us.
Meet the Micro-Glitch
If big Glitches are less of a danger, then the threat this year and moving forward is from multiple, hard-to-identify micro-glitches.
As online commerce has grown exponentially, so has its complexity. Glitches have scaled down in size, but they’ve grown in number, sophistication, and—especially—identifiability. The fact is that multiple micro-glitches may cut as deeply, if not more so, than catastrophic failures.
Need a real-life example? A major online retailer I worked with saw a sudden 15% drop in sales over the course of one week. Data analytics was puzzled. Website management couldn’t find anything wrong. The pricing team found no market trends or competitor offers that would affect sales so dramatically. Sales continued to drop.
Finally, quite by coincidence, an employee tried to purchase a popular item from the site, but couldn’t complete the checkout process. Her enquiry led R&D to investigate, and the company found that the specific product, and some 20 others, had been incorrectly coded for checkout. Anyone that added these items to their cart would be unable to purchase anything, and were thus consistently dropping off.
Over the course of the incident, the company lost an estimated $150,000 in sales. A drop in the bucket for a major online retailer, to be sure. But how many micro-glitches like these, whose tracks are buried in mountains of big data from multiple internal and external sources, are never identified? How much revenue leakage do they represent over time, let alone during crucial periods like BFCM? And how poorly do such micro-glitches reflect on brand equity in the long-run? Is a customer who wants to buy, but can’t, likely to ever return?
To beat micro-glitches this BFCM, and throughout the year, online retailers need to learn to identify and rectify them in real time. This involves sifting through the massive amount of data that drives decision-making, and finding revenue-siphoning micro-glitches before they go rogue.
How can online retailers meet this challenge? To start, make sure that you’re closely tracking these five key parameters:
- Identify problems faster by watching exactlywhat your customers are doing. Track in aggregate exactly what customers are doing on your website or in your app. Watch for pages, page elements, or products that show sudden drops or spikes in traffic. For example, a run-up in sales against a drop-off in revenues could indicate product mispricing. Or, a drop-off in sales noticeable only for customers using a particular Android version could indicate a version-specific micro-glitch. The more granularly you look, and the faster you can identify problems, the more likely your response will be realistically impactful and on-track.
- Get the bigger picture by closely monitoring third-party data. Keep a close eye on competitor advertising bid data, weather data (which can cause physical store closures or online revenue-impacting power outages), fraud detection and security data, and any other data from third-party sources that can quickly and dramatically affect your business. You need to seamlessly connect all data sources and analyze trends in real time to understand how to best respond to findings from this data. For example, in response to a specific competitor challenge identified in third-party data, do you need to shift local inventory, change prices, promotions, bundling, or ad inventory bidding? In the old days, this used to be done by setting up a team of people to watch key KPIs [key performance indicators]. Today, no team can integrate enough data quickly enough to effectively respond—certainly not during the hysteria of BFCM. The solution needs to be technological.
- Listen to what people are saying on social media. Data from social media monitoring tools has too long been considered a vanity metric. The fact is, this data has real value to sales and marketing operations—but only if used correctly. Working in real time, correlate social media data with changes in product demand or revenue. For example, if a celeb overtly or inadvertently promotes your brand via social media, you need to quickly identify the actual business impact. That way, you can more effectively leverage the momentum in-store and online: adjust inventory to meet expected demand, tactically bundle products to grow basket size, and more.
- Cover all data sources, in-store andonline. Today’s most successful retailers are successfully leveraging both online and physical presence. The data generated by online and offline efforts are the building blocks of overall brand value. Retailers have robust systems for tracking in-store and online activity—yet too often this data is stored and analyzed in separate silos. To get the bigger sales and revenue picture, while heading off trouble from potential micro-glitches, you need to be able to combine, slice and dice this data in aggregate and in real time. By way of example, if a storefront runs out of a popular product over the BFCM weekend, but you know it’s available online, you can direct in-store customers to buy online, or shift stock locally to meet demand.
- Remember that DevOps and IT data matter only in how they impact the business. Integrating and closely monitoring data from DevOps and IT is a key first step toward finding the root cause of any micro-glitch. But DevOps/IT data doesn’t exist in a vacuum. Without the context of your business KPIs, you only get half the story. Is an alert from DevOps or IT during BFCM a priority, if it’s not impacting business KPIs? Is the sudden drop in sales of product X the result of a server problem or coding error? Or is it because a celeb dissed it on social media or stock in a strategically located store ran out? The answer is in the data, but that data must encompass the bigger business-technical picture that’s the key to effective online commerce.
The Bottom Line
This year, when preparing for a record-breaking Thanksgiving weekend, retailers need to keep in mind more paraphrased wisdom from Dr. Seuss: a glitch is a glitch, no matter how small. Be aware that revenue leakage from micro-glitches adds up fast. Adopt technology and workflows that enable you to identify and respond in near real-time. Neither Grinch nor Glitch—large, small, evil or just misguided—will prevent BFCM from happening. The question is, how ready are we?
Anodot provides automated anomaly detection systems and real-time analytics.