Menu

IoT FEATURE NEWS

Needles in the Data Haystack Can Prick You: The Necessity of Real Time Anomaly Detection

By Special Guest
Anna Jones, Special Correspondent
April 06, 2017

Today, businesses collect all kinds of data about themselves, and they do this to detect glaring problems and discover hidden opportunities. Those problems and opportunities often appear as anomalies - data points that are unusual or unexpected – which tell you something has changed in that very specific aspect that you've been monitoring. It makes perfect sense then that detecting these anomalies as soon as they occur is vital for businesses when information can ripple across the globe in an instant.

Stemming the Bleeding
Many anomalies are caused by problems affecting (directly or indirectly) your revenue stream. The longer those problems go unfixed, the more money is lost, and this just as true for dramatic anomalies as it is for more subtle anomalies. Dramatic anomalies cause bigger losses, but they are also easier to spot because they cause such large deviations in your key performance indicators (KPIs). Subtle anomalies however, can go undetected for much longer because they’re much harder to detect in the data. And a subtle anomaly that persists for a long time can do just as much damage as a drastic, but brief one. 

These types of under the radar subtle anomalies often impact just one facet of a particular segment of your operations. For example, a drop in in-app purchases on your mobile game, but only from iOS users—and only those iOS users whose operating system version is below a certain number. 

A pricing short-circuit
Sometimes the problem is triggered by a specific sequence of events, like the price glitch that hit Circuit City in 2007 which allowed online shoppers to grab a discount on Nintendo DS games of up to a third off an already discounted price. Price glitches like these might be very specific in the type of product affected, but the instant communication speed of the Internet amplifies the effect of the price glitch by rapidly increasing the number of transactions which take advantage of the glitch. It only takes one (pleasantly surprised) online shopper to share the glitch on a web forum.

From that one mention, other forum members try their luck, and when they succeed, they tell everyone else on the forum as well as on their social network feed, increasing the feeding frenzy. Soon, this stampede reaches the critical mass necessary to be picked up on major online news sites (in this case, Engadget) and now entire classes of people who never would have bought video games from Circuit City join the flood of new shoppers who’s only purchases are ones which take advantage of this absurdly low price.

Price glitches like these are pretty common in ecommerce. Had Circuit City been using real time anomaly detection, not only could it have stopped the price glitch snowball from becoming an avalanche, but it also may have been able to detect and adjust to the more fundamental market changes which ultimately pulled its plug (pun intended). Although it’s too late for Circuit City to take advantage of real time anomaly detection, it’s not too late for your business. 

When a blip is not just a blip
Real time anomaly detection can also help you make money, not just alert you that you're losing tons of it. For example, a small bump in sales revenue from Canada following a new product launch originally meant to fill an unmet market need in the southeastern United States may signal pent-up demand for your product across the whole North American region. Targeting this untapped pool of potential customers with an effective continent-wide marketing campaign could generate a wave of new sales revenue.

Humans just can’t detect anomalies as fast as businesses need to react to them. This is even more true when you factor in the volume of business metrics (sometimes numbering in the millions) which need to monitored for anomalies. As real time business incident detection and analytics firm Anodot explains, this is why real time anomaly detection requires an automated system: you would never be able to hire enough humans to meet these twin requirements of speed and scale.

Get the alerts, then set the priorities
Besides satisfying a crucial need for businesses to save and make money, real time anomaly detection also allows you to better triage the anomalies that are detected. This is because you can always postpone action on an instant alert, but you can never act instantly on a delayed alert. This is a key advantage of employing real time anomaly detection. Detecting anomalies in real time can help your business save and generate money, enabling more efficient operations and giving you a much needed competitive advantage.

Oh, and it’ll also catch those fluke discounts on video games.




Edited by Ken Briodagh


SHARE THIS ARTICLE
Related Articles

IoT Evolution World Week in Review: Intel, Nokia, Telefonica

By: Ken Briodagh    6/24/2017

Welcome to the IoT Evolution Week in Review, my friends. This week, we've been talking about global cybersecurity, the IoT Evolution Expo, healthcare …

Read More

IoT Time Podcast S.2 Ep.33 Trusted Computing Group

By: Ken Briodagh    6/22/2017

On this episode of the IoT Time Podcast, Ken Briodagh sits down with Steve Hanna, chair of TCG's embedded systems and IoT work groups, and principal a…

Read More

IoT Time Preview: The Cloud

By: Ken Briodagh    6/22/2017

In this weekly series, we'll be previewing chapters of "IoT Time: Evolving Trends in the Internet of Things" for you to read in the hopes that you'll …

Read More

IoT Evolution Speakers: Lee Gruenfeld Speaks Out

By: Ken Briodagh    6/22/2017

Lee Gruenfeld is Managing Partner of the Cholawsky & Gruenfeld SaaS/IoT consulting firm, and is a Principal with the TechPar Group in New York, a bout…

Read More

IoT Evolution Speakers: Jeff Liebl Speaks Out

By: Ken Briodagh    6/22/2017

Jeff Liebl is President of Anaren IoT Group, which designs, manufactures and sells custom high-frequency solutions and standard components for the wir…

Read More