Wednesday, March 30, 2011

Inventory Control and Business Analytics

In this age of cost cutting measures, every aspect of business operations must be considered.  It is in business functions that have a high degree of variability that cost control can be a difficult endeavor, but yet provide a tremendous opportunity for savings.  One of these business functions is the management of inventory.  Inventory can be a physical or abstract concept.  It can represent real raw materials or products ready for sale.  It can also represent intellectual property or data that may have storage costs and a limited shelf life.  The general management of inventory, however, comes down to this trade off: If you have more than you can use, then there are storage or disposal costs.  If you don't have enough when a customer comes knocking, then there is lost revenue.  In rare cases inventory has low storage and disposal costs so excess quantities are kept on hand for future use.  In  most cases though, whether due to fads, styles, age, or market forces, inventory perishes and has a carrying cost.

Managing a complex base of inventory is difficult and costly.  One tool that can support this effort, however, is business analytics.   Business analytics offers a  framework for lending insight to trends, inefficiencies and and opportunities for the  management of inventory.  Since inventory control is a highly variable activity, analytics can help understand the drivers and magnitude of this variation.

Several different analytic methods are available to provide this insight.  The most simple and straightforward method for understanding inventory levels is historical reporting and analysis.  This will illustrate periods when inventory was either deficient or in excess, and will clearly show seasonal or business cycle variation.  It can also identify poor or successful inventory decisions that have been made in the past; these can inform better future decisions.

A historical perspective does not necessarily give full insight to current inventory decisions, however.  Real-time monitoring of current inventory levels with key metrics such as Current Inventory, Working Capital, Backlog and End of Life Inventory provide an illustration of where inventory may by actively lacking, in excess or at risk.  Through regular review of these metrics, real-time inventory control decisions may be made to mitigate potential future waste or lost opportunity.

Without insight into future needs, however, current inventory monitoring has limited benefit.  In order to truly stay ahead of demand for inventory, a reliable and accurate inventory forecast is essential for proactive decision making.  The nature of forecasts, however, is that they become less certain the further into the future they predict.  As such regular update and review of forecasts is required to allow for re-alignment of future inventory requirements.  Proactive inventory management is effectively a bet on future needs.  This bet needs to be balanced with the variability inherent in the inventory forecast: The bigger the forecast uncertainty, the smaller the bet, and visa versa.  The net effect is an optimized inventory program.

This article just begins to touch on the analytic methods that could support the management of inventory.  The actual techniques will vary depending on business needs and business objectives.  It is these analytic techniques allow for the control of uncertainty and variation that is so present in inventory management.

Wednesday, March 2, 2011

Social Media Analytics

How can you use social media to improve your marketing reach?  By working with an online marketing strategist, you will be able to define a program that best suits your business.  But how do you KNOW that your social media campaigns are effective?  That is where social media analytics are essential.

Social media analytics have various components, none more important than the other, that are all interconnected, informing and driving your social media campaigns. 

At the root of these analytics is, of course, data.  Most data related social media campaigns are "machine" data (see previous blog post) - data about some human activity that is captured by a machine .  For example, if you were click on the "Share" or "Like" buttons at the top of this page (go ahead, click them!), Facebook and LinkedIn machines would create data about this blog post: who liked, when they liked, what they liked, why they like (if they left a comment), and who they know that "liked" before them.  A tremendous amount of information for a single button click!

But what do you do with this information?  How can analytics help?  When thinking about social media analytics, one must consider three questions: What campaigns have you done (that work)?  How are your campaigns doing right now?  How do you expect campaigns to turn out?

The first question "What campaigns have you done (that work)?" goes beyond basic reporting.  It requires in depth statistical analysis that ties a particular campaign or marketing channel all the way back to earned revenue (or equivalent business metric if revenue is not a key business objective).  It is only through this linkage that you will be able to determine how successful a campaign has been.  Naturally this is easier said than done.  But going through the effort of linking relevant corporate data sources (marketing, sales, accounting) to make this connection provides exponentially more value than simply reporting on arbitrary (and self-fulfilling) key performance indicators (such as Tweets this month).  Despite all the "noise" that comes in between a Tweet and cashed check from a client, the right statistical analysis can tease out those campaigns or channels that contribute to increased revenue.

The next question of  "How are your campaigns doing right now?", gets at the "proper" use of key performanace indicators.  Through the statistical analysis of the previous question, you will be able to determine the relative effectiveness of various social media campaigns (on revenue) and work backwards from revenue goals to set smart and informed campaign targets (i.e. how many tweets, blogs and friends do I need to drive my revenue target?).  The collection metrics that are expected to contribute to revenue should be monitored in concert, against targets or goals, as no single campaign will drive revenue alone.  By the way, revenue should be monitered with these metrics as well.  Afterall, this is what we are ultimately trying to impact, right?

The last question, "How do you expect campaigns to turn out?" requires the modeling of social media campaigns or channels before they are initiated, or while they are in progress.  This will allow you to tweak or turn the dials of the campaign initially or mid-stream to influence the expected (revenue) outcome.  This type of decision support is critical to making smart spending decisions relative to social media campaigns.  The analytics involved in this effort include predictive modeling of revenue for a particular or group of campaign strategies, marketing resource optimization models that allocate budget and staff in a manner that maximizes (revenue) impact, and adaptive models that will rechart a course when a campaign gets off track or is derailed by unexpected events.

With the right analytics strategy of "looking back, being present and looking ahead", you can take the guess work and leaps of faith out of your social media campaigns.  Facebook does not have to be a "huge waste of time" as Betty White quipped in her Saturday night Live monologue.  It can, with social media analytics, be a powerful tool!