Management Information, Analytics and Reaction Latency

Why delivering "the right data, to the right people, at the right time"
isn't as simple as some would have you believe.


Humans with their inbuilt data processing capability can react to external stimuli, sometimes, within milliseconds e.g. the reactions of a formula 1 racing driver.  Other reactions to events can take significantly longer  e.g. governmental response to macro economic cycles. However in each of these examples the decision making process is the same: identification of an event, collection and analysis of the facts (data), understanding that action is required, and finally, an action implemented.

Business decision making follows exactly the same principles, although decision makers are normally assisted by Management Information Systems. These systems capture the occurrence of business events: record, classify, organise and analyse the data, then present it as consumable information.  When optimised, these systems can deliver reliable information to help form a business decision.  All this however takes time.   The time between an event and the business decision maker reacting to it is called Reaction Latency.

Reaction Latency is the time it takes for a decision maker
to understand and react to a business relevant event.

Reaction Latency is important  The slower a decision is taken the lower the business benefit achieved.  It is important to understand and reduce the latency inherent in all business information systems and decision making process.

Latency can be attributed to different parts of the management information system and decision making process i.e. data acquisition, data analysis and decision making.  The latency associated with each results in a loss of business value as shown below.

Note: Implementation Latency is dependent on individual
business dynamics and so is not discussed further here.


To explore Data, Analysis and Decision Latency further:


 Data Latency

Data Latency is the time between a business relevant event
taking place and information about that event being acquired and
 recorded within the management information system.

In the past, data acquisition was relatively simple and accurate with the majority of business relevant events being triggered within the business e.g. The receipt  of a customer order (or lack of an order) etc.  With the growth in social media, big data and the "Internet of Things", the recognition of a relevant business event is increasingly occurring outside the organisation; increasingly closer to the customer. When capturing data from outside the organisation, extra care has to be taken regarding its validity, provenance and accuracy.    The risk of poor data quality needs to be balanced against the advantages of early recognition of relevant events.  Data latency is reduced and competitive advantage increased  the closer to the customer an event is recognised.

The focus has to be on understanding how a business relevant event can be identified (as close to the customer as possible) and then how it can be accurately captured within an organisations management information system.

Data Latency - In Summary:


Analysis Latency

Analysis Latency is the time it takes for event related
data  to be processed  into consumable information
and presented to a business decision maker.

Information system architecture has significant influence on analysis latency.  The infrastructure architecture has to allow for quick recording, analysis and information presentation with minimum delay and probably minimum movement of data (and associated risk of data loss).

Historically, analysis latency (particularly within accounting systems) has been limited by financial and management accounting processes. Many business events are only recognised on a monthly basis.  This necessitates continued emphasis on reducing accounting close times plus re-thinking how financial records are finalised i.e. weekly, daily or even "real" time.

The growth of in-memory databases has improved the opportunity to shorten analysis latency.  At present only a limited number of organisations have invested in such technology (though increasing as the relative costs decrease). In-memory data processing has also stimulated the growth of predictive analytics (faster algorithm processing).  Predictive analytics is increasingly used to recognise events and/or trends that may lead to changes in future business outcomes.  In response data capture systems must flexibly recognise and retrospectively reclassify events, not originally thought relevant.

Analysis Latency - In Summary:


Decision Latency

Decision Latency is the time taken between the presentation
of information to the business decision  maker and the
 decision maker understanding action is required and why.

Decision latency is less of a technical issue and more of a psychological issue. Decision makers need information that minimises the cognitive effort required to understand when action is needed.  If the cognitive effort is too high then important information may be ignored or, at worst, misinterpreted. 

The designer of the analytics interface must understand the techniques required to engage and hold the attention of decision makers.  This is achieved  by  delivering very concise analysis and information. Irrelevant or difficult to assimilate data should be avoided  (avoid being data rich and information poor).  The objective is to minimise the decision makers cognitive effort when interacting with the information presented.

Decision Latency - In Summary:



"the right data, to the right people, at the right time"
  isn't as simple as some would have you believe.

All the above could be summarised in the familiar saying, "the right information, to the right people at the right time".   Contrary to the enticing simplicity of this phrase, there is no single solution can addresses Reaction Latency and it's three sub latencies (despite vendor efforts to sell systems solving all three at once).

Data, Analysis and Decision latency need to be understood and addressed in  order to reduce Reaction Latency.  In my experience looking for one solution to solve all three at once is unrealistic and leads to failure.  Each has to be addressed individually, applying the correct level of data management, technical infrastructure and psychological expertise as required. 

Conclusion - In Summary:


Business Process Analytics : Michael zur Muehlen & Robert Shapiro

Originally downloaded on 3/8/2015 from :

Get a free giffgaff Sim