Through The Looking Glass

secretlondonglasses_previewFinally, it happened. I had to get my first pair of reading glasses a couple of months ago. I got the lightest pair I could find given what I had read and heard about comfort levels.

I need them to read and read only. I am absolutely fine watching TV or moving about the house without them. In fact, the lenses distort the perception for anything beyond reading. What surprised me were some of the behaviors I developed.

As time moved along, I realized that I was leaving them on for some other activities. Even though the view was distorted, I was not conscious of this and went about my business. This happened until something made me realize the glasses were on and I removed them.

This got me thinking! What kind of glasses do I carry around with me in life? All perception is necessarily through the lenses of my collective experiences, but what distortions and prejudices have crept in that I am not conscious of? How aware do I need to be to have an open mind?

Applying this to a professional setting, do our built-up biases unconsciously drive our actions and decisions? What needs to happen for us to realize that we are carrying this baggage and for us to unload it? Mind you, this is not about ignoring all previous experiences. This is about identifying and filtering the parts that are appropriate in a given situation. This is about identifying whether the biases are taking us away from where we want to be.

Being aware of this seems to be the first step in building the self-consciousness to identify the biases. This will allow us to be more open to input as well as contrarian evidence. Having a confidante on the team who constantly challenges your leanings and status quo would also help. But, what helps most is our desire to do the right thing.

So, 9-to-whatevers, time to identify the lenses and figure out how to undistort the view? Let me know what you think.

Getting from Data to Decisions

This blog is the outcome of thoughts, discussions and interactions which began with the last two posts on The Velocity of Information Part 1 and Part 2.Given that information needs to flow into the decision matrices and the speed is important, the thought has been dubbed with this title.

The insights obtained from these discussions resulted in the below chain of thought.

  • Data as an end state is of no use. It needs to be converted to a usable format. i.e. Information. This conversion is best done by the people who are close to and understand the data
  • Information as an end state is of no use. It needs to be analyzed and insights created. These creations are motivated from the perspective of  the results. In other words, the analysis and insights need to be drawn (or at least vetted) close to the decision makers
  • Insight as an end state is of no use. It becomes an academic exercise if it does not flow into any decision matrix. Remember, a decision to do nothing is also a decision
  • It is difficult to predict which piece of information will be relevant to which decision a priori. Therefore, the appropriate approach would be to de-couple the information creation and the information use stages. These two stages would then be connected through context

As we increase the velocity, the idea is that decision making will be more effective. The context of the decision will drive what information/insights get into the matrix and the increased velocity will improve the coverage and recency aspects.

So, the question is what does this mean. We need to dig deeper into 3 areas.

1. Information Creation

There are a plethora of tools available for this step comprising the first two layers of the proposed model. The “Business Intelligence” world is exactly about being able to extract information about data sets (which are getting ‘bigger’ all the time). The majority of this world revolves around structured data and structured analysis, but unstructured data analysis is beginning to come into its own at this point. There are strengths and weaknesses in this area which need to be addressed, but there are already several threads on that.

In my mind, the importance here is to ensure that the information extracted is rational. This means accurate, timely and correctly categorized. Categorization relates to defining the context(s) within which a particular piece of information could be useful along with the standard tagging/metadata pieces.

2. Information Delivery

This is also a critical stage of the process. This is represented by the cloud (layer 3) in the previous posts. However, this layer defines the contextual language, provides for connecting the suppliers of information to the buyers of information, is the plumbing in the scheme of things. In essence, it is a marketplace to make decision-making more effective.

3. Insight Creation

At the heart of decision making is converting information to insight represented by layer 4 in the model. This is a process which is completely manual and is often the basis of what people mean when they say “out-of-the-box”. This layer will need to have bright, experienced people who understand the context. But, the quality and coverage of the information coming into this is very important. A missed indicator or a false positive can throw the whole process out of kilter.

Therefore, the ability of these “experts” will be to create the correct contexts. What they need to depend on is the quality of information they get when they send a context into the Information Delivery system or the aforementioned cloud.

It seems to me that it is quite imperative that organizations look at what can be done to improve the velocity of information and make decision making more effective. De-coupling, as above, probably makes for a smoother implementation; technically, as well as organizationally.

However, the opportunity here is not necessarily limited to internal implementations. The power of this can be stretched a little bit. How many data-providers do we have in the world today? Can they improve the offerings by converting to information before they publish? The analytics can be done of their own accord or could be added as per the specification of the client. Further, they could publish the output (non client specific information) to somebody who runs an information mart! The information mart sets up the contextual language and provides fully baked information to clients who need it; the information could be a mix of the clients internal analytics as well as external open market feeds. The possibilities here are extremely intriguing…

So, how many of the 9-to-whatevers would be OK to take a plunge into something like this? Comments? Thoughts?

The Velocity of Information (Part 2)

There have been a few interesting offline discussions on this topic – thanks to those who helped shape the below thoughts.

Here a visual representation of the framework.

The Layers

The Layers

So, what are the thoughts driving the framework at this point?

  • The output desired here is a framework and not a product roadmap. Hence, the thoughts are limited to principles with an attempt towards completeness
  • The solution is not looking at a BI tool. While these tools today (including Big Data tools) are sufficiently advanced, they still represent only the bottom layer of this framework
  • Work should be done as close to the source as possible. Hence the ownership of the data, analysis and accuracy needs to be at the information curation layer. These concerns need to be satisfied before hitting the cloud
  • The two processes of generating the information and using the information for decisions need to be de-coupled. The impact of this will be shown below

If we look at the Analytics layer, we see several disparate data sets. This indicates that we could be creating information within disparate silos. As an example, the operations department could provide the information on production data, the marketing department on sales data; the macro-economic data for the geography and any relevant news tidbits could come from a completely different data set.

The silo generating the information should not limit themselves in knowing who needs the data a priori. Anybody within the organization should be allowed to access the information.

Similarly, the decision curators should not limit themselves to knowing the range of information that they would like as inputs a priori. A context search within the cloud should give them all that is relevant within the context. Then, their function would be to filter what is relevant from the results available. As an example, the decision on a product mix within a particular geography would not necessarily approach the legal department to determine if any new laws are expected. This structure would allow that information to be assimilated automatically. Don’t we all want to increase the serendipity within our organizations?

The basic underlying idea is to increase the velocity of information by getting it to the right place at the right time. This will also improve the return on investment in the creation of information; reduce duplication of datamarts and analysis across silos; improve the speed at which information gets incorporated as well as increase collaboration across the organization.

Do the 9-to-whatevers think that the objectives here are desirable? Do you think that this framework will move us in that direction? The time has now come to start thinking beyond the basics into some level of detail on the layers and the interactions between them. Do you have any thoughts you would like to share?

The Velocity Of Information (Part 1)

Decisions are made everyday. Actually, they are made several times a day by most individuals. This is also true within a corporate setting.

So, what are these decisions based upon? When we talk of taking “informed” decisions, they are typically based on information. Crisis management apart, decisions are typically taken after careful deliberation of the available information. The information may not provide a complete picture, but the more coverage you have, the better the decision.

How does information get to the decision makers? In yesteryear, careful paths were laid down to elicit this information. In a top down approach, the data to be analyzed was identified, the framework for analysis was agreed, and the data flowed from the point of data availability to the point of decision making. In this journey, it was collated, checked, summarized, categorized, sliced and diced, ad nauseum until it reached it’s destination. The time it took to reach the decision maker often made the information irrelevant, or at least resulted in lost opportunity. For information to be useful in decision making, it has to be timely.

Then, the technique advanced. Vast stores of data were made available close to the decision makers. They could apply the frameworks necessary to elicit the information in a more timely manner. However, the volume of data grew resulting in incorrect and irrelevant information being included in the decision. For information to be useful in decision making, it has to be relevant.

Also, in todays world, there are multiple sources of information and multiple decision makers to consume it, making pre-defined flows difficult to identify and implement. The question is what can be done to get relevant information in a timely  manner to the decision makers, even when they might not know they want it. This is what I would like to discuss and build on. My initial thoughts follow…

I am not big on industry jargon, but three things which are big today might contribute to the solution (not necessarily in the way they are understood currently though!). In my proposal, I create four layers. However, a lot of thought needs to be put in to how this will function. Early days yet…

1. The information creation layer represents the “big data” gang. However, this can be decentralized. This layer extracts information based on the data. The rules and frameworks used to extract information are in the hands of those who are closest to the data.

2. The information curation layer. These are the people/systems that validate and categorize the data for accuracy and relevance. eg. eliminating co-incidences and non-causal co-relations.

3. The “cloud” layer. This allows for multi-point to multi-point flow of the information.

4. The “decision maker curation” layer. These are people/systems who subscribe the information from the cloud, establish relationships and curate it for the decision makers.

Will this allow us to increase the Velocity of Information allowing for better quality decisions? What to the 9-to-whatevers think?