Recently, I was talking to a high profile client in Kennedy world about database marketing, segmentation and data buckets, and I said 'You know how PRISM works right?' and INSTANTLY he said, 'YES, YES, I want to run my data through your data prism'. Soon the conversation turned in to a project and I am working on dissecting his list(s) in to multiple slices to create segments and sub-segments of data.
From 1989 and now, this is what I have done. I have taken large data-sets and sliced them in to smaller data segments so the marketer (copywriter, designer, content creator) can create marketing assets that would resonate with the audience.
I went on Amazon.com to find a real prism, and here's what I found.
I spent some time talking about these four crystals with my team members and explained to them how database marketing and data science works in the life of a small business. We spoke about for key concepts: data sphere, data pyramid, data prism and data cube.
In this blog post, I will explain these four key concepts. Most data scientists who do data analysis and data science might find these concepts a bit underwhelming and the shape comparison a bit childish. I found that most business owners who make between $1M and $5M and don't have a background in data science, statistics or data modeling find this language comfortable and easy to understand the complex world of data science. So, here we go. I hope you like it!
The Data Sphere
I learned this technique from a friend of mine who was head of data science at Reed publishing. He asked me: "Parthiv, what is data to a business? We talk about data all the time, but what is data? How can we control something we are submerged in? What is air to a human? What is water to a fish? How does a fish explain the concept of water to another fish?" He continued, "You got to look at data from multiple perspectives. You see, data is a point that is connected to a string of other data points and the string is part of a sheet that is made up of other data strings that are made up of other connected data points. And the point, the string and the sheet look differently from every vantage point. In order to analyze the data, you got to observe data at the spherical from every vantage point concurrently, and not just look at the data point but the string it is part of and the sheet that string is part of. The data as an absolute point is useless". I actually understood what he meant, and we then went in to some math theories on how the formula for analytics will change as you shift the vantage point.
They say 'I don't have a crystal ball' when they can't forecast the future. But crystal ball is actually a data sphere that gives you access to data point, string and sheet from multiple vantage point so your guess is better than someone who is not looking at the same data from multiple vantage points. The astrologers and psychics depend on common math and deductive reasoning to arrive at the guess work they sell. They don't have any super-human powers. They are math producers and magic performers.
As a CEO or CMO of your organization, you are submerged in data. It is hard for you to see the thick band of data that is around you. However, it is important to know that the data exists around you and data will make very loud noise before arrival of money and after departure of money in and out of your life. More important, that data noise also leaves an ever lasting trail so one can go back in time, export elements of data, study the data noise around those data points, data strings and data sheets and arrive at reasonably accurate, conclusive decision as to why money moved the way it moved.
Here is why this matters: Money will move AGAIN exactly the same way, when the same data conditions are reenacted. When you know WHY money moved in certain way, you can recreate that scene in your life again if results were desirable and avoid such a scene if the results were undesirable. From that perspective, the data sphere is a time machine of a sort.
The Data Pyramid
Dan Kennedy talks a lot about Pyramid. Examples are "Moving money up the pyramid" "Top of the pyramid marketing" or "When the base is compromised" are some of his signature topics of discussions. Why does he talks about Pyramid? What is so interesting about that geometrical shape that we never paid attention to after we left middle school? What does a Pyramid has to do with business?
In my opinion, he talks about a Pyramid because a Pyramid is a multi-directional triangle so you can look at upward movement of data from every vantage point. Your business has entry and exit points for data from everywhere (except bottom) and based on an RFMp based Customer Value Matrix (one that measures Recency, Frequency, Monetary Value and Propensity To Refer) you can determine your bottom 80% of pyramid and top 20% of pyramid. At the top, you can further segment your data based on top 1%, middle 4% and bottom of top 15% giving you 4 bands of data: 1%, 4%, 15%, 80%. Once you plot your current clients on the 1%-4%-15%-80% scale, you would plot your alumni (where would you place them if you did not lose them?) and based on the placement of current clients and alumni you can identify prospects that match and mirror your 1%,4%, 15% and 80%. Based on this segmentation you can determine which prospects will get Fed-Ex, which prospects will get direct mail, which prospects will get emails and which prospects will not be touched until they self-approach you. Data Pyramid helps you allocate and budget time, money and energy on specific prospects and help you create a well thought out marketing mix based on desired results.
The Data Prism:
Prism is one of the weirdest shape we studied in school. We know that prism is a 3 dimensional triangle shape that has a unique ability to intake a white ray of light and display a multi-color 'rainbow'. Rainbow is called rainbow when water droplets act as a prism in the sky that splits the ray, so the rainbow is displayed.
So, what is a data prism? Data Prism is your MYSQL or MS SQL algorithm that looks at your historical data points (and strings and sheets) and analyzes your last 100 days, last 300 days, last 1000 days or your last $1M or $5M in transactions and analyze them on RFMp based Customer Value Matrix.
Yes, you do need a qualified DBA (database administrator) to build and run your prism. You would export all data from all sources and import in to SQL. You study the record layouts of all sheets and do some basic data hygiene to spot data contamination or data truncation issues. You also take an inventory of what is called 'Known Data Issues' so you can compensate for them. You run some 'quick search' queries to build confidence and compensation for your hypothesis. (Confidence = how confident am I about what I see) and (Compensation = How do I adjust the query knowing ___ is true/false). Once you do the basic data analytics you are ready to dissect your transnational data through prism and begin arriving at trend spotting and decision making through deductive reasoning.
The Data Cube
Your data cube is where all your data lives. This is your data warehouse. This is the backbone of your pixel-estate. HubSpot gives you a superb Data Cube where your prospect and client can enter your Pixel-Estate, a gated campus of built environment and are under constant surveillance while they are inside your data cube. They can enter your crystal data cube at will from any data point on the surface of the data cube, they can leave your data cube if they wish from any data point, they can go up and down, side to side or front and back and they can but while they are inside the cube all their movement within the cube and consumption of content is tracked. For example, as you are reading this blog article, a system is keeping track of your consumption of this knowledge so when you talk to a human at eLaunchers.com, they can have an intelligent conversation with you knowing what you already consumed. We prefer HubSpot to be the data-cube of choice because it is more inclusive compared to other systems, but one could connect multiple systems (Wordpress, Click Funnels, Active Campaign, Infusionsoft/Keap) to simulate HubSpot's data-cube. We jokingly call it 'FrankenSoft' :-).
