Data.
A word that keeps slithering its way into business models the world over. I have been in business long enough to know the ins and outs of business operations. To make them work, you constantly have to iterate and have the flexibility to adapt to the morphing landscape of the market dynamics in your industry.
Enter, data. How can we use it as a measure for the effectiveness of our business models? There are a few underlying elements that we must adhere to. To establish the certainty that the data we are using is in fact helping our cause, and not hindering it.
1. Data Is “a” Resource, not “the” Resource
Like any other commodity in a supply chain, data is not “the” resource, but “a” resource. For traditional supply chains (raw material to finished product), you need to look at each link of the supply chain as a cohesive part of the whole. Hence, any rupture or anomaly in a link will disrupt the efficiency of the final build.
Rightly so in the data supply chain. I recently took the opportunity to analyze some numbers in the airlines and aviation industry. An industry colleague shared his access to Flight Global open portal data source. They have information on airlines worldwide fleets and their distribution throughout the world. You can read the report that contains that analysis here:
Will Bombardier CSeries Jets Ever Takeoff ?
First and foremost, the major and primordial assumption to my analysis is that this data source is one, up to date, and second, accurate. I had no way to dig into the metadata (data about data) to research the source and its validity. This assumption is cited in my report.
After some exchanges on twitter with some airlines and aviation enthusiasts who were questioning the report analysis, I decided to dig further. I went into a database sourced directly on an aircraft OEM (Airbus) website. It referenced the worldwide fleets of a certain segment of commercial aircraft (single aisle narrowbody jets with 100-150 seat configurations). The results are posted in the blog entry cited above.
The reality was that the data sets referenced were not congruent. Even when applying the same restrictions, assumptions, and filters, they were not aligned. Each analyst has their own methodology for data interpretation. Whether the data results be 1% off, or 28% off, the important thing to remember is that they were NOT the SAME. Different source, different data. Hence, the data feeding the supply chains are skewed somewhere along the line.
If the methodology applied to two or more different raw data sets (that are reporting on the same metric) is the same, and the results are different, then it is safe to assume that the inputs into the development of the raw material are different. As in my analysis of the Airbus In-Service Single Aisle narrowbody world fleets with 100-150 seats configurations.
The data sets that I applied my ideology and methodology to were in essence reporting the same thing. One set was from Flight Global, and the other from Airbus website. However, once the analysis applied, the output was different.
This leads me into the importance of making sure the sources of data in your analysis and business plans are verified as accurate and true. To do this, there are a few rules to follow:
2. Get as Close to the Source as Possible
Make certain that the raw material feeding your analysis and business plan is not skewed or has been altered in some way, shape or form. An anomaly at the trail head of the data supply chain could skew the final results (and along the data pathway) by magnifying it substantially.
3. Always be Auditing the Data Supply Chain
As in any other supply chain governance model, the raw material inputs sources and quality must always be audited in perpetuity. Should one of the elements change, the integrity of the end product will also be affected.
4. Validate and Manage Assumptions globally
Often enough we get so involved in our work and data models that we can lose site of the very reason of why we are building them. Always be focused on the end result. Also, involve all the stakeholders from the outset and govern the process and build with checkpoints along the way. Make sure that all stakeholders question the process, and data validity from the outset and during the process.
By getting different points of view, inputs, and questions, the teams will be able to hypothesize on different scenarios to establish congruency on which one gets the best vote. There will be discussions and interpretations. This is the driver of success. Look at all the options. Analyze, verify and govern them. Select and proceed with the option that gets the best vote from participating stakeholders.
In my opinion those are the main underlying principles of making sure your project, analysis, business model are going to have a solid build with the least amount of noise and variance as possible.
Thanks for reading. Comments, ideas and questions welcome.
