Every smart business owner and corporate leader is talking about big data. It’s a buzzword that has been beaten into the business community over the past few years. There are more than 1.7 million results for “articles about big data” online.
In fact, it’s in danger of being overhyped. There’s been so much of a focus on big data collection, which poor data analysis is threatening to undermine the reputation of big data in corporate decision making.
Contextual data needs to become the new buzzword, because it is far more powerful and effective. And helps to facilitate more competent data analysis.
What is contextual data?
Contextual data is information that has been gathered and analyzed without isolating it from the broader scope of information. Isolated big data is the opposite – it’s information that has been gathered and then analyzed in a vacuum.
Could big data be saved from overhype by contextual data?
Think of contextual data as the white knight riding in on a horse to save the bloated, isolated business intelligence systems that simply fail to live up to the hype. Humans generate more than 2.5 quintillion bytes of data every day. It’s literally impossible to analyze all this information in real-time or analyze past information without missing evolving insights highlighted in new information.
The first ingredient in updating your data analysis and collection efforts with contextual data is investing in systems that instantly collect and transfer data to your data storage systems. Stop thinking of data analysis as a weekly or bi-monthly process. Focus your data engineers on providing micro-insight based on targeted data that is fresh – collected within hours of analysis – within the context of your historical data.
Can dark data be contextualized?
Dark data, or information that has been collected and archived without comprehensive analysis, represents the vast majority of information that governments, businesses and organizations currently possess.
There are two reasons that this data is turned into dark data. First, it may be that too much data is coming in. This isn’t a bad thing, but it’s an economic and technological reality that only so much information can be thoroughly analyzed before newer information replaces it. Second, some data is deemed irrelevant to current business objectives. In these cases, analysts look at this data as a good addition to the archive, but irrelevant to the needs of their current projects.
The nurturing of contextualized data involves taking recently gathered data and comparing it to historical data. By its very nature, this type of data curation involves reviving dark data – the kind of information that only becomes relevant as the needs of decision-makers changes.
For retailers, contextual data is there for the taking – everywhere!
One of the industries that most benefits from contextualized data is the retail space. Every time a customer shops or interacts with a brand, they leave breadcrumbs of data behind. Modern POS software can compile contextual data that includes basket size, promotions leveraged, and demographical information about the individual that completed the transaction.
This data can be used to personalize individual customer experiences by predicting future needs of consumers. For example, Amazon looks closely at how consumers behave on their platform. Based on historical interactions, Amazon will alert consumers that specific items have gone on sale.
One of the most exciting aspects of this involves predictive shipping. In other words, Amazon will send items to warehouses located close to consumers that they believe are likely to purchase the items. This dramatically reduces shipping times and increases customer satisfaction.
Regression Analysis Helps Turn Down the Noise and Focus Data Collection Efforts
In fact, there’s so much data flying through company data systems that it can be hard to understand what is important and what can be ignored. Regression analysis helps companies leverage their investment in big data storage.
There are two types of regression analysis – linear regression and multiple regression analysis. Both involve using “independent variables and “dependent” variables.
To isolate the type of data that is most important to today’s business objectives, data analysts use a sample of data to try and identify causality. In some ways, it’s the data analyst equivalent of going fishing. You make an educated guess, and then throw the numbers up on the board to see if they change in a way that proves a statistical relationship.
Consider this example: If you can statistically prove that there’s a relationship between the number of billboards your company rents, and the number of customers entering your store on a daily basis, it might be a good idea to investigate which types of billboards are most effective. Should your marketing budget be used to invest heavily in highway adjacent billboards, or do you want to shift some dollars into public transit billboards?
These types of answers wouldn’t be possible without tapping into previously dark data and then analyzing it based on the focus of today’s decisions.
Most data gathered by online customer interactions is already contextualized.
The good news is that creating contextualized data doesn’t have to involve sending a data analyst fishing. Instead, you can leverage data that’s already contextualized. For example, a purchase transaction has multiple variables tied together – items purchased, customer data, total spent and coupons used. There’s also data on payment instruments – here’s a hint: There’s a reason that every store you shop at encourages you to sign up for their credit card. They have contextual data that proves customers who own a store charge card will purchase more, and are significantly more likely to shop their store before the competitors.
In conclusion, contextual data is critical to competent decision-making in 2018. Don’t let go of all that dark data – even if you’re tired of paying data storage costs. You never know how future needs will change the relevance of old data – and there’s no such thing as too much historical trend data.