Applying the right approach to unlocking the secrets in your data can avoid getting carried away with the latest data fads. Building a data lake when all you really need is a tidy pond will not only waste money and time, it can flood your business with unnecessary infrastructure and ongoing costs.
According to CIO magazine, companies who rate themselves substantially ahead of their peers in the use of data are three times more likely to rate themselves ahead on financial performance.
But for most of us, data remains locked in silos, trapped in spreadsheets, legacy databases, and incompatible formats. And even when data is accessible, often it can be unintelligible, hiding its important insights in eye-glazing rows of numbers and fragmented matrices.
So how can you get to this data goldmine that is essential to being competitive in the big data age?
“Big Data,” “Data Lake,” “Unstructured Data,” “AI” – these are terms that get us excited, promising easy fixes to unlocking our valuable business data and providing instant insight.
But there is no quick fix. Getting carried away with the latest data fads can lead to enormous expenses with little to show for it. Building a data lake when all you really need is a tidy pond will not only waste money and time, it can flood your business with unnecessary infrastructure and ongoing costs.
These six steps can help you make sure you are taking applying the right approach to unlocking the secrets in your data assets, without blowing up the piggy bank.
Rome wasn’t built in a day. Similarly, great data-focused companies started by tackling a small, important problem first. But they also had a vision of where they wanted to go. Netflix has used its data strategy to perfect its content programming; Oracle to enhance its lead generation pipeline. Whether its improving operational efficiency, boosting sales, and innovating new products, data is at the heart of today’s successful strategies. Pick your area of focus, and then start with a small, well-defined project.
Let’s say you want to use your marketing and sales data to understand your most successful marketing campaigns. Your first step is to determine what data you’ll need, what state it’s in, and how you will match it up between silos. Unfortunately, most of us don’t have data schemas for all of this just sitting on the shelf. Now’s the time to create one: export the data sitting in all those SaaS applications and internal databases, and then map the data coming from the necessary sources as you would chart the seas before setting off on an intercontinental voyage. Some data is unstructured? That’s fine – try applying some off-the-shelf AI or neural net software to see what taxonomies could emerge. You’ll soon discover that the taxonomies and data tables you need to hold even the simplest data fields are not the ones you might have thought.
Once you know how the data is structured, start with one simple KPI or metric, and map it all the way through to how you’d like that metric to be displayed to an end user. Don’t take a complex metric that builds on several other metrics (cost of customer acquisition, product recommendation, etc.) – start with a straightforward one, like say, average click-through-rate or maximum purchase value by customer type. Pull the data together, match the fields, put it in a useful data store, and display a chart that an end-user will find interesting. Show this to the end users until they tell you you’ve got it right. Excel may suit your purpose fine, here, or a simple sql database tied to a BI tool; that’s ok.
Now that you’ve seen how the data needs to flow, get a list of all the KPIs you want to present. Prioritize this list into three tranches – metrics we can’t live without; those that provide useful insight to everyone; those that someone will want to see but others won’t need.
Now you are ready to build version one of your dashboard that focuses on how to display tranche one. You won’t need to worry about tranches two and three until the end of this process.
How will you proceed? Start by taking each KPI and determining where the data behind it will come from, how it will calculate, and how it will stay updated. Then you have a real understanding of what it will take to get there.
Now you are ready to display your metrics. This is where the UI design and data visualization techniques come in. You may soon discover that certain dataviz tools don’t display the data in ways that are required for the users to get a full insight. Or maybe there are simple tools out there that do all that you need. Select the least expensive approach that makes your data instantly understandable.
Most people want to start their data journey here at visualization – and that’s a mistake. Not until you’ve actually wrestled with data and how you will get to your KPIs can you pick the appropriate visualization tools. Doing it the other way around will leave you in agony as you realize that the tool you’ve chosen doesn’t work with the data formats you have.
Finally, we need the highway to bring the data to the masses – your customers, or your employees. Now you are finally ready to bring in the programmers to examine the best data management and information architecture. Perhaps you’ll want a cloud-based micro-services architecture. Perhaps you just need a simple mysql database. Perhaps, in fact, you do need a Hadoop big data store to harness millions of data-points streaming in by the second. But by this point, you have a solid understanding of how your data needs to expand, as well as the visualizations needed to make it useful. With the addition of the right architecture, you are now prepared to expand the data management platform to contain the totality of your project.
Whatever you do – don’t chose your technologies until you get to the appropriate step. Starting to build your data warehouse before you’ve mapped your data is an exercise in wastefulness. Exploring your data through this process will force you to make decisions on which technologies to use at the appropriate times, when you have the right understanding of the issues presented by your data.
Standing at the beginning of your data journey, the road ahead can look intimidating. Juxta Digital has helped hundreds of companies map their data, determine their KPIs, and build the right visualization and architecture to make their data apps sing. Don’t go into the unknown alone when we can guide you to success.