The wrong investment can lead to negative outcomes. Here's how to avoid that.
sponsored by Evolving Solutions
It's become widely accepted that incorporating analytics or Big Data into your business processes and decision-making creates differentiation. Many businesses find it critical both to respond to customer demands as well as to keep competitors at bay.
Once you've decided you need to incorporate analytics into your business, another problem arises: there's an embarrassment of riches when it comes to selecting analytics tools and platforms. Facebook, Google, IBM, Microsoft and other tech giants are offering solutions and some are providing advanced algorithms for free to the community. There are clusters, data warehouses, appliances for your data center and cloud based solutions from companies from traditional IT providers to startups barely a year in existence.
Identifying, selecting and implementing the right tools to support your strategy quickly becomes as challenging as the business processes and personnel required. Investment in the wrong tool plagues many projects, leading to unplanned expense and outcomes that don't meet stakeholder expectations.
There are no easy answers, but there are a few important principles that apply to analytics projects.
1. Begin with the end in mind
The sage advice from Seven Habits of Highly Effective People applies doubly to analytics and data projects. Determine what your business goal is first before becoming overwhelmed with your choices for how to get there. It can be useful to consider the innovative, disruptive things others inside or outside your industry are doing. It can spark ideas that you might not have otherwise considered, but it's easy to get lost in the tools and processes they used. Adrian Cockcroft, former Netflix Chief Cloud Architect said "People try to copy Netflix, but they can only copy what they see. They copy the results, not the process." Others chose their tools and process based off of the outcome they wanted. Your process, and therefore your tools, will be unique to what you are trying to accomplish.
Once you are clear on the desired results, it's critical to identify a "quick win", something small that can show tangible value relatively quickly. This serves three important purposes. It assures stakeholders that you're on the right path, it clarifies process gaps and identifies what data might be missing.
2. Know what you have and what you need
If this seems obvious, it is. But it's often overlooked. In a rush to innovate, it's not uncommon to neglect the assets you have available in your own digital history. If you have 5 years of sales data for swim wear in March and April, do you have enough information to determine which styles in what quantity to stock in Atlanta this year? Back to beginning with the end in mind. If your goal is 5% growth and that matches historic trends, you are probably set. What if your supplier base has changed completely? Historic data isn't irrelevant, but it's probably not sufficient. Are there call center logs, web chat logs, customer emails or comments available that could give you additional perspective? What else do you need? Where can you get it?
Take advantage of both the data and systems you already have. Often, a company has a business intelligence system or data warehouse that can provide a piece of what is needed. Development and data teams have tools for other projects that can be applied to the task at hand. Don't be afraid to challenge assumptions that may be holding the project back, but don't dismiss them out of hand, either.
3. When developing the solution, consider "where" when you consider "what"
As the team narrows down the tools and processes needed to support your project, be sure to consider the best location for your systems at the same time. If you've determined a Spark or Hadoop cluster is the best solution but you don't have any internal Linux skills, a cloud-based solution could significantly reduce your risk and time to value. A NoSQL database to hold your product catalog results might be simple to set up in the cloud, but if every interaction needs to talk to your on premise order management system, you may be better off with a local solution.
Again, there are myriad options, but the best choice comes down to the goal you set out, the incremental wins you're looking for and the assets you are leveraging. Whenever possible, it's also important to take reuse into account. That's a key value of taking incremental steps. A capability built to deliver one solution can be used by another more easily.
4. Taken together, these considerations point to a logical conclusion
The need to identify an important project and get a quick win, leverage what you have and integrate what you need, account for a variety of technologies and hosting emphasizes an important trend often called hybrid data warehouse or federated query. Those are just fancy terms to highlight the value of being able to ask questions of all of the data you need, regardless of location, to arrive at answers or output in one place. This is the key filtering the options and getting to the expected results from analytics and data projects.
Michael Downs is a Solution Architect for Evolving Solutions, a Twin Cities-based, full service technology solution provider delivering best-of-breed solutions, with exceptional technology serviced by expert talent. Our dedication to our clients is second to none. We take the time to fully assess your data and business objectives, develop customized software solutions based on your needs, and provide ongoing support.