Top 5 Reasons Not To Roll Your Own
Okay you have a crack data engineering team that knows all the open source big data tools inside out and can build anything. Why not roll your own system to deliver data insights to the business about the behavior of customers, users and devices? Won’t that cost less and be more tailored to your business? Well, no.
Here are the top 5 reasons you shouldn’t, regardless of what technology route you are considering for your homegrown analytics project—this applies whether your taste runs to Hadoop, Cassandra, Druid, Spark, Redshift, Google BigQuery or any other open source or cloud data engine out there, and whether you’re planning to build your own reporting interface or grab something off-the-shelf.
1. End User Empowerment
Your users should be able to explore and ask new questions on their own. Homegrown systems rarely manage to offer more than limited filtering and pivoting of metrics and dimensions that were manually defined by development teams in advance. New questions usually mean asking for development of a new report and often takes days or even weeks or months to deliver. By the time the business gets a new report they usually have had to act without data to keep things moving.
Homegrown projects will never empower end users to ask completely new questions about behaviors and get back answers in seconds through visual interfaces without coding. Your end users will be hostage to the availability of development resources to inform new kinds of decisions and your development resources will be burdened with ad hoc requests. Ironically, this is an area where homegrown will result in less ability to adapt to your unique business than using a highly flexible and powerful commercial solution like Interana.
2. Technology Superiority
Only Interana has a visual interface tightly coupled with a back end designed to answer behavioral questions completely on-the-fly with no coding and speed-of-thought query performance, so users can ask lots and lots of questions iteratively. Its patented technology encompasses innovation all the way up and down the stack. Every other approach requires explicit coding to model new behavioral constructs and nearly all require significant batch processing behind the scenes resulting in delays to answers to new questions at scale. There is no way to put this power behind an off the shelf reporting interface without losing speed, power or flexibility to end users.
No matter what you are told, there is no way for a homegrown project using any type of open source or cloud data warehouse back end and either off-the-shelf reporting/BI front ends or homegrown interfaces to deliver the speed of asking new questions about behavior to an end user through a visual interface that is necessary to match the power of Interana—at any scale.
3. Time to Value
Interana can be up and running against your existing data pipelines in weeks or even days. Most integration projects for building a data warehouse for event data and integrating or building a front end interface take 6 months or more before the first report—we’ve seen many take 18 or 24 months.
You can’t afford to go another month without a data-informed culture in today’s competitive environment. There isn’t time to tinker.
4. Time to Insights
Time and again we hear customers who used to have custom event data infrastructures tell us they used to wait a week for a report to answer a new critical business question because of a combination of the backlog in engineers queues, the time to develop and debug custom queries, and the processing time for the queries themselves at scale. Then when the report that comes back generates follow-on questions or needs refinement, it’s another week or more for the next round. More often than not the business gives up and flies blind.
You need a system that lets users answer dozens of questions in minutes not one question in days.
5. Ongoing Innovation
The state of the art of visualization, query building, types of behavioral questions and dashboarding is constantly advancing. The scale continues to grow. The integration points keep evolving. A commercial vendor like Interana is invested in ongoing R&D and committed to staying at the forefront of analytics. Homegrown tools get stale and even companies with the best intentions tend to let development teams atrophy over time as other priorities compete for scarce resources.
You can’t afford to fall behind as the state of the art of data-informed culture advances.
Fundamentally, delivering behavioral insights to the business today is not about engineering a solution to deliver specific metrics efficiently in pretty dashboards. It is about empowering teams to ask lots of new imperfect questions about behavior easily and quickly so that they use data all day every day. This is the heart of the movement toward a data-informed culture and it is why Interana and interactive behavioral analytics is fundamentally different than what you get when you roll your own analytics.
Oh and what should those engineers be doing instead? How about some machine learning and AI work that are value-add within the service you provide itself rather than internal tools.
If you’re interested in learning more about Interana, be sure to reach out!