Building the perfect analytics stack: It’s now about choice and customization

Several years ago, we had a television network client who, the morning after their show aired, would eagerly ask, “How did I do last night?” What she meant was: How did the network’s shows perform the previous evening?

But answering that basic question wasn’t so simple a few years ago. Why? Because inherent in that one simple question were a bunch of others that unraveled into a complicated narrative exercise. And, of course, what she really wanted, after all that exercise, was something that boiled down to: “You had a good night,” or maybe, “Not so good, but here’s why.” Coming up with that answer required that we arduously compile information from a bunch of different tools and mash them together, quickly, into what she needed to know.

[aditude-amp id="flyingcarpet" targeting='{"env":"staging","page_type":"article","post_id":1981595,"post_type":"guest","post_chan":"none","tags":null,"ai":false,"category":"none","all_categories":"big-data,business,marketing,","session":"C"}']

That arduous process vexed everyone. Unsurprisingly, as a market grew for answering those questions, companies tried to build all-in-one solutions, setting off a wave of mergers and acquisitions. Salesforce bought Radian6, Oracle bought Collective Intellect and so on. The result was a bunch of very large dashboards of information resources that try to be everything to everyone.

It’s understandable how that happened, and it has benefited countless industry professionals. But, for many media companies that we deal with, it also has become increasingly obvious that one size of data dashboard doesn’t (in fact, can’t) fit all of their respective needs.

As a result, a significant trend has emerged that is headed in the opposite direction. That trend has many implications for networks as well as for the vast array of digital creators trying to become networks — as well as for the vendors that want to provide all those companies their data.

Call it Build Your Own Stack, and more importantly, build that stack from best-of-breed, highly focused vendors that provide exact pieces of what your company needs. I’d say that every major media company we talk with is already building its own customized collection of data services, in one fashion or another.

About half of those companies are building their stack in a rather old-fashioned way (or at least, old-fashioned for a business that’s only a few years old). That means taking our dashboard of information, and the dashboards of other analytics providers, and manually collating information from each one separately to build a coherent answer to “How did I do?”

In another 30 percent to 40 percent of cases, a media company will rely on aggregators, such as Tableau.com or Domo.com, that provide a sort of meta-platform combining all the a la carte data sources that a company wants, or at least the ones that are compatible with that platform’s requirements. That can be effective, and is certainly simpler than juggling a bunch of separate interfaces from a bunch of vendors.

Then there are what I’ll call the technologically aggressive 10 percent — media companies that are building their own data teams and analytics platforms to pull in data from diverse sources. These custom dashboards further integrate all those streams automatically to create a very sophisticated and company-specific answer to “How did I do?”

It’s these latter two groups that are forcing analytics companies to focus on two key areas: portability and research.

[aditude-amp id="medium1" targeting='{"env":"staging","page_type":"article","post_id":1981595,"post_type":"guest","post_chan":"none","tags":null,"ai":false,"category":"none","all_categories":"big-data,business,marketing,","session":"C"}']

By portability, I mean each analytics company needs to identify its true strengths and then make it easy to integrate that information with other data sources, whether through aggregators or a custom platform. That means creating application program interfaces that simplify getting all of what I have here into what you have there. Without portability, analytics firms may be left behind as custom stack-building comes to dominate the business.

But growing portability means you have to conduct continued research and development on your products. Because clients will have far more choices in which data, and data providers, they include in their perfect stack, it will be more difficult for companies to rely on institutional inertia and technology lock-in to keep clients using their products when better ones come along.

The real power of any analytics company lies first in its deep mastery of a specific sector, and then in building on that mastery with continued learning. It’s the only way even the best tool can remain competitive over time, especially as stack-building becomes a more fluid process.

Customers like our aforementioned client still need a simple answer to “How did I do?” but that’s where a lot of analytics companies fail. They think what they do is so important that the user on the other side should care about them. But clients don’t. Data only become insights when the data can be tied to business outcomes. They care about the answer, a defensible and clearly understandable answer. The provider of that answer matters far less to them.

[aditude-amp id="medium2" targeting='{"env":"staging","page_type":"article","post_id":1981595,"post_type":"guest","post_chan":"none","tags":null,"ai":false,"category":"none","all_categories":"big-data,business,marketing,","session":"C"}']

Companies that are part of the new stacks need to make sure they can simply answer their part of that question — and integrate their part smoothly with other companies answering other parts.

All of this, of course, may just set the stage for another round of M&A deals in a couple of years as the pendulum swings back toward concentration. But it’s clear that being all things for all clients often leads to a not-completely-satisfying product for just about everyone. So go deep, keep learning, and play well with others.

Jared Feldman is founder and CEO of Canvs, a language analytics technology platform created to measure and interpret emotions.