Who, Exactly, Is the BI User?
BI use cases and BI user constituencies have changed radically in just the last few years. Is that why BI adoption seems to be stuck?
- By Stephen Swoyer
- September 8, 2015
At the annual Pacific Northwest BI Summit in July, industry luminary Claudia Imhoff led a round-table discussion in which participants tackled that most intractable of problems -- namely, the seemingly lackluster rate at which business intelligence (BI) tools and services are actually used in enterprise environments. Imhoff's discussion wasn't so much concerned with lackluster BI adoption, per se, as with whom we mean when we talk about BI "users" or "consumers."
This "whom" has changed radically over time, Imhoff and other participants agreed, so much so that surveys of BI adoption don't -- and, in a sense, can't -- tell the whole story.
In too many cases, in fact, they amount to pointless exercises in comparing apples to oranges.
"The metric has remained constant, but the pool of users has changed. It's [remained constant at] 20 to 25 percent, but it's not the same 20 to 25 percent in the organization," argued industry veteran Suzanne Hoffman, vice president of business development with RadRac Industries, a sales management firm. Hoffman, who's worked for vendors such as the former HyperRoll (acquired by Oracle Corp.), the former Applix (acquired by IBM), the former Hyperion (acquired by Oracle), and Tableau Software Corp., has examined BI usage through several lenses. What she saw at Tableau wasn't what she saw (a decade ago) at HyperRoll, which specialized in OLAP acceleration.
"It used to just be the upper echelon of the organization, from senior vice president to director, it's actually shifting in terms of who the users are," Hoffman continued. "We've compressed the number of employees in most organizations, [so the proportion of people using BI] actually ends up being a better representative share of decision makers than it was before. The other interesting thing is that the rest of the market is expanding. [We used to speculate about] data visualization cannibalizing reporting, [but] that ... hasn't taken as much of a toll because the user pool has shifted."
The classic problem, as formulated by industry watchers such as Cindi Howson (now of Gartner Inc., and previously a principal with BIScorecard.com), Gartner, and others, is that BI usage can't seem to crack the 25 percent threshold. Survey after survey finds the pool of potential BI consumers stuck at between 20 to 25 percent of all employees (or, put differently, of "potential" consumers).
By far the biggest problem is that potential consumers might not even be aware that they're accessing -- let alone using -- BI reporting or insights. On the one hand, this is an indication that BI is actually succeeding -- i.e., it's embedded, more or less transparently, in the context of business processes. On the other hand, it frustrates the attempts of IT teams (or interested business stakeholders) to market the successful use of BI to C-level decision makers and other important stakeholders inside the organization.
To wit, Hannah Smalltree, director of product marketing with big-data-as-a-service upstart Cazena Inc., referred to a survey on BI usage that she recently helped conduct. "It was hard to [determine] the difference between accessing a BI tool or getting BI delivered via an application. [We asked] 'How much of your organization is using BI?' but we didn't ask it in a way that differentiated [on] that [basis]. I wonder if there's a ... trend of BI being more embedded in apps and biz processes, or [of users] who don't think of it as just going to a BI tool."
The availability of free and/or open source software (F/OSS) likewise complicates the issue, noted industry luminary Merv Adrian, a vice president with market watcher Gartner Inc.
Vendors such as InfoBright, JasperSoft, Pentaho, and Talend -- to name just a few -- market OSS "Community Editions" of their enterprise-grade products. Users are able to download and use this software at no cost. (In some cases, source code from these projects has been appropriated by or implemented in other F/OSS projects.) This is just the tip of the proverbial iceberg, however. From the "R" statistics environment and its vast ecosystem of libraries and services to a slew of BI-related Apache projects such as Kylin, Mahout, Lucene, Solr, and, of course, Spark -- these and other projects either explicitly address BI and analytics or incorporate BI/analytic functionality -- there's a surfeit of F/OSS BI alternatives. "[F]reely available open source software ... is used by people who don't contact people like us to ask what to buy and [by] people who aren't answering surveys. The use of this software is far higher than the dollars that we count up ... [there is] far more use of open source than paid software, and most of it is not measured very well," Adrian pointed out.
Just as the "whom" of the typical BI user is changing, Adrian argued, so, too, is the "what" of BI itself.
"I think we've moved from an era when [the term 'business intelligence'] just meant we did reporting. We called it BI even then. When we were writing FOCUS reports, we thought we were doing BI," Adrian continued, referring to FOCUS, a seminal mainframe-based reporting application developed by Information Builders Inc. "There was a lot of executive stuff, [such as] the early [executive information] systems, [and] the first dashboards were executive-focused," he said.
"I think ... people define [using a BI tool] as doing BI because I'm using a BI tool. If they're picking up Qlik or Tableau, they know they're doing BI. If they're accessing [BI] inside an ERP system, they're just like, 'No, I'm just using my corporate application, I'm not doing BI."
Glen Rabie, CEO of platform-as-a-service (PaaS) BI specialist Yellowfin, sought to call into question one of the most powerful ideas in contemporary BI -- that of self-service visual discovery. In point of fact, Rabie claimed, the potential pool of self-serving "analysts" comprises just a fraction of the overall pool of potential information consumers. In other words, the self-service model caters to a comparatively small pool of users -- let's call them the "techno-haves" -- that was frustrated by the shortcomings (or dysfunctionality) of traditional BI tools. True, visual self-service a la Tableau or Spotfire freed these techno-haves to do things on their own terms. But the vast majority of potential BI or analytic consumers would prefer not to have to interact with tools of any kind, Rabie argued.
"It's not even antipathy [to business intelligence as such], it's 'My job is not to be an analyst, my job is to run the business.' As vendors, we all pretend that [people] want to be analysts, but they don't. We think, if it's pre-packaged, we can give it to them, [they'll be] happy, problem solved," he said.