Enterprise app stores: Deployment tools, requirements goldmine

November 14, 2012

Anyone who’s ever clicked an “add to cart” button after reading an online review knows that such feedback can be a powerful motivator. For companies looking to give employees access to business intelligence (BI) insights via mobile apps, the kind of feedback that shows up in app store ratings is likely to be a powerful development resource, according to Michael Goul, professor of Information Systems at the W. P. Carey School of Business.

Goul teamed with Olivera Marjanovic from the University of Sydney Business School, Susan Baxley of Teradata Corp. and Karen Vizecky from Target Corp. to see if app store feedback might be useful for requirements engineering on corporate BI tools. The quartet of researchers concluded that, yes, BI is likely to be delivered to employees via apps, and the app store model is a probable means that companies will use to get it into workers’ hands.

What’s more, employees will probably praise or pan the corporate apps the same way they’d comment on apps purchase from Apple or Amazon. Since there’s going to be feedback, Goul and his team think companies should mine it and apply findings to fine-tuning app development. The team’s research into the subject indicates this approach could close the gap between BI developers and the employee who make decisions based on information delivered by analytics.

There’s an app for that … or will be

The researchers started their investigation with the premise that the consumerization of IT – where trends start in the consumer market and spread to the corporate world – combined with bring-your-own-device preferences, will be a driving force in corporate application development. Citing Gartner analyst findings, the researchers noted in a recent paper that “by 2013, 80 percent of businesses will support a workforce using tablets, 75 percent of business will plan for end users who will be using personal devices to connect to the enterprise network, and the majority of end-user devices will be employee-owned.”

Gartner analysts also have estimated that, with the growth in importance of business intelligence, the percentage of BI applications used on handheld devices would be in the 33 percent range by 2013. “SAP already has a whole bunch of BusinessObjects mobile apps in the Apple app store,” notes Goul. “There also are companies like Alteryx that specialize in analytic apps. It’s game on.”

Not only are BI tools likely to be deployed as apps, Goul maintains they’ll be distributed through corporate app stores. So do Gartner Analysts. They named app stores and marketplaces fifth on a list of 10 strategic technologies for enterprises in 2012.

Among the advantages of corporate apps, Gartner analysts cite security. If you build or deploy the apps yourself, you’re less likely to see employees downloading the malware that lurks in the iTunes store, Google Android Market and other app-filled places.

But, Goul sees another advantage to corporate apps that are offered via an app store: feedback and user ratings – and rantings. “There’s a bottleneck in business intelligence requirements engineering, and that bottleneck can be solved by using analytical approaches to examine feedback,” he says.

Proven, but bloated

According to the research team, requirements engineering equals the sum of all activities used to gather information about the requirements of a system or component, and the term implies a repeatable methodology. But the researchers also found “there is too much reliance on methods and techniques that are generic to all systems development.”

Goul explains what he means by generic approaches: “Before, you’d go around and hold focus groups and try to surface the mistakes and deficiencies in your software, or people would try to build in as much functionality as they could upfront before they implement something.” The problem with this approach is that it’s keyed to larger applications, not the small snippets of code that make up a mobile app. Because of their small size, apps evolve more quickly he says. The feedback mechanism of app stores works well to refine user requirements because spontaneous user commentary makes the feedback cycle quicker, more compressed,” he adds.

App store feedback is valuable for other reasons, too, the research team maintains, and several theories playing to the validity of online commentary as a requirements engineering tool. Among them are connectivity behavior, which outlines how people use mobile technology, and conservation-of-resources theory, which holds that people use tools that protect things they value, like time, energy and even their jobs themselves. The conservation-of-resources theory predicts that connectivity behavior will be high if apps deliver job value. And, if connectivity behavior is high, feedback is likely to be based on solid interaction with the app itself.

Other factors are at play, too. For one thing, people are really attached to their mobile devices. In fact, they’re emotionally attached. “The device has become a social actor – almost a person – in your life,” Goul says. Since people are attached to the devices, they also become attached to the apps those devices deliver. The more attached folks get to the apps, the more valuable their comments will be. Consequently, the researchers examined not just the sentiments expressed but also how embedded app usage was in a person’s daily routine.

Fine whines

To evaluate users’ app store comments as a potential tool that could be used by BI developers, the research team chose productivity tools to use as proxies for corporate analytic tools. In this study, the app of choice was Docs to Go for iOS users.

Next the team scraped comments on the app from the Apple iTunes store. A random sample gave the scholars 5,036 comments out of 23,434 that had been posted. To these, Goul and his team applied sentiment analysis, which also is sometimes called opinion mining. App store reviews turned out to be very effective mechanism for gathering users views, the team found.

“We derived that sentiment analysis is good way to tell what people truly feel about apps, and that it provides input about what next generations ought to take into account,” Goul says. In Docs to Go commentary, the researchers saw users asked for specific enhancements, such as the ability to pull email attachments out of in-box folders, edit Excel spreadsheets and drag folders onto the iPad from a laptop. With detailed suggestions like these packed into them, the researchers concluded that sentiment analysis on reviews could support requirements gathering and point developers in the right direction for app enhancements. The team used several sentiment analysis and clustering tools, doing much of their analysis using Clarabridge’s text analytics suite.

In fact, the researchers think sentiment analysis might have more value than focus groups in some ways. For one thing, focus groups query a small number of users, while app store reviewers just “pile on” their views even when those views have already been expressed by someone else. “This is an important point for sentiment analysis, Goul says, adding that it means reviews could be mined as an input for a crowdsourcing approach to requirements gathering.

Goul does, however, have a caveat on using app reviews for requirements gathering. “With bring-your-own-device trends, users aren’t discerning between business and pleasure use of apps,” he notes. “One guy complained about an app not working on his Yahoo account. As a business, I might not care so much about his personal Yahoo account, but knowing exactly how users are using the apps could be a challenge for organizations.”

Still, he thinks the sentiment-mining approach has validity, even if it might mean some uneasy transitions for IT professionals. “If I’m a developer and I see that feedback every day, I know everyone else is looking at it too,” he says. “That means my whole job is a lot more transparent than it used to be, and my job function is more open to market forces.”

Bottom line

  • Enterprise app stores are on the rise, and they’re likely to be one way companies deploy business intelligence (BI) tools to workers.
  • Just as users post reviews in Amazon or the iTunes store, corporate workers are likely to post comments about the BI apps they’ll download and use.
  • Researchers tested whether sentiment analysis could be used to find out what people think of enterprise BI apps and support requirements engineering.
  • Study findings indicate that app reviews do contain enough valuable information that they could be used to help software developers fine-tune corporate apps.