Meet Your New Office Assistant: He's An Algorithm

Work relationships are crucial to employee performance which, in turn, affect productivity, efficiency, commitment, turnover and a host of other metrics that influence organizational outcomes.

A person's relationship with his or her first line supervisor has been reported to be the most significant in terms of overall job-related effectiveness, but culture, collaboration, cooperation, affinity with co-workers, among others, have also proven to be indicative of results related to customer satisfaction and other external impacts.

So, it is not surprising to learn that, as the saying goes, there is  now an app for that. Or more precisely, there is software that helps people better manage these relationships by sifting data that may identify opportunities to sell, increase efficiency or otherwise improve behavior related to output.

If this sounds a bit Orwellian, well, how is that different from credit card companies, social media and your favorite coffee shop or shoe store knowing the same things? The fact is that much of managerial progress now is based on data analysis designed to identify trends that may provide leads, tendencies, threats and options. The information, in itself, is neither good nor bad. How it is used and the degree to which it is  manipulated may have personal implications. These appear to be in line with broader societal trends, however unnerving they may be to some. The release of knowledge about government security monitoring created an uproar because of widespread fears of abuse that are strangely absent when it comes to almost identical private sector snooping to gain marketing advantage.

In the realm of office relationships, the question is more likely to be about effectiveness (does any of this really add value?) than privacy. The larger issue is what we may be losing by subcontracting yet another aspect of our interpersonal skill set. JL

Evelyn Rusli reports in the Wall Street Journal:

The next frontier for data is improving your work relationships.

Jon Porter, the CEO of private wealth-management firm Three Bell Capital, used to keep track of clients by manually typing information about meetings and leads with software from Salesforce.com Inc. CRM +0.48%

This spring, he got an algorithm to do the work.

Software from startup RelateIQ Inc. now looks at every digital scrap of Mr. Porter's work life—incoming emails, social-network contacts and phone calls—compares it with his colleagues' data, and figures out what and who is important. Two weeks ago, the algorithm prodded Mr. Porter to follow up on a time-sensitive question from a client.

"Had we not been on top of that, our client would have missed the window" for an investment, said Mr. Porter, who estimates he saves about two hours a week with the software.

Data scientists are beginning to peer into work relationships, trying to identify patterns that can improve how employees collaborate with peers, manage sales relationships, or see how they stack up against colleagues. It is a nascent market, but up-and-coming startups have their eyes set on upending established business-technology companies like Salesforce, which are also increasingly digging into data.

"We wanted to build an algorithm that could do what a highly trained relationship manager, with 20 years of experience, could do," said Adam Evans, a co-founder of RelateIQ, which has raised $29 million from investors including Accel Partners, Allen & Co., Battery Ventures, and Facebook Inc. FB -1.69%co-founder Dustin Moskovitz. The investment values the startup at $100 million.

How RelateIQ Works

• Users sign up, connect their email and relevant social-media accounts.
• Contacts and data are automatically pulled in from these sources, creating an address book.
• Users create lists, or project pages, where they can define objectives and add companies or contacts to that page.
• Users download mobile app to log calls and missed calls.
• As users interact with contacts, communications are logged automatically.
• Users decide whether to share their entire list, some of it or none of it with colleagues. In the context of a group, lists track items such as sales partners or recruits.
• RelateIQ's algorithm constantly collects data signals to identify whether relationships are cooling and if list members should be prodded to take action, such as respond to an email.

Source: RelatedIQ

Elsewhere, Boston-based Sociometric Solutions Inc. uses physical sensors to collect data on employees' movements and the tone of their conversations to tell managers where interactions are dipping and where employees are congregating. In San Francisco, tenXer Inc., a program for computer engineers, tracks code modifications and hours spent in meetings to help them see how their productivity stacks up against colleagues. And Boston-based Yesware Inc. helps employees track emails, monitors how many times their emails are opened, what devices recipients are using, and provides analytic reports on the email traffic of colleagues.

Though harvesting of such employee data may raise eyebrows of some privacy advocates—particularly in light of the recent debate over technology companies' involvement in National Security Agency data-gathering programs—the startups are emerging as a hotbed for venture-capital dollars.

"It's a trend toward a more transparent workplace," said Michael Abbott, a general partner at Kleiner Perkins Caufield & Byers, which is currently on the hunt for investments on the theme.

The idea is that software can detect patterns that humans can't.

Angus Davis, the CEO of online payments service Swipely Inc., used Yesware during his last fundraising round to determine which venture capitalists were reading his emails, how many links they were clicking and if they forwarded it to others in the office. "When I saw an email opened 30 times, I thought, 'Wow' they are interested," Mr. Davis said.

RelateIQ, which has been operating in "stealth mode" underneath a home décor store in Palo Alto Calif., for two years, is one of the most ambitious of the big-data work apps. Its software offers a central hub for work groups to see what their co-workers are doing and keep track of relationships in real time.

RelateIQ absorbs massive amounts of data—it scans about 10,000 emails, calendar entries and other data points per minute at first run—but does offer privacy controls for employees, such as the option to hide the content of email messages from colleagues.

Of the company's roughly 30 employees, four make up the data science team, a group with Ph.D.'s in statistics, fluid dynamics and physics. The team constantly tweaks the software to better identify patterns, such as the average time it takes for a person to respond and what types of punctuation and phrases typically elicit responses.

It also tries to detect sarcasm and words that are usually associated with important questions. These data can reveal, for example, whether a relationship is stagnating or progressing.

The data startups will have to tackle entrenched business-software companies.

Salesforce, which has more than 100,000 customers using its customer relationship management software, is increasingly mining data to provide recommendations. For example, its communication tool Chatter recommends people and topics users should follow, based on patterns in their past activity.

"Everyone is chasing us," said Kendall Collins, an executive vice president of Salesforce. "I've never seen another player in the market, at scale, with the level of social integration and the level of data enrichment."

Oracle Inc., ORCL -1.18%also a major provider of such software, declined to comment. SAP AG, SAP -0.29%another competitor, says it is increasingly working with relationship management startups to provide some back-end tools to process huge data sets.

RelateIQ, which has about 100 clients, says its software requires companies to enter in less information to extract useful insight. "We automatically capture everything that is happening with your relationships and surface only what you should worry about," said Steve Loughlin, RelateIQ's CEO and co-founder.

Another big challenge is identifying the right communication data and organizing it in ways that make sense to humans who, unlike computers, aren't very predictable.

"We have very complex communications environments," said Tom Davenport, a professor in management and information technology at Babson College who studies technology in the workplace. Machine learning, he says, can result in recommendations that seem to come out of nowhere.

Mr. Loughlin of RelateIQ says he's aware of the limitations of an algorithm to read actual relationships. For instance, software could interpret a long delay from a client as a negative signal, even if he was just out with an illness.

Because of the squishiness of human relationships, the company says it is careful about language. "We don't say you 'need' to follow up, we say 'we suggest,'" Mr. Loughlin said.