Venkatesh Rao’s recent post on The Real Reason Enterprise Search is Broken struck a chord with me. A very discordant chord.
Venkatesh writes that while the technical challenges faced by enterprise search are daunting, they can be solved with enough engineering talent. He thinks the real reasons for failure lie in organizational behavior.
I desperately want him to be wrong about office politics undermining enterprise search, but I suspect he’s right. I’ve experienced it.
In a previous job, I served as a consultant deploying enterprise content management systems. The company was making a big push to get their software installed on every workstation. Getting the software onto more computers meant more employee licenses meant more money.
Most end users disliked the content management system. Putting stuff into the systems was time-consuming and difficult process. Unless managing corporate records or documents was your job, it competed with all the other work you had to do.
The enticement for registering material in the content management system was that it became searchable and sharable. That’s nice, but many employees had worked out some method of finding stuff on their own computers, however inefficient. And you can always share stuff by sending endless streams of email back and forth, however much people say they hate that. No, the real draw of enterprise search was being able to find other people’s stuff.
My job was to battle the adoption problem. You needed a large base of employees contributing material to build the network effects. You needed enough other people contributing their stuff to entice you to search and contribute your stuff. If you didn’t get a critical mass, the system wouldn’t work.
So you’d think that the software company I worked for would have invested massive amounts of time and effort in making the search and sharing features as attractive and usable as possible, right?
Wrong. The search and sharing features sucked.
It took me several years to figure out why.
more users = more licenses = more money equation seems simple enough, my old software company knew that the secret to sales was getting upper management to sign the check. And their reasons for paying for a system are quite different than employees’ reasons for using the system.
At one company I worked with, the biggest champion of the content management system was the corporate counsel. She wanted all important records and documents collected in one system. This would bring the company into compliance with Sarbanes Oxley and a host of other government regulations. Public companies are required to keep records.
But public companies hate being sued. And they really hate the “discovery” process by which lawyers pick through corporate records looking for misdeeds. An ideal content management system, she explained, would be one where you could add records (as required by law) but never get them out again (to keep the company out of trouble).
At another company I worked for, a senior leader wanted more information about far-flung satellite offices. He had responsibility for service delivery and quality control, and wanted to monitor everything. His position and authority depended on him being “in the know”.
But it also depended on others not knowing the things he did, or being able to draw their own conclusions from his data. He particularly didn’t want these satellite offices coming up with their own ways of doing things, without help from headquarters and from him. An ideal content management system would vacuum data directly from all his service techs machines — ideally without their intervention — and piped directly to him for analysis. Only he or members of his staff were to have access to the information in the system after it had been entered.
For these check-signers and management champions, the fact that our enterprise search sucked was a feature, not a bug.
Most organizations simply aren’t ready to adopt the radical transparency and decentralization that pervades the web, even if they accepted these things as worthwhile goals. Many wouldn’t think it necessary. And some will actively resist it.
The same features that engage knowledge workers may actually put off those in management. This inevitably leads to compromises in the design and implementation of enterprise search tools.
And if enterprise search isn’t comprehensive, a lot of what makes it compelling as a solution is lost. It becomes yet another repository that must be consulted and cross referenced, in addition to all the other partial solutions that already exist within the organization. It will add to the noise, rather than reduce it.
Venkatesh is surely right that if your organization’s culture is hostile to information sharing, or even ambivalent about it, your enterprise search initiative is likely to fail. You’ll be forced to make compromises that undermine the system. And there are plenty of consultants and software companies that will help you sabotage your effort. You’ll get a broken system, because on some unconscious level, that’s what your company wanted.
But if there’s a shred of hope, it’s this: Not all corporate cultures are hostile to information sharing. Some enterprises thrive on it. And politics and culture can be changed. It just isn’t as easy as buying and installing software.