The Practice Yields the Process

In my MBA class this past semester I introduced the Enterprise IT (EIT) module about a third of the way through the semester.  I explained that we’d be studying information technologies that let business leaders impose new ways of working —  new processes — on their organizations.  To get them thinking about what lay ahead, I asked them something like:

"When should you deploy Enterprise IT?"

For the most part I got what I was expecting —  answers included:

"When you have to comply with regulations like Sarbanes-Oxley"
"When you want to be sure that a process is done the same way everywhere"
"When you want consistent data throughout the company"
"When current systems are legacy spaghetti."

One of my students was Polish, and so presumably had some experience with authorities imposing their ideas.  He said simply "When you’re sure you have the right answer."

This caught me up short because it was so true and so glaringly obvious, but I’d never thought of the issue that way.  Like a lot of people involved with EIT, I’d been carrying around the unexamined assumption that if smart people spent time thinking about how they wanted a business process to flow, they’d get it about right.  I’d seen this approach work so often in the EIT efforts I’d studied (for example, at Cisco, CVS, and Evergreen Investments) that I’d started thinking of it as the default for introducing new workflows.

But what if you don’t have any good idea what the new workflow should be?  Or you’re not sure what the critical considerations are?  What if you’re trying to do something novel, something for which EIT vendors don’t yet have good templates?

In these circumstances, do you simply have to take your best shot at pre-defining the right processes, then embedding them in Enterprise IT?  Or do you have to give up on the idea of  using IT to facilitate amorphous or highly innovative processes?

In their wonderful book The Social Life of Information, John Seely Brown and Paul Duguid make a distinction between the process —  a cross-functional sequence of tasks that need to be executed in order and with efficiency —  and the practice.  A practice is a community of people who have something in common:  an interest, a job description, a body of expertise.  Whereas the process crosses borders, the practice stays within them.  

Seely Brown and Duguid argue that current corporate IT does a much better job supporting the process than the practice.  For the past dozen years, businesses interested in improving their processes have been able to select, adopt, and exploit ERP, CRM, SCM, eProcurement, and all the other flavors of Enterprise IT.  

Businesses interested in improving their practices, meanwhile, have had what?  Knowledge management systems were sold and bought on the promise that they’d strengthen communities of practice, but have had a dismal track record.  As I explain in my recent Sloan Management Review article, this failure stems in large part from the fact that KM systems presupposed (in other words, imposed) the structure of the knowledge they were intended to capture and disseminate.   They were in fact another flavor of EIT.  

So have any information technologies really helped foster communities of practice?  Adam Cohen’s book The Perfect Store covers the early history of the global phenomenon that is eBay.   Founder Pierre Omidyar believed deeply in fostering community, and in not imposing lots of process up front.  Cohen writes "Omidyar’s routine when he received an email with a complaint about another user was to respond to the author, send a copy of the email to the other person in the dispute, and tell them both ‘You guys work it out.’"  After doing this for a short time, Omidyar had the great insight make the discourse among community members, both positive and negative, part of the community itself.  The process of leaving feedback and the member ratings that are so central to eBay’s success are the result

I saw something similar when I studied Alibris, the online intermediary for rare, used, and out-of-print books.  The company started as Interloc, a simple inventory listing service for professional used bookdealers, who could post the books they had for sale and check if any of their colleagues had a book they were looking for.  Only after watching this technology support the practice of booksellers for over four years did Alibris executives decide to try to support the process of bookselling.  They  built an eCommerce engine and shut off the old listing service.  

The new system was clearly an EIT —  it assigned users into roles (e.g. buyer or seller), arranged tasks into sequences, assigned decision rights (for example, Alibris started monitoring fulfillment and gave itself the right to delist dealers with low fill rates), and increased all parties’ interdependence.  And it succeeded far, far better than most other independent eMarketplace / B2B exchanges.1  Alibris learned about the process by watching the practice.  The practice was visible because it made use of what I call a Network IT platform, or an online forum where participants can engage in unstructured interactions.  

Two of the biggest NIT platforms on the Internet currently are Craigslist and Wikipedia.  Craigslist is almost certainly the world’s largest and most popular source of classified ads, and Wikipedia is the planet’s biggest reference work.  Finding fault with both of these platforms has become a cottage industry (and some of the critiques have merit), but the fact that they’re both enormous and continually growing should surely tell us something.  Either millions upon millions of people are being duped  by these two sites, or they’re demonstrating real value to their users.  I suspect it’s the latter.

From the start, both Craigslist and Wikipedia have benefited from and facilitated strong communities of practice.  And at their start both sites cared very little for process.  Like eBay and Alibris in their early days, they were largely freeform environments; they allowed people to come together, but didn’t tell them what to do once they got there.

Over time, however, the leaders of both Craigslist and Wikipedia decided that they needed to institute some processes.  As I used Craigslist a couple weeks ago to unload an old TV, I learned that I couldn’t post the same ad in both the ‘free’ and ‘furniture’ areas of the ‘for sale’ section.  I also learned that I couldn’t delete my ad then instantly re-post it in order to have it show up at the top of the chronological listing.

I was trying to game the Craigslist system in some of the most obvious ways possible.  And after watching countless similar attempts, the leaders of Craigslist decided sometime in the past that it wasn’t in the best interests of the site to allow such shenanigans, so they put in place some automated checks on the behaviors.  In other words, they watched the practice to learn about appropriate processes, then embedded these processes in software.  

Wikipedia has done precisely the same thing over its history, and it has also codified a lot of its rules in words, rather than in code.  Wikipedia has extensive sections on official policies, guidelines, and proposals that contributors can refer to.  It’s easy to contribute to Wikipedia without being aware of any of them; they serve primarily to settle disputes or questions that have come up repeatedly in the past within the community.  

So it’s clearly possible for a freeform online community of practice (what I call a pure Network technology) to evolve some processes (and so become something closer to Enterprise IT) without losing its soul.  What’s critical, it seems, is for the leaders of the community to observe it over time and keep asking themselves what rules, policies, constraints, workflows —  in other words, what processes —  would be helpful.

eBay, Alibris, Craigslist, and Wikipedia show how smart business leaders have migrated their organizations’ emergent structures to imposed ones.  The impositions aren’t abrupt or made in a vacuum; they’re done for demonstrably good reasons.  When the transition from emergence to imposition is managed this way there’s really little tension between the process and the practice.  In fact, they reinforce each other with when they come together with a bit of technology and a bit of leadership.


1Alibris was sold to Oak Hill Partners in May of 2006.  It still operates as an independent eMarketplace for books.