Balancing Act: Capturing Knowledge Without Killing It

10/16/2000 Process-based reengineering and practice-based knowledge management are profoundly different, say Xerox P !"#s John Seely Brown and Paul Duguid$ %ut rat&er t&an c&oose between t&em, successful managers need to strike a balance$ 't#s a delicate art, t&ey say( one t&at begins wit& )knowing w&at you know$) by *o&n +eely %rown and Paul ,uguid "Process and practice," write John Seely Brown and Paul Duguid, "do not represent rival views of the organization. Rather, they reflect the creative tension at the center of innovative organizations."

The anager!s challenge, they say, is to "eep #oth approaches in view at the sa e ti e. "$or in the delicate art of #alancing practice and process lies the eans #oth to foster invention%#y allowing new ideas to spar"%and to furt&er it%#y i ple enting those sa e ideas." &ere, Brown and Duguid show how processes that support how people wor" can #enefit fro new perspectives on what they already "now. -nowing .&at /ou -now 'dentifying a company#s best practices is not easy, for a couple of reasons$ 0irst, t&ere#s a large gap between w&at a task looks like in a process manual and w&at it looks like in reality$ +econd, t&ere#s a gap between w&at people t&ink t&ey do and w&at t&ey really do$ ctual work practices are full of tacit impro1isations t&at t&e employees w&o carry t&em out would &a1e trouble articulating$ 2&e manager w&o wis&es to understand t&e company#s best practices must bridge bot& of t&ose gaps$ 2o illustrate t&e difficulty of identifying best practices, we#ll look at t&e customer ser1ice representati1es w&o fix Xerox mac&ines$ 0rom t&e process perspecti1e, a rep#s work can be described 3uickly$ "ustomers &a1ing difficulty call t&e "ustomer +er1ice "enter$ 2&e center, in turn, notifies a rep$ 4e or s&e t&en goes to t&e customer#s site$ .it& t&e &elp of error codes, w&ic& report t&e mac&ine#s state, and documentation, w&ic& says w&at t&ose codes mean, t&e rep diagnoses t&e problem and follows instructions for fixing it$ Practice &ere would seem to in1ol1e little more t&an following t&e map you are gi1en and doing w&ate1er it tells you to do$

't would seem t&at way, if someone &adn#t bot&ered to look more closely$ *ulian 5rr, formerly an ant&ropologist at Xerox#s Palo lto !esearc& "enter 6P !"7, studied w&at reps actually did, not w&at t&ey were assumed to do$ nd w&at t&ey actually did turned out to be 3uite different from t&e process we#1e 8ust described$ 2&e reps# work is organi9ed by business processes, wit&out a doubt$ %ut t&ey succeed primarily by departing from formal processes( t&ose processes followed to t&e letter would soon bring t&eir work 6and t&eir clients# work7 to a &alt$ 0or example, t&e company#s documented repair processes assume t&at mac&ines work predictably$ /et large mac&ines, made up of multiple subsystems, are not so predictable$ :ac& reflects t&e age and condition of its parts, t&e particular way it#s used, and t&e en1ironment in w&ic& it sits, w&ic& may be &ot, cold, damp, dry, clean, dusty, secluded, in traffic, or ot&erwise$ ny single mac&ine may &a1e profound idiosyncrasies$ !eps know t&e mac&ines t&ey work wit&, 5rr suggests, as s&ep&erds know t&eir s&eep$ .&ile e1eryone else assumes one mac&ine is like t&e next, a rep knows eac& by its peculiarities and sorts out general failings from particular ones$ "onse3uently, alt&oug& t&e documentation gi1es t&e reps a map, t&e critical 3uestion for t&em is w&at to do w&en t&ey fall off t&e map;w&ic& t&ey do all t&e time$ 5rr found a simple answer to t&at 3uestion$ .&en t&e pat& leads off t&e map, t&e reps go $ $ $ to breakfast$ .&en t&e <oing <ets 2oug&$ 5rr began &is account of t&e reps# day not w&ere t&e process 1iew begins;at nine o#clock, w&en t&e first call comes in;but at breakfast before&and, w&ere t&e reps s&are and e1en generate new insig&ts into t&ese difficult mac&ines$ 5rr found t&at a 3uick breakfast can be wort& &ours of training$ .&ile eating, playing cribbage, and gossiping, t&e reps talked work, and talked it continually$ 2&ey posed 3uestions, raised problems, offered solutions, constructed answers, laug&ed at mistakes, and discussed c&anges in t&eir work, t&e mac&ines, and customer relations$ %ot& directly and indirectly, t&ey kept one anot&er up to date about w&at t&ey knew, w&at t&ey#d learned, and w&at t&ey were doing$ 2&e reps# group breakfast s&ows t&at work goes on t&at formal processes don#t capture$ %ut it s&ows more$ 't demonstrates t&at a 8ob t&at seems &ig&ly independent on paper is in reality remarkably social$ !eps get toget&er not only at t&e parts drop and t&e customer ser1ice center but also on t&eir own time for breakfast, at lunc&, for coffee, or at t&e end of t&e day;and sometimes at all of t&ose times$ 2&is sociability is not 8ust a retreat from t&e loneliness of an isolating 8ob$ 2&e constant c&atting is similar to t&e background updating t&at goes on all t&e time in any ordinary work site$ 2&ere, too, c&atting usually passes unnoticed unless someone ob8ects to it as a waste of time$ %ut it#s not$ 5rr s&owed t&at t&e reps use one anot&er as t&eir most critical resources$ 'n t&e course of sociali9ing, t&e reps de1elop a collecti1e pool of practical knowledge t&at any one of t&em can draw upon$ 2&at pool transcends any indi1idual member#s knowledge, and it certainly transcends t&e corporation#s documentation$ :ac& rep contributes to t&e pool, drawing from &is or &er own particular strengt&s, w&ic& t&e ot&ers recogni9e and rely on$ "ollecti1ely, t&e local groups constitute a community of practice$ 60or a detailed description, see )"ommunities of Practice= 2&e 5rgani9ational 0rontier,) 4%! *anuary-0ebruary 2000$7

Storytelling. >uc& of t&e knowledge t&at exists wit&in working groups like t&e one formed by our Xerox reps comes from t&eir war stories$ 2&e constant storytelling about problems and solutions, about disasters and triump&s o1er breakfast, lunc&, and coffee ser1es a number of o1erlapping purposes$ +tories are good at presenting t&ings se3uentially 6t&is &appened, t&en t&at7$ +tories also present t&ings causally 6t&is &appened because of t&at7$ 2&us stories are a powerful way to understand w&at &appened 6t&e se3uence of e1ents7 and w&y 6t&e causes and effects of t&ose e1ents7$ +torytelling is particularly useful for t&e reps, for w&om )w&at) and )w&y) are critical but often &ard matters to discern$ .e all tell stories t&is way$ :conomists tell stories in t&eir models$ +cientists tell stories in t&eir experiments$ :xecuti1es tell stories in t&eir business plans 6see )+trategic +tories= 4ow ?> 's !ewriting %usiness Planning,) 4%! >ay-*une 1@@A7$ +torytelling &elps us disco1er somet&ing new about t&e world$ 't allows us to pass t&at disco1ery on to ot&ers$ nd finally, it &elps t&e people w&o s&are t&e story de1elop a common outlook$ 5rr found t&at war stories gi1e t&e reps a s&ared framework for interpretation t&at allows t&em to collaborate e1en t&oug& t&e formal processes assume t&ey are working independently$ Improvi ation. Bot all of t&e reps# problems can be sol1ed o1er breakfast or by storytelling alone$ :xperimentation and impro1isation are essential, too$ 5ne day, 5rr obser1ed a rep working wit& a particularly difficult mac&ine$ 't &ad been installed recently, but it &ad ne1er worked satisfactorily$ :ac& time it failed, it produced a different error message$ 0ollowing t&e establis&ed process for eac& particular message ;replacing or ad8usting parts;didn#t fix t&e o1erall problem$ nd collecti1ely t&e messages made no sense$ 4a1ing reac&ed &is limits, t&e rep summoned a specialist$ 2&e specialist could not understand w&at was going on, eit&er$ +o t&e two spent t&e afternoon cycling t&e mac&ine again and again, waiting for its intermittent cras&es and recording its state w&en it did$ t t&e same time, t&ey cycled stories about similar-looking problems round and round until t&ey, too, cras&ed up against t&is particular mac&ine$ 2&e afternoon resembled a series of alternating impro1isational 8a99 solos, as eac& man took t&e lead, ran wit& it for a little w&ile, t&en &anded it off to t&e ot&er, t&is all against t&e bass-line continuo of t&e rumbling mac&ine$ 'n t&e course of t&is practice, t&e two gradually broug&t t&eir separate ideas closer toget&er toward a s&ared understanding of t&e mac&ine$ :1entually, late in t&e day, e1eryt&ing clicked$ 2&e mac&ine#s erratic be&a1ior, t&e experience of t&e two tec&nicians, and t&e stories t&ey told finally formed a single, co&erent account$ 2&ey made sense of t&e mac&ine and worked out &ow to fix it$ nd t&e solution 3uickly became part of t&e community lore, passed around for ot&ers in t&eir group to use if t&ey encountered t&e same problem$ s 5rr#s study s&ows, executi1es w&o want to identify and foster best practices must pay 1ery close attention to t&e practices as t&ey occur in reality rat&er t&an as t&ey are represented in documentation or process designs$ 5t&erwise, t&ey will miss t&e tacit knowledge produced in impro1isation, s&ared t&roug& storytelling, and embedded in t&e communities t&at form around t&ose acti1ities$ ,oes t&at mean process &as no importance in t&is contextC 5f course not$ %ut t&e processes t&at support &ow people

work s&ould be deeply informed by &ow t&ey already work;not imposed from abo1e by process designers w&o imagine t&ey understand t&e work better t&an t&ey actually do$ rmed wit& a sense of w&at really &appens on t&e ground, it#s possible to design processes t&at prompt impro1isation rat&er t&an ones t&at are blindly prescripti1e$

:xcerpted from t&e article )%alancing ct= 4ow to "apture -nowledge .it&out -illing 't) in t&e !arvard Bu ine "eview, >ay-*une 2000$ D 5rder t&e full article E John Seely Brown is c&ief scientist at Xerox and director of Xerox#s Palo lto !esearc& "enter 6P !"7$ Paul Duguid is a &istorian and social t&eorist affiliated wit& t&e Fni1ersity of "alifornia at %erkeley and Xerox P !"$ 2&is article is adapted from t&eir book The Social 'ife of (nfor ation 64ar1ard %usiness +c&ool Press, >arc& 20007$ ll images G :yewire unless ot&erwise indicated$