t shaped managers

106 HARVARD BUSINESS REVIEW

The knowledge economy demands a new kind of executive, onewho freely shares ideas and expertise across the company whileremaining fiercely committed to business unit performance.But T-shaped managers must be carefully cultivated. Here's how.

Introducing

"pShapedManagers

Knowledge Management'sNext Generation

MARCH 2001

by Morten T. Hansen andBoiko von Oetinger

DESPITE THEIR BEST EFFORTS, most Compa-nies continue to squander what may be their

greatest asset in today's knowledge economy: thewealth of expertise, ideas, and latent insights thatlies scattered across or deeply embedded in theirorganizations. And that's a shame, because capital-izing on those intellectual resources-using existingknowledge to improve performance or combiningstrands of knowledge to create something alto-gether new-can help companies respond to a sur-prising array of challenges, from fending off smaller,nimbler rivals to integrating businesses shoved to-gether in a merger.

Many companies have tried, with mixed success,to leverage this underused asset by centralizingknowledge management functions or by investing

107

Introducing T-Shaped Managers

heavily in knowledge management technology. We sug-gest another approach, one that requires managers tochange their behavior and the way they spend their time.The approach is novel but, when properly implemented,quite powerful.

We call the approach T-shaped management. It relies ona new kind of executive, one who breaks out of the tradi-tional corporate hierarchy to share knowledge freelyacross the organization (the horizontal part of the "T")while remaining fiercely committed to individual busi-ness unit performance (the vertical part). The successfulT-shaped manager must leam to live with, and ultimatelythrive within, the tension created by this dual responsi-bility. Although this tension is most acute for heads ofbusiness units, any T-shaped manager with operating unitobligations must wrestle with it.

You might ask. Why rely so heavily on managers toshare knowledge? Why not just institute a state-of-the-artknowledge management system? The trouble is that,while those systems are good at transferring explicitknowledge - for example, the template needed to performa complicated but routine task-direct personal contact

is typically needed to

The entire history of the effectively transfer im-_ , , ^ plicit knowledge-theT-shaped manager at

BP is one of continually atj gly applied to par-fine-tuning the tension ticular business prob-between the manager's lems or opportunitieshorizontal and vertical and is crucial to theroles, an evolution that ^""^^^ « innovation-

^ ^, . , driven companies. Fur-cor,tinues to this day.

ing documents aroundcan never engender the degree of collaboration that'sneeded to generate new insights. For that, companiesreally have to bring people together to brainstorm.

Effective T-shaped managers will benefit companies ofalmost any size, but they're particularly crucial in largecorporations where operating units have been grantedconsiderable autonomy. Although giving business unitsgreater freedom generally increases accountability, spursinnovation, and promotes sensitivity to local marketconditions, it also can lead to competition betweenunits, which may hoard, rather than share, expertise. Byencouraging collaboration, a T-shaped management sys-tem can be a powerful counterbalance to such negativebehavior.

Morten T. Hansen is on leave as an assistant professor ofbusiness administration at Harvard Business School and Iscurrently a manager in the San Francisco office of the BostonConsulting Group. Boiko von Oetinger is a senior vice pres-ident in the Munich office of the Boston Consulting Croupand is the director of the firm's Strategy Institute.

Our research over the past six years suggests that fewcompanies have recognized T-shaped management as akey to success and even fewer have enjoyed its benefits. Sohow do you successfully cultivate T-shaped managers andcapitalize on the value they can create?

Energy giant BP Amoco-a sprawling enterprise withover 100,000 employees and operations in loo countries -provides some provocative answers. Our in-depth exami-nation of BP's management practices, including inter-views with more than 25 business-unit and corporatemanagers, highlighted five specific types of value thatT-shaped managers can generate. BP's experience alsosuggests guidelines for creating an environment in whichT-shaped managers will fiourish. Such guidelines are im-portant because the benefits of T-shaped managementwon't be realized - will even become liabilities - if the con-cept is poorly implemented. A key insight: senior execu-tives must put in place mechanisms that simultaneouslypromote and discipline managers' knowledge-sharingactivities.

BP's Evolving ApproachThe story of BP's ongoing effort to create an effectiveT-shaped management system is instructive in part be-cause of a number of initial missteps the company madealong the way. Indeed, the entire history of the T-shapedmanager at BP is one of continually fine-tuning the ten-sion between the manager's horizontal and vertical roles,an evolution that continues to this day.

The seeds of BP's T-shaped management approachwere planted in the early 1990s at the oil and gas explo-ration division of what was then British Petroleum. In asuccessful bid to cut out layers of management and im-prove performance and financial accountability, the divi-sion-known as BPX and headed at the time by BP's cur-rent chief executive, John Browne-was divided intonearly 50 semiautonomous business units. But becausebusiness unit leaders were personally accountable fortheir units' performance, they focused primarily on thesuccess of their own businesses rather than on the successof BPX as a whole.

With the ultimate aim of making BPX more valuablethan the sum of its business units' results, Browne and hisexecutives set out to encourage greater understanding ofthe goals and challenges of other units and of BPX as awhole. Early in 1992, they established "peer groups," inwhich leaders of roughly a dozen business units engagedin similar types of businesses met to discuss the strategicand technical challenges they all faced. The importantthing about these meetings was that senior managementwasn't allowed in the room. That reduced posturing andencouraged candor.

But over time, BPX's senior management realized thatsimply sharing knowledge for knowledge's sake-the cre-



ation of "learning loops," as one executive put i t -was only marginally productive. So in 1994, thepeer groups became more results oriented, as-suming responsibility for allocating capital re-st>urces among business units in the groupand for setting unit performance targets.In both cases, the aim was to further thebroader interests of the entire peer groupand help it meet its own goals, which wereset by the BPX executive committee.

Not surprisingly, this approach initiallycaused considerable friction among thebusiness unit leaders. "People were verycombative about why their pet project mer-ited investment," recalls John Leggate, who ranseveral BPX business units during this period."Butover time, and with some due process instituted withinthe groups, business unit leaders gradually became lesspartisan and thought about the bigger result." In fact, theunit heads began to see the benefits of this collaborativeapproach, including the opportunity to tap the knowl-edge and expertise of other units.

In 1995, when Browne became BP's CEO, he rolled outthis system of collaboration and networking across the en-tire company. Since then, BP has become known for itsknowledge-sharing practices.' Less known is BP's realiza-tion that, despite the clear benefits of such a system, youcan overdo it.

For instance, BP had encouraged the formation notonly of peer groups but also of cross-unit networks fo-cused on areas of shared interest. Over time, this ideafiowered into an unforeseen number of networks andsubnetworks (the "helicopter utilization network" wasone), which consumed increasing amounts of managers'time. An audit within BPX alone identified several hun-dred of these networks. "People always had a good reasonfor meeting," says Leggate, who is now BP's group vicepresident for digital business. "You're sharing best prac-tices. You're having good conversations with like-mindedpeople. But increasingly, we found that people were fiyingaround the world and simply sharing ideas without al-ways baving a strong focus on the bottom line." So thecompany again tightened the reins, reducing the numberof networks and limiting cross-unit meetings to those con-cerned with specific business results.

Tecbnology has played a role in these knowledge-shar-ing activities. BP has an electronic "yellow pages" thatidentifies experts in different areas. And the companyearly on developed sophisticated digital-networkingcapabilities, such as multimedia e-mail and desktop video-conferencing, which enable managers to gather and workacross business units in virtual teams. But technology hasits limits. The expert directory quickly falls out of date andoften fails to fully capture exactly what each personknows. And executives say that well-developed relation-

ships, nurtured through face-to-face interaction, arefundamental to successful virtual teams.

As shortcomings have emerged, BP hasmodified its knowledge-sharing system. Onesign of the system's strength has been itsability to function even as BP undertookthe mammoth task of integrating the dis-parate cultures of two acquired rivals:Amoco in 1998 and ARCO last year. "Thereal value of mergers lies in the scope theyoffer for learning from a wider base of

BP has become known for itsknowledge-sharing practices.

Less known is BP's realization that,despite the clear benefits of such a

system, you can overdo it

experience," says Browne. "We've begun to do that and,as we live through the process, we see more and morepotential. Still, the process of leveraging the learning isitself a learning experience." The leap in scale and globalpresence that resulted from the mergers, for instance,has prompted BP to reorganize business units into newgroups more strategically focused than the former peergroups. The evolution will continue, executives say, asstrengths and weaknesses of the new system appear.

The T-Shaped Manager in ActionTo get an idea of how BP's system works in practice, let'stake a detailed look at an individual T-shaped manager.'David Nagel, BP's gas business unit head in Egypt, joinedthe company as a result of the Amoco merger and quicklyfound a key difference between his former and currentemployers. "Before, if you needed help in a particulararea, you'd go to Houston for assistance,"he says, referringto the former headquarters of Amoco's oil and gas busi-ness. Today, Nagel typically seeks help from his peers inother business units and often reciprocates, as well.

Like all BP business unit managers, Nagel has a two-part job description. He is effectively CEO of his businessunit, with profit and-loss, balance sheet, capital expendi-ture, and other responsibilities. These are spelled out in apersonal annual performance contract be has with bisboss, one of BP's group vice presidents. At tbe same time,Nagel is expected to engage in a variety of cross-unitknowledge-sharing activities, which he estimates con-sume somewhere between 15% and 2(>% of his time.

To ensure that his horizontal activities don't under-mine the goal of outstanding unit performance, Nagelmust carefully manage his time and energy. (See tbeexhibit "A T-Shaped Workweek.") That means continualself-monitoring to be sure that cross-unit activities in fact

MARCH 2001 109


Companies on the Road to T-Shaped Management

Our close examination of BP grew outof a study we conducted of a group oflarge, global companies that seemedlikely to have effective practices forsharing knowledge across units. Weconducted interviews with execu-tives at 30 companies, about one-third of them located in Asia,one-thirdin Europe, and one-third in the UnitedStates. The companies representeda wide array of industries, includingcomputers, biotechnology, paper, steel,Pharmaceuticals, consumer goods,banking, and high technology

Although we had, through mediasearches and discussions with ourcolleagues, identified in advancethose companies that empha-sized cross-unit links, wefound that few actuallybenefited broadly fromcross-unit learning andcollaboration. Many hadtried to implement link-ing mechanisms; theyjustdidn't work very well.For example, several exec-utives described attendingfrequent off-sites, informa-tional meetings, and confer-ences convened in the beliefthat such gatherings would resultin greater cooperation and knowledgesharing across units.The problem wasthat these, on their own, weren't effec-tive. I n the words of one manager, "wemeet to agree to disagree and sched-ule another meeting."

In fact, only BP regularly realizedthe full array of value-creation possi-bilities we identified in our research.But a handful of companies have hadsome success collaborating acrossbusiness and country boundaries.• GlaxoSmithKline. The pharmaceuti-cal company has benefited by encour-aging the cross-pollination of ideasthrough information matchmakers,which we call "human portals." Oneday, for example, V. Thyagirajan, anarea director located in Singapore, re-

ceived a phone call from Claxo's man-aging director in the Philippines, whowas looking for new products formu-lated for his local market. Thyagirajanset up a meeting with the Glaxo man-aging director in Mumbai, India,some-one Thyagirajan knew was also inter-ested in local product development Ina visit to an R&D lab outside Mumbai,the Philippines executive saw that theIndian product developers were work-ing on line extensions of existing anti-tuberculosis medication, a family ofdrugs not emphasized at the corporatelevel ofthe UK-based company. The lab

visit sparked a joint effort between theteams in India and the Philippines, anda Philippine researcher moved to theMumbai facility to help develop newproducts. The team first came up witha modified antJ-TB medication formu-lated specifically for the Philippinesand has subsequently developed otherproducts for the Philippine market.- Siemens. Three years ago, the largeGerman industrial company launcheda training program that brings high-potential managers from differentdivisions together in small teams tosolve a problem facing one ofthe busi-ness units. So far, more than 100 teamshave been formed. Team memberswork together for about a year, whichincludes attending several weeklong

meetings at an off-site corporate cam-pus. They then make recommendationsto the business unit manager involved,who serves as the team coach duringthe project. Through the program,team members develop their businessskills, build informal relationshipsacross business units, and save thecompany money - more than $10 mil-lion so far- by solving real businessproblems.

• Ispat International. Senior execu-tives at this London-based global steelmaker have institutionalized severalsimple mechanisms for sharing knowl-

edge across their far-flung units thatcould easily be implemented in

companies from many otherindustries.One is Ispat's pol-icy of cross-^directorships,which requires the generalmanager of every operat-ing unit to sit on the boardof at least one other unit.The managing directorsof Germany and Trinidad,

for instance, sit on eachother's boards because they

both produce "long" steelproducts, such as rods and

other structural materials. Tbispeer oversight encourages units to

adopt best practices from other units-for instance, Germany's successfuldownsizing initiative. Managing direc-tors of each operating unit also jointogether every week for a phone meet-ing that lasts no longer than twohours. Managers report exceptions,nonroutine activities, and things that,in company parlance, "keep themawake at night." In one recent call, themanaging director in Trinidad men-tioned problems he was having with atransformer that repeatedly failed. Asit turned out, managers in Mexico andCanada had similar transformers andwere having similar problems. Thethree units ended up cooperating onboth troubleshooting and buying theexpertise to perform repairs.

no HARVARD BUSINESS REVIEW


serve an important business purpose."We've tried to elim-inate the peer group meetings that are held just for thepurpose of saying, 'We had a peer group meeting,'" Nagelsays. "We recently canceled one that made sense when wescheduled it but that doesn't make sense anymore be-cause tbe issues have changed."

The dual demands also have required him to delegatesome business unit responsibilities, particularly gas ex-ploration and production, to two trusted lieutenants. Thatfrees him for tasks extending beyond his business unit.

The ones that involve knowledge sharing can be char-acterized quite simply: in a broad corporate context,Nagel collaborates, connects, gives, and takes.

He collaborates in a peer group comprising his businessunit and seven others throughout the Mediterranean andAtlantic regions that, like his, are focused on increasinggas production. In fact, he is the facilitator of the group,responsible for convening the meetings and workingto move its members toward agreement on the oftenthorny and challenging issues of exactly how to allocatecapital and meet those peer-group production targets setby his division's executive committee. In September, thegroup determined, based on each unit's projections, thatit was likely to fall about 3% short of its goal for 2001.Through some intense discussion over the course of thenext two meetings, the group determined whicb businessunits were in the best position to close that gap.

Nagel also occasionally connects people from differentparts of the company. For example, he may get a call froma legacy BP engineer seeking the name of a legacy Amocoengineer who could offer advice in an area in whichAmoco was known to have particular strengths.

Nagel gives advice to other business units, when re-quested by individuals both within his peer group andbeyond it. Last year, he and his managers were involved inroughly 20 such "peer assists" to other BP units; Nagelwas personally involved in three of them.

And Nagel takes advice from other units. Last year, bisbusiness unit benefited from roughly ten peer assists, inwhich people came from around the world to offer spe-cific ideas on such issues as his unit's marketing plan.Sometimes the help comes more informally. Shortly afterthe merger, an engineer in Nagel's unit tapped into hisnetwork of BP contacts and determined within severaldays that the productivity of a particular type of wellbeing drilled in Egypt could be tested without "flaring" it-that is, without opening it up and burning off some gas.This allowed speedy evaluation of the well while avoidingenvironmental concerns about flaring that had arisenunexpectedly.

Why didn't Nagel seek help from headquarters, as hemight have done at Amoco? "The model here is an openmarket of ideas,"he says."People develop a sense of wherethe real expertise lies. Rather than having to deal with thebureaucracy of going through the center, you can just cut

across to somebody in Stavanger INorwayl or Aberdeen[Scotland] or Houston and say, 'I need some help. Canyou give me a couple of hours?' And that is expected andencouraged."

Creating Horizontal ValueT-shaped managers like David Nagel create vertical valuefor BP in the form of strong business unit results-a topmanagerial priority at a company that emphasizes busi-ness unit accountability. They create horizontal value inone of five distinct ways, which progress up the value-added ladder from the exploitation of existing resourcesand knowledge to the exploration of new opportunities.The benefits at the lower end stem from traditionalknowledge transfer; those at the upper end require col-laboration to create new ideas.

Increasing Efficiency Through the Transfer of BestPractices. Deborah Copeland, head of BP's business unitfor retail operations in the southeastern United States,was looking for ways to improve the performance of herregion's BP and Amoco service stations. Through her peergroup, she learned of pilot programs at BP stations in theUnited Kingdom and the Netherlands that were testingsome innovative ways to order and deliver conveniencestore supplies. So last summer, she sought a peer assistfrom ber counterparts in those two countries, as wellas from BP retail executives in seven other countries. Theymet and recommended best practices in such areas as sup-plier management and store layouts. Copeland thenlaunched three pilot programs at several stores in the At-lanta area. The results, she says, were dramatic. The pilotstores stocked 26% fewer stock-keeping units (or SKUs)than similar control sites; this inventory reduction led toa 20% decrease in working capital even while sales rose10%. Copeland is currently rolling out the practices acrossanother 62 sites in Atlanta and Orlando, Florida.

Improving the Quality of Decisions Through PeerAdvice. When Anne Drinkwater became head of the busi-ness unit responsible for transporting Alaska oil from theNorth Slope to refineries in the western United States, viapipeline and tankers, she didn't know much about ship-ping. Her previous job had been in the Gulf of Mexicooverseeing deepwater oil wells. But a key task afrer shearrived in Alaska was to decide the number and sizes oftankers her unit needed to handle oil production outputover the next 20 years. In addition to enlisting a few ex-pierts from the corporate center in England, her projectteam identified six people from other business units whocould help with the decision. In a number of face-to-faceand virtual meetings last summer, the group coveredissues ranging from long-term oil production forecasts tofinancing options for the new ships. The final recommen-dation: buy three new tankers and take options onanother three."In a very supportive way, they challenged

MARCH 2001 111


some of our thinking and pointed us in the right direc-tion," says Drinkwater, who recently left Alaska to heada BP unit in Norway.

Growing Revenue Through Shared Expertise. In thelate 1990s, Graham Hunt was the leader of a BP petro-chemical business unit responsible for the design and con-struction of a $200 million acetic acid plant in westernChina, to be run as a joint venture with Sinopec, the Chi-nese petrochemical company. The complexity of bringingsuch a plant on-line In so remote a location made it a rel-atively risky undertaking, so Hunt sought BP expertisefrom a number of operating units around the world. Overa 30-month period, about 75 people flew to the site inChina from different parts of BP for visits lasting from aday to several weeks. They gave advice on technical, legal,tax, safety, accounting, and financial issues. Largely be-cause of this peer assistance. Hunt says, the two-year con-struction project came in on time and under budget. Pro-duction began in November 1998, and the business brokeeven after only several months of operation. Says Hunt,now chairman of BP China: "We needed expertise and,given our organization, it was often a phone call away."

Developing New Business Opportunities Throughthe Cross-Pollination of Ideas. Vast but often far-flungpools of ideas and expertise are one of the greatestcompetitive advantages a large company has in today'sknowledge economy. Fruitfully combining this knowledgecan produce what might be called "epiphanies of scale" -

creative insights that

Early on, people on the ^ hot start-up com-BP side made it quite P^"^ ^at has fewer

, 1 ^ • I ^ mtellectual resourcesc ear that you might _ ^ u u^ ^

^ ^ may not be able tohave spectacular achieve, m the spring

individual business- of 1999, John Meio,unit performance, but then BP'S director of

if you weren't seen to be brand development,making contributions *^^'P^^ """^''^^ ^" '"'•

, J . tiative to develop newbeyond your own unit, ^-businesses using ex-

yOUWOuldntbe jsting BP assets, inviewed favorably, typical BP fashion,

managers from some15 business units met to brainstorm. The effort producednearly 600 ideas, out of which 150 are currently being de-veloped. For example, BP is working on Flightneeds.com,which will help operators of small to midsized jets plantheir trips and fueling needs, then furnish fuel at theplanned stops. Melo says that the managers from differ-ent units "could look at our existing business through a va-riety of lenses and thus identify hidden opportunities."

Making Bold Strategic Moves Through the Prom-ise of WelKoordinated Implementation. David Eyton,currently head of a gas business unit in Trinidad, wasn'talone in wondering how long a massive integration of

companies the size of BP and Amoco would take. "Itseemed possible that the two companies and culturescould continue to exist indefinitely as parallel universes,"he says. But Eyton's peer group at the time-comprising12 business units, roughly half of them legacy Amocounits-Immediately embraced the integration process.Working witbin the well-established peer group norm ofa freewheeling and candid exchange of ideas, the groupresolved the staffing issues in two months. One monthlater, other basic elements of the combined business wereput in place. BP executives say it was such flexible peergroup behavior that permitted a nearly complete inte-gration of the companies within just 90 days. And that ex-perience helped give BP the confidence to launch anothermajor merger, this time with ARCO, only months later.

Designing the Right Organization:Promote and DisciplineSenior executives who are eager to create new valuethrough a T-shaped management environment must, likethe T-shaped managers themselves, find ways to managesome inherent tension in the concept. BP has done thatthrough the combination of promoting and disciplininghorizontal management behavior. Top executives canpromote this bebavior in several ways without creating anossifying layer of bureaucracy.

Create clear incentives. Business unit managers at BPare judged on their ability to meet specific performancetargets for their units. But they also are rewarded andpromoted according to how effectively they-and theirstaffs-share knowledge with others outside their units.Such behavior "is a key test of a manager's performanceand potential," says Nick Butler, the main policy adviserto CEO Browne. "Lone stars"-those who deliver out-standing business unit performance but engage in littlecross-unit collaboration-can survive within BP, but theircareers typically plateau.

Knowledge-sharing contributions are by their naturemore difficult to measure than success at meeting specificunit performance targets. But executives throughout thecompany say that bosses are generally well aware ofthe level of their subordinates' cross-unit contributions.And the policy of promoting T-shaped behavior is rein-forced by BP's corporate culture. David Nagel recalls hisexposure to this shortly after the Amoco merger: "Earlyon, people on the BP side made it quite clear [to legacyAmoco managers] that you might have spectacular indi-vidual business unit performance, but if you weren't seento be making contributions beyond your own unit, youwouldn't be viewed favorably."

Develop economic transparency. Good T-shapedmanagers don't just provide assistance across businessunit boundaries; they also seek help themselves. And oneway to encourage such requests is through corporatewide


Introducing T-Stiaped Managers

A T-Shaped Workweek Atypical workweek for David Nagel, BP's gas business unit head in Egypt,shows how he balances his vertical (business unit) and horizontal (knowl-edge-sharing) responsibilities. (The workweek in Cairo runs from Sundaythrough Thursday.)

SUNDAY, OCTOBER 22

AM* Egypt gas business-unit team meeting^prqgress

versus ttie performance contract

PM• Meet with UK representative from Barclays Bank• Meet with UK government trade mission to Egypt• Team meeting with Egypt oil business unit

leader on safety, commercial deals, and governmentand public relations

• Note to fellow board members of local communityservices association (which provides services to expatriateBP employees) on simplified accounting practices

MONDAY. OCTOBER X-

AM• Meet with Egyptian gas utility (partner on new project)

PM•Chair Egyptian Petroleum Industry Environmental

Protection Committee meeting• Meet with BP oil traders frorri London regarding

opportunities in Egypt

TUESDAY, OCTOBER 24

AM• Meet with BP project management leader in Cairo to

discuss new approach to project management in Egyprt• Check roles and responsibilities on BP emergency

plans for Egypt• Conference call with BP downstream organization

(refined products and retailing) about opportunitiesin Egypt

PM• Career discussion with gas unit staff members• LNC (liquefied natural gas) project review• Review financial projections for peer groupSeek peerjnputyia e-mail on next steps for key project

mr WEDNESDAY. OCTOBER 2

AM• Fly to London; review business unit correspondence

PM• Peer group teleconference on financial submissions,

upcoming meeting agenda• Meet BP Algeria oil business unit leader (peer group

member) to discuss future production opportunities

THURSDAY. OCTOBER 26

AM• Peer assist to Algeria gas business unit• E-mails on business unit promotions, individual

development plans

PM• Discussion with Spanish gas and power business unit

leader on LNC opportunities• Fly to Cairo; review speech for upcoming BP CasTech

Conference in Houston

FRIDAY, OCTOBER 27

AM/PM• Finalize peer group^perfbrmance submissions for 2001

KEY• Business unit activities• Cross-unit activities

internal-benchmarking systems, which spur managers ofunderperforming units to ask for help.

When in 1999 Jeanne Johns took over as business unitleader for BP's $1.5 billion oil refinery in Toledo, Ohio, shechecked a company database that lists numerous perfor-mance and cost metrics for BP's 19 refineries around theworld. Some items, such as salaries, clearly were notcomparable because of differing business environments.Still, she found that the Toledo facility was lagging far be-

hind its counterparts on a key performance metric-something called "cents per equivalent distillation capac-ity," which compares the cost of running refineries of dif-ferent sizes and complexity. The best-performing facilityboasted an index of 9 cents per EDC; the Toledo refinerywas at 14 cents.

At Johns's request, a team that included some key peo-ple fi-om the top-performing refinery, located in Texas,spent a week in Toledo helping her staff identify possible

MARCH 2001 113


From: Larry WatsonTo: Les OwenSubject: Lightning Protection

We haven't talked in awhile. Do you know anyonewell versed in lightning protection equipmentand practices? We've had a lot of problems recentlywith lightning strikes damaging our pumpingfacilities.

Human Portals at Work

From: Les OwenTo: Ian FrenchSubject: RE Lightning Protection

How are things in Houston? I thought you might beable to help with the attached request for information,given your past experience in Larry's part of the worldand with lightning-related problems.

From: Les OwenTo: Larry WatsonSubject: RE Lightning Protection

Received the following reply on lightning. Isuggest you contact Nigel and Ian directly, whoare both in Houston, if you have questions.

BP cultivates what we call "human portals," a particular type ofT-shaped manager who helps people identify third parties in the orga-nization that can provide needed information. Les Owen, a BP engi-neer in Alaska, is such a manager; likethoseof his counterparts acrossthe company, his role is informal and in addition to his regular busi-ness unit responsibilities. Recently, Owen put a BP engineer seekinginformation about protecting pipeline facilities from lightning strikesin touch with two engineers elsewhere in the company who were ableto help. The exchange-reproduced here in an edited version in whichnames, locations, and commercially sensitive facts have beenchanged-is noteworthy not because of its extraordinary results butrather because it's typical ofthe way Owen and many people like himthroughout BP regularly serve as information matchmakers.

From: Ian FrenchTo: Nigel Wallacecc: Les OwenSubject: RE Lightning Protection

Les Owen, who works for BP Pipelines in Alaska,has been asked by Larry Watson, an engineer atthe pipeline unit in Siberia, about recommendedpractices for lightning protection. Before I followup, could you respond with any thoughts? They'relikely to be particularly interested in equipmentand the associated electronics needed forprotecting pipeline pumping facilities.

From: Nigel WallaceTo: Ian Frenchcc: Les OwenSubject: RE Lightning Protection

There are many claims for exotic lightningprotection systems. I prefer to approach the issuewith good earthing and earth-bonding practice andthe protection of particularly vulnerable componentswith surge diverter devices. I attach a copy of someuseful presentation slides from the BP ElectricNetwork forum in Windsor in November. Call ifquestions. I'm always happy to provide help in thearea of lightning protection - often seen as a blackart, though it shouldn't be!

operating improvements. As a result, within nine months,the Toledo refinery reduced its costs to n cents per EDC,saving $24 million a year.

Formalize cross-unit interactions. Institutionalizingcross-unit behavior can help sustain something thatmight otherwise remain ad hoc. The difficulty is to avoidcreating the very bureaucracy that working across units ismeant to cut through. With its peer groups, BP has triedto achieve this balance. One key, managers say, is the pro-hibition against managers'superiors participating in peergroup meetings, which leads to the mixture of robust con-frontation and collegia! support that characterizes suchgatherings.

David Eyton was the business unit leader ofthe V\^chFarm oil field in southern England at a time, in the late1990s, when oil production was beginning to decline inthis mature field. Eyton presented to his peer group a

1999 operating budget, excluding onetime items, of $30million-the same amount as the previous year, eventhough production was falling. But the n other membersofthe group, in a challenge to the long-standing assump-tion that unit costs rise as oil production falls in such ma-ture fields, declined to support the proposal. They pushedEyton to reduce ongoing costs by 20% to keep unit costsflat'i was shocked,"he says."I had absolutely no idea howto do what they were asking me to do."

Over the next several days of meetings, two membersofthe group met with him privately and offered not onlysupport but also tips on how to pull this off. After the peergroup meeting, Eyton passed the challenge on to his ownteam. They formed a task force, which visited other busi-ness units, including the ones whose leaders had offeredthe initial challenge. Through such measures as cancelingleases on buildings and renegotiating service contracts.



Eyton's unit cut its operating costs to $24 million in 1999.More important, his team devised a long-term plan to en-sure that operating costs declined in parallel with pro-duction for the remainder of the field's life.

Eyton's peer group experience may serve as a convinc-ing example of the benefits of cross-unit behavior. But BPhas learned that encouraging such behavior is only halfthe equation. Senior executives must also monitor andoccasionally rein it in. They can discourage ineffectivenetworking in at least two ways:

Curb cross-unit interactions. T-shaped managersneed to know that it is not only acceptable but also some-times wise to refuse requests from other units for theirtime or the time of their staff. "1 could easily wake upevery day and find virtually my entire leadership team offat a meeting somewhere around the globe," says DeborahCopeland, the retail unit executive. "While each one ofthose meetings would benefit the company in some way,there would be hardly anybody left in the office to keepmy own business running. In addition, you risk burningyour people out with atl of this collaboration."

BP executives even talk of occasional outbreaks of"peer assist mania." This is wben managers request toomany assists, either because they bave become swept upwith excitement over the potential benefits or, more cyn-ically, because tbey tbink peer assists are a key to earninglegitimacy within the company. In either case, tbe exces-sive requests waste precious time and effort and can un-dermine faith in the process. Anne Drinkwater, the for-mer business unit leader in Alaska, recalls receiving aninvitation to participate in a peer assist gathering that itwas clear she could not contribute mucb to. In fact, shefelt tbe request amounted to no more than an attempt bytbe requesting manager to check off a box on his annuallist of professional objectives. She declined. "You onlyhave so much time," she says, "and you have to prioritizearound the right things."

Managers also emphasize that knowledge-sharing ac-tivities should focus on business results rather tban socialevents. While social gatherings, like company confer-ences and off-sites, do help create some comfort level,they also leave managers frustrated because tbe meetingsdon't really achieve any business goals. Solving problemstogether through a results-oriented approach to knowl-edge sharing is a more potent way to create trust amongpeople from different business units because achievingresults together creates a track record showing that peo-ple are really helping one another. Trust is a byproduct ofeffective collaboration.

Replace bloated Rolodexes with "human portals."Another potential danger for companies seeking the ben-efits of cross-unit collaboration is ovemetworking, withevery manager fiipping through a bulging Rolodex when-ever an issue arises. This is a particularly inefficient wayto share expertise: if loo people want to keep in touch

with one another directly, they would need to maintain4,950 direct relationships-a challenge for even the mostambitious group of T-shaped managers.^

Senior executives can minimize this problem by iden-tifying and cultivating a particular type of T-shaped man-ager, one who connects people seeking Information withthose who can help them-effectively serving as humanportals in the companywide knowledge web. Given theimplicit nature of the advice that's typically called for,human portals can't simply reroute information like aswitchboard. Rather, they must use their extensive knowl-edge about who knows what and their understanding ofwhat actually is needed to creatively make connectionsbetween information seekers and information holders.

Although business unit heads may be outstandinghuman portals, often these people connectors can befound further down in the organization. Take Les Owen,an engineer responsible for pipeline technical and regu-latory matters in the Alaska business unit that was headedby Anne Drinkwater. "Les is better than a Web site," shesays. "He's always help-ing other people con- T-shaped managersnect. He knows every- need to know that it isone and everything not only acceptable butthat s gomg on. Like , ^. . .others at BP who serve ^'^^ sometimes Wise tothis function, Owen refuse requests for theircarries out his human time or the time ofportal responsibilities their staff.as a sidelight to his pri-mary iob. Drinkwater had informally ensured that he hadtime for this role. "I was careful when we agreed on workallocation that I didn't fill \(xm of his time," she says.

On average, Owen says, he fields maybe ten phone callsand as many e-mails a week from people outside his busi-ness unit trying to locate someone who can help with aproblem. Owen, who built up his network of contacts over26 years at BP in a variety of jobs and locations, says hisrole goes beyond the company's electronic directory ofexperts. "Using a database is like picking a name out of aphone book - you don't know the person," he says. "To getan answer they'll have confidence in, people would rathergo to contacts they know." (See the exhibit "Human Por-tals at Work.")

A Behavioral OverlayBP's T-shaped management system isn't the only way acompany can share its intellectual resources across busi-ness units. One traditional alternative is to centralizeknowledge management and decision making, which canbe done in a number of ways. The top management teamcan amass a large number of experts at corporate head-quarters. The team can order business unit heads to col-laborate. Or it can even combine two or more business

MARCH 2001 115


units to coordinate their operations and reap the benefitsthat come from sharing intellectual resources.

While centralizing creates certain economies of bothscale and scope, it has a number of serious drawbacks.Centralized knowledge management often breeds aknowledge bureaucracy that is slow to respond to re-quests for information from managers in the field. It alsomay fail to capture the latest expertise and ideas from theoperating units, where on-the-job learning is constantlyevolving. And centralized decision making about knowl-edge management undermines the benefits of an other-wise decentralized organization, hindering the innova-tion that can well up when autonomous business unitsare free to experiment.

By contrast, what BP's T-shaped management ap-proach does, at its core, is to overlay a flexible "behavioralnet" onto a decentralized organization structure. Mecha-nisms such as peer groups, peer assists, dual promotioncriteria, and human portals are designed to change man-agers' daily activities rather than the organizational struc-ture in which they work. And herein lies the key insightfrom BP's experience. With its behavioral overlay on adecentralized structure, the company can realize thebenefits of cross-unit learning and collaboration withouthaving to institute top-down approaches that could

undermine the freedom andaccountability needed toproduce outstanding indi-vidual unit performance.

Of course, BP does cen-tralize many decisions andactivities, including the de-sign and oversight of infor-mation systems and muchof its procurement. It evencentralizes some specializedtechnical expertise that istoo nartow to be carried byany single business unit.But cross-unit learning ismainly a decentralized anddispersed activity that isbased on managers' willing-ness to work across organi-zational boundaries.

BP's approach refiects asubtle but vita! shifr in thesources of advantage forlarge global companies. Inthe past, the key advantagelarge companies had wastheir ability to pool volumeacross business units andcountries to lower purchas-ing, component, and pro-

duction costs and better leverage their brand. Theseeconomies are still important, but the benefits of cross-unit learning and collaboration have become much moreimportant in many increasingly knowledge-intensiveindustries. And while pooling purchasing volumes toget better supplier rates is often best handled by a centralpurchasing department, the benefits of cross-unit learn-ing tend to be best achieved through decentralized andhorizontal networking-that is, directly among employ-ees in operating units who are able to learn, teach, andcollaborate.

1. See "Whaf s Your Strategy for Managing Knowledge?" by Morten T. Hansen,Nitin Nohria, and Thomas Tiemey, Harvard Business Review (March-Aprii1999).2. See "Unleashing the Power of Learning: An Interview with British Petro-leum's John Browne," by Steven E. Prokesch, Harvard Business Review (Sep-tember-October 1997).

3. We use the term "T-shaped" to refer to a manager's behavior, in contrast toprior uses of the term, where it has typically referred to a manager's skill base.See, for example, the discussion of "T-shaped consultants" in "McKinsey &Company: Managing Knowledge and Learning," by Christopher A. Bartlett,Harvard Business School Case 9-396-357 (1996).

4. "1 he number of connections needed for everybody in an organization toknow everybody else is represented by the equation n(n-\)l2, where n is thenumber of people in the organization.

Reprint R0103GTo order reprints, see the last page of Executive Summaries.

"Ifyou saw the size of the cubicle they're putting me in you wouldn't ask why I'm here."


t shaped managers

Documents