Systems for Managing Knowledge Management

August 31, 1999

After several years as Sales and IT Director at LegalInformation Resources, since 1994 part of Sweet & Maxwell, Derek Sturdycontinues to develop legal database and editorial systems, within the Sweet& Maxwell family, as Product Development Director for Westlaw UK.

Did we win the battle, or the war?

If any of us were asked to summarise the state of play in KM today, we mightcome up with something like this:

‘The technical aspects of managing know-how are no longer arcane rocket science. The technology is available and is not expensive. People have built what used to be called ‘know-how systems’, usually relabelled them ‘knowledge management systems’, installed them and used them. Law firms have grasped the nettle of culture change and realised that effective exploitation of good work done previously can make money, keep and win clients, and open up new opportunities for work. Individual lawyers increasingly enjoy reducing handle-turning and welcome the increases in standards brought by adopting best practices. In short, KM has become just another tool for lawyers, like using word processing instead of typewriters – which looked like a revolution until suddenly it became completely standard practice. KM is a done deal.’

Really? Here I will discuss the essentially backward-looking aspects of KM,and show what has to be done to deliver long-term value from know-how systems.Behind the hype of miracle solutions, there lies an unpleasant truth about thehard work that has to be done, by humans, to make KM effective.

What’s the problem?

The problem lies where it always has done: in updating and weeding thecontent of the system. Leaving aside issues of culture, Luddism, andretentiveness, the questions we have to answer about know-how remain:

  • Is it still true?
  • Is it still (was it ever) relevant?
  • Will the system scale up to the whole firm?
  • What about the sheer volume of text that the passing years will bring?

Now that firms have the opportunity to swamp fee earners with text, how doyou get the quality into the quantity? The solutions are still largely to dowith human brains. We can help them with technical goodies, but we can’t do itall automatically.

A new know-how (KM) system looks so good in the early days. It contains onlythe latest stuff, so it’s up to date; and the amount of it is quite modest –at most a few tens of thousands of pages, and only a few thousand pages in manycases. This is not going to stay true. In a matter of months, stuff which is nolonger best practice, or even good practice, will need to be dealt with, and itwill be clogging up your know-how (KM) system.

Forward looking or backward looking?

In one of the Financial Times supplements on knowledge management, theterm ‘managing collective intelligence’ (MCI) was used to describe processesthat are forward looking, and ‘knowledge management’ was described asbackward looking. While the allocation of jargon in this way is rather anuisance, the distinction is an important one. If KM is about not reinventingthe wheel, MCI is about improving the wheel each time it is given a turn. Sincewe manifestly need to retain exact copies of pieces of work done for clients,clearly know-how systems must contain additional documents that increasinglywill become templates rather than examples, and that are specifically writtenand updated for the job. This negates the often-advertised idea, therefore, thatknow-how systems can be run simply by a few fixes (and some fancy relabelling)of existing document management systems using documents produced by existingprocesses. Instead, we need to create new documents, albeit based on old ones,and new processes to manage them, specifically for the know-how system. It alsoraises the question: so who will do this?

The knowledge management/fee-earning time conflict

That question (who will do this) leads us straight into an awkward and narrowplace. On one side, there is the Scylla of making the best use of your bestbrains for earning fees and delivering the solutions to clients. On the other,we have the Charybdis of the highly inconvenient fact that, if your best brainsdon’t update the know-how system, equally they obviously might fight shy ofusing it. You can’t expect the world’s expert on restructuring sovereigndebt to accept the templates, work-flow or even phraseology of someone theyconsider second-rate. How do we avoid the know-how concept being crushed betweenthese clashing rocks? The solution to updating and improving lies in managingthe things that change and they are:

  • how to do it: the workflow, the business process
  • is it still true: the primary law (legislation and cases) and related codes of practice that provide the legal envelope for the work.

Some answers: start with process

The general principles change relatively slowly and must remain the provinceof your key fee earners. Nobody else can do this for them convincingly, so youneed to build in, to the display or availability of each know-how document, anassociated ‘re-use’ facility. This will provide a constant check on thequestion: ‘is this still the way to do this job?’ You are really quitelikely to get an answer to this question, especially, and fortunately, if youhave annoyed the fee earner by delivering to her an out-of-date or sillydocument, provided you make it quick and simple. Your know-how system must becapable of eliciting a very quick, knee-jerk response on this matter. Here’s atypical tick-the-box list:

  • Downright dangerous
  • Useless – discard
  • Needs updating for this process – not too useful until this has been done
  • Question mark about this – talk to me
  • Still perfectly useful
  • Good but my work will improve this for reuse in this [specify]… type of work

Ticking a box on a nicely-designed, slightly sycophantic screen, is not goingto take a second of the time of someone who has the answer in her head; afterall, she must already have decided what to do with the piece of know-how offeredto her.

Process management

Although this may sound a tautology, someone has to collate the results ofthe process review described above. This may seem the traditional role of theknowledge officer or PSL within a team. But the hard bit is always going to begetting the person who signals the need for, or the availability of, animprovement to the process – the lawyer working on a matter – to completethe detail. Completing the detail is what makes the improved process availableto future users (ie updates the know-how for the process). So such anintermediary must know enough to approach the right people, and needs apersonality which will get this task accomplished, week after week. A Rottweileris the last person required, but a Barbie won’t get there either.

On the positive side, process does in fact normally change slowly; so this isnot likely to be a huge percentage of the total time spent, by any PSL orknowledge officer, on know-how matters. The vital point is to ensure thatsomeone has clear responsibility for doing the job. Otherwise you run the riskthat the process, the workflow, for more and more of the matters handled by thatdepartment will sooner or later become obsolete, and with that obsolescence willcome the decreasing use and final abandonment of your expensive KM system bythat department.

Vetting is clearly an integral part of process management. If a fee earnerhas taken a template based on work done for a specific matter, and hasskillfully adapted it for use on a slightly different matter, both templatescould be valuable. But if you just present a busy Head of Department with twotemplates and ask for approval or otherwise, it will sit in the in-trayconcerned for ever. Instead, proposals, based on summary documentation, andpointing out the differences between templates and reasons for them, which busy,top fee earners can vet in a very short time, will need to be prepared – allof which takes work by knowledge officers.

Summary of forward-looking process management:

  • Top fee-earners create content and workflow
  • For each new matter, existing templates are reviewed and adapted to specific matters by whoever is charged with the work
  • Review process is quick and easy by means of a review screen
  • Follow-up process ensures that adaptation and improvements are properly organised
  • Vetting process for changes involves ‘can we – can’t we’ proposals carefully prepared by knowledge officers during the follow-up process
  • Technology can supply only the review screens and reporting functions, and support the follow-up process. All the rest is human brain and therefore ‘business process’
  • The law firm has to do this. Without it, KM systems will be a waste of money. It can’t be contracted out, and there are no ‘packages’ you can buy.

Is it still true?

This is the area where automation of the procedures by applying software todatabases is easier, and much more productive. It is also the area where lawfirms can make use of outside resources. The core issue is the dependency of alegal argument on ‘law’, that is, on legal precedents, case law, legislationand codes of practice.

Clearly, if an argument is founded upon authorities which are themselvessubject to change, whether by over-ruling, repeal, amendment, or even hostilecomment, then the argument itself potentially needs review, or amendment, whensuch an event occurs. Fortunately, there’s a way to handle this easily,provided one precondition is met.

The Cite is Queen

That precondition is the imposition (not in all firms: some do this already)of a discipline in the citation of authorities. If you want to base an argumenton the Estonian Copyright Act of 1996, at some point you must make quite surethat you have cited the thing in the proper way, so that later automated workcan establish that that exact piece of legislation is one of the authorities ofyour argument.

All this means is that, given accurate citation of authorities, and given away in which you can access any change in the status of the authorities, you canhave automated systems to announce, to the author of an argument, that thefoundations have shifted. Legal publishers are not blind to this concatenationof circumstances, and those with large databases are rapidly developing systemsto assist law firms with this issue. Of course, there is an issue withcomprehensiveness. I suspect that you need a very large, comprehensive, onlinedatabase to do this job, and the more experience I get with the incredibleinterweaving of primary sources, the more convinced I become that the days ofthe small, specialist database are over. Specialist services cannot include allthe vital, odd, effects of unlikely documents, and this makes them dangerous touse.

Analysing the effect of a change

Again, you can buy this. But whether you buy the analysis or do ityourselves, it has to be done. The fact of the change is not always simple. Acase that you employed as an authority in your argument might have beendistinguished in a subsequent hearing, but for reasons connected with anotherpoint of law than the one with which you were concerned. In such a case yourauthority stands – but it has to be assessed, and you might still need tobolster your argument with another, untainted, authority.

This is where the traditional one-to-many activities of the major publisherscan assist you. One of the things they do is to provide this kind of changeanalysis – though whether they do it fast enough, yet, is a question for youto answer and about which they need to provide convincing proofs. But in themore esoteric areas of modern commercial legal practice, the burden of thisanalysis is likely to remain, at least in part, with the fee earners of thedepartment involved.

How can automation help you more in the analysis of change? Already, you canmake use of tools and databases which will assist you to identify change in theauthorities you have used. Although not technically simple to implement nor isit overly complex, provided that your citation of your authorities follows theaccepted formats. Well-designed software and access to well-designed databasescan replace a lot of person-hours per year, in this area. In the next section, Iconsider a specific technique.

Central linking methodology and automatic back-editing

KM systems can be designed to assist their potential beneficiaries ifspecific rules for identifying citations are followed. Clearly, if a particularcase is overruled, one of the first and simplest things you need to know is:‘where, in my know-how systems, is this case hearing mentioned or discussed orused as an authority?’. It might be in quite a large number of differentplaces. Again, suppose that a department has followed a particular case hearingbut has consistently used a bad citation, derived from a mistake somewhere alongthe line which has been slavishly followed – as happens. In both instances,the last thing you want to have to do is to trawl through all the documents inyour know-how system every time any one single event occurs, and update thedocuments – hundreds of such events can happen in a week.

The methodology is quite simple, though it relies on the availability of arelational database to control the handling of the citations and their locationwithin your know-how system documents (ie handling the metadata, a word I hopeyou are all grateful to me for avoiding up to now). You identify the citationonce and for all, using some neutral list of authorities, and store the resultin a table in a relational database. The reference back to your own document isin the form of ‘the nth citation in document XYZ in document collectionABC’. Notice that you do not put any formal citation in your own document –just the identifying information. The reference to the thing cited (legislation,case hearing, etc) is similarly in the form of an identification of the databaseand the document identification. Thus if a correction is made by the originatorsof the second part, you have to do nothing – it just comes out right when youuse it.

When a change occurs to the thing cited, it is simple to get a list of allthe documents in your know-how system that might be affected. A grading systemis already applied by some primary law publishers that can alert you toreversals or less dramatic effects on authorities cited, and this trend islikely to be expanded in the future.

Summary of the ‘is it still true’ issue

  • Ensure accuracy in the original citations you make to authorities
  • Employ linking methodologies that allow you to be alerted simply to changes in the authorities cited
  • Do not rely on human intervention to discover the fact of the change – you can obtain this from publishers more cheaply
  • You may need human intervention to assess the effect of the change on your argument


Both the operations discussed above – ensuring that process and authoritiesare up to date and maintaining them in that state – require human resource.This is neither infinite nor cheap.

So you are bound to need an evaluation of the know-how system extent for eachdepartment. The criteria for evaluation, to avoid prohibitive costs in pointlessareas, will be:

  • How often does the business process change?
  • How often do the authorities (case law, legislation, codes of practice) change?
  • How many people are required, given best software assistance and data sources, to maintain currency and to discard bad practice?

Funnily enough, if the answers to both the first two questions are ‘veryslowly’, you probably don’t need know-how or knowledge management systems inthat area.

What is the resource required? It is divisible in this way:

  1. The two tasks of getting KM going, and keeping it going, require different skills and, quite often, different people. Therefore…
  2. The initiators and creators of a system will simply waste their time if the business processes to maintain the system, along the lines discussed here, are not planned right from the start. We do not have to look very far around us to see examples of apparently wonderful systems which have failed, or are about to fail, as revenue-generators, simply because this critical piece of initial planning was not carried out.
  3. The inheritors of the system will need to feel something other than that they are second-class citizens. These are the people whose day-to-day application of skill will turn fashion statements into revenue, and KM into the forward-looking, updated, managing of collective intelligence. Probably sourced as one person per department in a large firm – wherever they actually happen to work or to whom they happen to report – these people will ensure the success or the failure of the system. Identifying them, grooming them, and getting their buy-in early on, is more important to the long-term success of the firm’s KM operations than the sexy initial set-up stuff.

Accepting the need for these resources from day one as part of the system,and including their costs in the business case for each KM rollout to eachdepartment, will help to make your investment in knowledge management forwardlooking, deliver a constantly improving resource, and avoid what could otherwisebe a huge waste of money.