The Challenges of Developing and Maintaining a Precedent Database

Law firms have found, by hard experience, that creating and then – more important – maintaining a precedent library is a challenging task. Some firms have foundered in the seemingly straightforward process of simply creating firm precedents. Others that have succeeded in that task have found that ongoing maintenance is, if anything, even more challenging.

One firm that seems to have got it right is Gowling WLG. The person primarily responsible for this task is Graeme Coffin, the National Business Law Precedents Partner, Mark Tamminga, who is Partner, Leader of Innovation Initiatives, is working with Graeme on further automating the Gowling WLG precedent asset.

I hoped that they might be willing to share some of their secrets with the Slaw readers, so I contacted them by email and asked if they’d be willing to outline some of what they’ve done and they graciously agreed to do so. For the benefit of those who might be interested in more details rather than less, I have not edited down any of their responses. (The text is primarily courtesy of Graeme, who speaks in the first person, with some additions from Mark.)

Gentlemen, can you describe the current state of your firm precedents and perhaps outline what your status was ten years ago?

Today we have a mature system of over 1,700 annotated model precedents, which cover the field from annual resolutions to syndicated loan agreements. There are still some substantive gaps to fill, but the system is now transitioning from being under construction to being maintained, what I call the “care and feeding” phase, which for a system this big is a substantial effort. The big effort going forward will be in finding new ways, technologically, to deliver the content and ease its use, through automated processes, and linkages to other systems, such as our legal project management software. A key component of that effort will be to “productize” bundles of precedents, by linking them to document assembly, data bases, and workflow tools.

Ten years ago? Ten years ago, though efforts had been made, we had literally nothing, no precedents, and no knowledge management generally (at least no firm-wide knowledge management). I was hired at the end of 2006 to do something about that. Gowling WLG has been extraordinary in giving me the resources I needed to give them the current system.

It seems to me that a good precedent process is the result of figuring out the proper interrelationship between four key aspects, namely content, people, process, and technology. Would you agree? Are there other key aspects?

I suppose those aspects are central.

But there’s something else – marketing and education. When we started this, internal PR was crucial, and you always have to be reminding people that the precedent system exists, that it’s always growing and changing, and how to use it. We hold quarterly training seminars on line for anybody who wants to get a refresher, and announce major changes and new documents.

Technology is now becoming prominent, but really wasn’t, over the decade in which we created the system. The documents respond to macros on a custom toolbar that strip out annotations and the like, but my slogan starting out was “so simple even a lawyer can use it”, and we always have to keep in mind that anything even remotely complicated to use will put most professionals off. It’s all I can do to get them to type a word like “indemnity” into a search engine, and look at a set of folders to find the one labelled “shareholders agreements”. No joke.

So we look for certain qualities in the software and systems we acquire – they should be attractive to the eye, and fun and simple to use. We can’t turn drafting into a video game, I guess, but we can make it, from the mechanical standpoint, almost as easy as buying a few books on Amazon, or at least that’s the goal.

Our KM system, “Gordon”, (where the precedent system resides) is accessed via pretty-looking tiles of several pleasing colours. The KM database you access has the colour of the tile, so you always get a visual cue as to which system you’re in – like Monopoly, you know that Park Place is deep blue and Kentucky Avenue is red. I once suggested that instead of an hourglass or rotating object to indicate the system was chewing on a task, we should have it throw Pong up on the screen. Pong is like watching a Zamboni; people like looking at it. This was rejected, of course, but quirky fellow that I am, I maintain that if it had been done that way, lawyers wouldn’t be impatient – they’d be sorry that the search engine was getting back to them too fast.

For me, though, this effort was mainly about building an expert team that could approach the drafting process with rigour and commitment to certain standards. The lawyers who work for me are not of the “square peg” sort who couldn’t really adapt to mainstream practice. They were all hand-picked, plucked from existing practices, and induced to change roles and come work for the team; and each was picked on the basis of being superior to his or her colleagues in the areas that matter most to precedents – drafting skill, rigour, work ethic, and legal reasoning, plus that special temperament that loves meticulous work and getting things elegantly right. They’re a special crew.

Process of course is central to this, see below.

I never found content much of a challenge. Even though we have a lot of documents, they almost all cover “meat and potatoes” aspects of business law, and the only somewhat esoteric items on the system are there because, while rarely used, they are things unusually likely to involve risk. We have an escrow agreement for use when we, the firm, act as escrow agents for any reason, something we heartily discourage, but just in case. We have special indemnities for use by our own lawyers when they serve on boards.

Could you describe what happens when you develop a new precedent?

The key questions before embarking on any project are who wants it, do we do much of that sort of thing, and will they use it if we give it to them? There is no sense drafting stuff nobody asked for, or isn’t really likely to be used much. That said, in general, if any group says they need a precedent, we’ll draft it for them.

If at all possible, we keep the drafting entirely within the group. The precedent team’s lawyers have a broad set of specialties, and generally we have all the expertise required to avoid resorting to busy practitioners at all, and if we do need outside help, it will usually be for discreet issues concerning things like tax.

The team member with the most relevant expertise gets to hold the pencil, and be the “primary” on the document. We gather all sorts of examples of the document at hand, many from our closing book collection (which per force contains documents drafted by other firms), and also from the DM system, and then do a compare and contrast, arriving at a consensus composite. I know that KMStandards has developed software to do the same thing, and we’ve looked at it (and will keep looking at it), but for now, we continue to do it manually – my people are extraordinarily good at it, and it never takes too long (though it would be nice to have a box with flashing lights do it for us, no doubt).

We have a particular team member whose responsibility is to act as “scrubber” – she cleans up the language, grammar, checks for mistakes, proofs, edits, makes sure the boilerplate conforms to our models, and generally cleans it up in both substance and formatting to conform with our models. She stays on board throughout, even if she’s not otherwise a drafter, keeping the document on the straight and narrow, and will check it one last time before we post it.

The “consensus draft” is just the starting point. It will still be nowhere near our standards, and we will end up redrafting pretty much all of it to some extent, augmenting it along the way, and adding detailed annotations. The annotations cover matters both practical and substantive, and I’ve always said you could get a pretty good legal education if you ever had the time to sit down with our collection and read their notes.

This phase is time-consuming, but the special value of our system is that the documents are not simply a composite of other people’s drafting. I guess that means there’s another quality I need in my team members – they have to be good writers. Most lawyers think they are, but, well, mine have to be able to write like they work for The Economist. I tell them, don’t bury the lead. Keep it snappy. Tell the folk why they care.

We have an evolved set of protocols that cover every phase of the drafting, posting, and up-keep of the documents. We literally have a rulebook in a binder. As an example, to maintain version control, every document in the drafting phase is stamped “PENDING” and marked with the initials of the primary drafter, while being saved under a specific document type. While we collaborate on the drafting, only the primary drafter can feed markups to our administrator, and only our administrator can make electronic amendments. The only lawyer on the team even permitted by our DM system to make amendments to documents is me, and I’m not allowed to. Every change is noted in a revision log, along with the reason behind it.

There are drafting rules too. We have a style guide that prescribes how we draft all sorts of standard contractual components. For example, plain language in all cases is mandatory. I have a list of “plague words and phrases” in our procedures binder, and not only do our drafters avoid them, but before posting a document our administrators do word searches to root out any that might somehow have slipped through. At training sessions we issue an open challenge to the users to find any instance of “whereof”, “hereinafter” and so on in any document.

When we think we’re happy and ready to post, there are a number of technology-based steps we take. On the mundane level, we check all spelling and formatting (and look one last time for any plague words to root out). Its document type in the DM is changed to the special category that applies to our models, certain things are added – every document gets a unique embedded tracking number, for example – and the various macros are run against the document to make sure all is in working order.

It then has to be inserted into our models database in the KM system, and we do this through a custom interface that allows us to place it in as many places on the folder tree as seem advisable, while applying tags, or “facets” to it that can be used to find documents. The existence of the new document is recorded in two master indices, one in Word and one in Excel, and a revision log, now blank, is begun.

When we amend the document the process is different based on the reason for the amendment. If it’s otherwise fine and we’re merely improving it in some way, the document remains accessible, but if the law has changed and the present draft is no longer correct, we lock it down and put a message to that effect in the document’s name on the interface.

All documents under revision are copies of the posted document, get their own distinct document type while work is done on them, are Labelled “REVISIONS TO: [system document number], again with the responsible lawyer’s initials on the cover page, and then all proceeds as with a new document, except there is a tricky process when we save the revised draft over the original, which tricks the system into believing that no new document exists. This is necessary to preserve the viability of internal hyperlinks between the documents.

In future, we will take a further step with some documents, and code them into document assembly software through which users will access and modify them. We have a number of promising proofs of concept about to go to market and we are committed to a putting great deal more effort into this obvious next stage in automation.