Graeme Johnston / 8 April 2022

This article combines two pieces I wrote in early 2020, just before the Covid emergency hit. Consolidating things on this blog, I thought they were worth republishing. I’ve made some minor edits to tie the two pieces together into one.

I’ve also updated it with some developments since 2020. But the gist remains unchanged.

Contents

Chapter 1: 100 years and counting

• GTR
• Purpose of this article
• GTR as a child
• Why do GTR? Back to the future
• The 1940 milestone
• Going viral: 1950s – 80s
• Smith 2.0: 1990s – today
• Two questions

Chapter 2: What next?

• The underlying needs
• ‘How can we see what people are doing?’
• ‘But we’re different’
• Enter GTR
• Problematic side-effects
• That’s all very well, but…
•  
• Automating GTR more cautiously
•  A further option – lose the ‘G’ from GTR
• Obstacles
• Concluding thoughts

 

 

Chapter 1: 100 years and counting

GTR

Discussion of what I’m going to call granular time recording – or GTR – in legal services — and of its progeny, time-based pricing — can be more than a little frustrating.

Critics understandably emphasise the negative effects on clients, lawyers and society more generally. Such criticisms are weighty. But the question always arises: what does ‘better’ realistically look like in an unpredictable world?

Failure to answer this convincingly has left GTR solidly in place; indeed, forever being optimised.

Time-based pricing has been challenged up to a point. However, the reality is that, for complex, non-repetitive, non-automated work, ‘alternative fee arrangements’ (AFAs) (i.e. fixed fees, fee caps and other variants) do not often depart significantly from time-based pricing.

The practice driven by some corporate clients of ‘shadow invoicing’ (i.e. requiring a record of time spent even where a fee has no time element) tends to reinforce the link because it incentivises over-working. The excessive time shown allows the client’s agent to demonstrate theatrically the ‘good value’ of the agreed fee against doubters, while also helping individual time-recorders within the law firm to meet their targets – a win/win/win for the seller and the respective agents of buyer and seller. 

Recognition of the seriousness of the problem, and ideas for challenging it, are not new. For instance, a report by a commission of the American Bar Association (ABA) almost 20 years ago listed many of the system’s corrosive effects while also articulating why it has remained so entrenched.

A working group of the New York State Bar Association radically recommended in October 2021 that billable hours be limited to 1,800 a year. We’ll see whether that gets anywhere. There is no sign of anything like that being considered in England, where the main legal regulator in February 2022 assumed that practising law involves ‘long hours’ and stated that it would only take disciplinary action against law firms if they ‘impos[e] wholly unreasonable workloads or targets’ on their people, without any attempt to quantify the meaning of this.

As Richard Tromans has recently observed in relation to the widespread wish on the buy side to end the ‘billable hour’ —

The problem is there doesn’t appear to be a clear path to reaching this outcome — and so we are stuck with ‘End of the Billable Hour Groundhog Day’, with the anachronistic approach to legal production always just about to come to an end, but never quite expiring.

Purpose of this article

I propose to explore whether there are systematically better, realistic ways to address the relevant problems with less harmful side-effects.

My tentative answer is ‘yes’. However, to set the scene I think it is useful to recap some history in order to understand how we got here and so that we can, perhaps, avoid implementing a higher tech version of the same old problem.

GTR as a child

This appears to be the approximate centenary of systematic time recording in legal work.

Its principal early promoter, Reginald Heber Smith, appears to have started work on it in earnest around 1920, shortly after he joined a New York business law firm.

However, Smith did not widely publish his method until 1940 and it seems not to have become dominant until the 1960s in America; later and more patchily elsewhere.

Why do GTR? Back to the future

GTR is best known these days as a basis for time-based pricing. However, one of the key talking points of today’s time recording software vendors is that, even as departures from pure time-based billing become more common, more detailed time recording is still worth optimising as a way to better understand how people are spending their time in order to give management the ability to do x, y or z in response.

It seems that Smith’s original aspiration was essentially the same — to use what we would today call ‘big data’ for varied management and process improvement purposes. ‘Taylorism’ in the jargon of those days.

The website of Smith’s old firm has an engaging summary, apparently based largely on a memo he wrote many years later (1966).

It seems that, when Smith started work on time recording around 1920, his motivations were indeed Taylorist and that he was drawing on his experience in the 1910s to improve the organisation of legal services for the poor.

As an indication of the spirit of times, the U.S. Chief Justice was already complaining in the 1930s about the emerging large law firms as ‘a new type of factory, whose legal product is increasingly the result of mass production methods.’ (Galantar and Paley, Tournament of Lawyers, 1991, p17).

There is no indication in those early days that Smith’s motivation was to promote time-based pricing. However, he and his firm appear to have ventured down that path at some point before 1940, perhaps in stages as the business logic of each step impelled the next.

The 1940 milestone

In my beginning is my end.

T.S. Eliot, East Coker, 1940

In seeking to understand how time-based billing became a dominant model, I found it illuminating to read Smith’s four part series ‘Law Office Organization’ published in the American Bar Association Journal of 1940. You can read it for free on JSTOR.

There is much value there, including on the breakdown of tasks and their allocation to appropriate staff. Lessons still not fully learned to this day.

However, for present purposes the important point is the emphasis put by Smith on two specific needs which could be met by time recording and time-based pricing.

The first need was to cover unpredictable costs and still make a profit. Specifically, a law firm’s costs (partners’ drawings, staff salaries, rent) are largely time-based yet, Smith observed, it was often unpredictable how much of the time in question will be spent on a particular matter. In Smith’s own words (from part 2 of his 1940 series):

If you ask an architect how much it will cost to build a house, he cannot give even an estimate until he has seen all the specifications. No client can give to the lawyer comparable specifications. For example, neither can then tell how much of a fight the opposing party intends to put up. If you beat him in the trial court and he accepts the verdict, that is one thing, but if he is determined to appeal, that is another.

The fact that the US litigation system was chosen as an example is perhaps revealing.

The second need emphasised by Smith was to measure relative contribution for the purpose of determining reward within a partnership. Smith expresses this as a ‘quota’ in $ terms but then, crucially, states (part 3):

Now we can take the next step. The lawyer with the $5,000 salary has a quota of $10,000. He has to earn $10,000 and he will try to do so by selling his time. Every man is expected to work a standard number of hours a year.

The series concludes by explaining how three factors (hours spent on client work, the amounts billed and the gross profits generated) can be analysed to ‘arrive at a judgment as to each lawyer’s worth to the firm.’

Smith admits that hours recorded, and the other two metrics, are ‘substantial’ rather than ‘conclusive’ evidence of contribution but suggests them as a simple way of prorating credit between partners while avoiding ‘ad hominem’ argument. The imperative for a managing partner of a growing firm (as Smith was by that time) to figure out an efficient way of minimising disputes is apparent.

Smith certainly should not be regarded as a simple solutionist:

all the qualities and values of a lawyer cannot be caught in a statistical net no matter how finely spun… But the question is, what better method is there? It is impersonal, it is open and above-board, it does yield a vast amount of information….

Nevertheless, as his old law firm observes (in the article already mentioned):

Extrapolated from one firm to an entire industry… and fueled by forces that he could never have foreseen, Smith’s formula for fairness and efficiency was to become something else altogether.

Going viral: 1950s – 1980s

The use of time recording data as the basis of a pricing model seems to have been adopted by various leading Manhattan firms in the 1940s. Assisted by promotion in a pamphlet issued by the American Bar Association to its members in 1958, time-based billing evidently became the norm in American firms in the 1960s.

It is revealing to note how little the basic numbers have changed over the years. Three examples make the point —

• Smith’s 1940 article reported billable hour targets in his firm of 1,600 for associates, 1,520 for partners (dropping to 1,200 for the over 50s who are expected to devote more time to charity and public services, while also prudently slowing down).
• There is evidence of billable hour targets rising to 1,800–2,000 hours in leading US firms by the end of the 1960s. (Tournament of Lawyers, pp 34–5).
• Decades later, a new bonus scheme announced in Jan 2020 by a US firm’s London office is based on billable hours of between 1,700 and 2,000.

Smith 2.0: 1990s – today 

When I started law firm work in New York in 1994 then London in 1995, time recording was still done via the weekly paper grid invented by Smith, still running on his original decimal unit system (1 unit = 0.1 hours = 6 mins). Time-based billing was also, by that time, the general rule.

Speaking to people with longer memories, it is clear that there was a significant lag, in London at least, between the introduction, first, of time recording, then of time-based billing as the default rule. But, as in Smith’s firm in the 1920s and 30s, it seems that one thing led to another.

More recently, attempts have been made to build on Smith’s system. I would sum these up as having two major strands, with a third possibly emerging.

Strand one: ‘capture more qualitative data’ 

This started to be emphasised in the 1990s and 2000s.

By the end of the twentieth century, most lawyers were expected to record narratives and/or work codes in a computer system. The codes were defined by the firm, the client or some other source of authority (e.g. court rules).

For some lawyers, this was in fact happening earlier (indeed, Smith’s original system made provision for it) but so far as I can figure out it didn’t become almost universal until around twenty years ago.

I imagine that the reason was that such recording imposes significant extra burdens on the individual lawyer and was regarded only as feasible when each individual had their own computer.

This development certainly created some insights, but having examined rather a lot of the resulting data over the years I would sum it up as being of mixed value and hard to digest.

Overall, I would say that strand 1 was not exactly a cost/benefit triumph though certainly some people made good use of it, and law firms have largely embraced it as a defensive mechanism in the event that fees are challenged.

Strand two: ‘capture more; and do it more smartly.’

Since the 2010s, significant efforts have been made by software vendors to

• capture what people are doing by more automated means so as to reduce the burden on individuals and hence the time available for other activities
• report more digestibly on what it all means e.g. by natural language processing of billing narratives and visualisation of the results

There are some useful early results from such products, though adoption is still limited and feedback appears to be mixed. Work in progress.

Strand three: the quantified self.

Tellingly, some law firm time entry vendors claim that their tool is helpful to the individual for the same reasons as personal fitness apps and the like.

It remains to be seen how much traction this will gain and what access to data will be offered to the individual user.

Two questions

The first hundred years of Smith’s experiment illustrate how observer effects and other unintended consequences arise. Time recording is not a neutral act. ‘You get what you measure’ is a cliché for a reason.

The two questions I propose to explore in a follow-up article are

1. What’s the real impact of going ever further down the automated time recording path?

• How is it likely to affect behaviour for better and worse?
• Can granular time recording be uncoupled successfully from time-based billing?

2. Could a law firm which refused to record time in a highly granular fashion succeed in competing on complex, non-repetitive work and growing beyond a handful of individuals?

• What are the real needs which such time recording addresses and can they be better met?
• What would happen if such a firm simply required a rough self-certification of the proportion of time spent on different topics?
• Or is granular time recording now so essential to legal work, so expected by clients, and so much an element of internal accountability that there is no realistic alternative other than for very small firms or those not undertaking genuinely complex work?

I don’t suggest there are easy answers but it seems worth talking about these things rather than just sliding uneasily into a second century.

Chapter 2: What next?

In a recent article, I traced the history of granular time recording in law firms. ‘GTR’ for short. It involves capturing a record of activity down to the hour and minute level.

GTR in legal work began life a century ago with reasonable intentions but is widely thought to contribute to over-working — ‘You get what you measure.’ This has obvious adverse consequences for lawyers’ clients (over-complication, priceyness) and some more insidious ones for lawyers themselves.

There is also a transaction cost — the time spent entering and describing time and (in some cases) analysing and arguing over it.

However, the big question remains — is there a better alternative to status quo GTR? One which can meet the relevant needs with fewer unhelpful side-effects.

This article discusses three options.

1) Automate GTR to the max

2) Automate GTR much more cautiously

3) Zoom out: make time recording less granular

Before assessing these, let’s look a little more closely at the needs which GTR seeks to meet.

The underlying needs

With the onset in recent centuries of more complex products reliant on inputs from many people, there has inevitably been a desire to see and understand individual contribution with a view to recognising and improving it; and also (let’s be realistic) in an effort to reduce slacking.

Several answers have been attempted over the centuries, but they tend to boil down to

• measuring useful outputs at a particular stage of a process
• making visible what people are doing, i.e. their inputs

The first is obviously more useful but can be hard because of the difficulty of defining and measuring quality. The temptation is therefore to focus on the second.

‘How can we see what people are doing?’

An early engineering effort to improve input visibility more efficiently than by employing large numbers of floor supervisors was the 1780s construction of a prototype Panopticon building by the English naval engineer Samuel Bentham to help supervise a shipbuilding project for Potemkin on the Dnieper. The notion became more famous later when his older brother Jeremy promoted it in England for penal use cases.

Panopticons proved to be a dead end in the context of physical manufacture. The idea manages to combine trust-harming surveillance with gameability (‘Look busy!’) and anti-efficiency incentives. Instead, nineteenth century (and subsequent) industrialists became better at observing and measuring the quality as well as quantity of outputs more closely.

That said, factories typically also maintained supervisors, assisted by time clocks to measure the basic input of presence on site or, in some cases, the time taken to complete a particular task (time clocks are a technology I’ve only come across rarely in law firms, though one large firm used them when I worked in Hong Kong and apparently a few others have tried it).

‘But we’re different…’

In the field of legal work, the nature of outputs presents more of a challenge than in manufacturing. The problem is not so much the intangible nature — that’s not an issue if the scope can be made clear and the process and time required can be highly defined. The problem, rather, is the need to deal with the uncertainty as to the level of effort imposed by the client’s particular situation and, sometimes, by the actions of an unpredictable counterparty or authority figure.

The definition of quality and the cost / benefit trade off in bespoke legal work can also be debatable — ‘quality’ in legal work is sometimes a question of reducing hard-to-define risk. And defining scope and value in bespoke legal work both take time, i.e. impose a transaction cost of their own.

The narrative that much legal work is inherently bespoke and unpredictable has, of course, been challenged. The challenges are, in my view, justified up to a point. But they’re also often exaggerated. However, I’m not proposing to go deeper into that in this article.

Enter GTR

My earlier article summarised the history of GTR. The major reasons for adopting it boil down, I suggest, to three.

1. Understanding what people you’re paying for are doing — with various motivations, such as challenging inefficiency or reassessing where to spend money because the value (or risk averted) simply isn’t worth the cost.
2. Pricing, in the absence of a better, mutually acceptable way to determine value which does not involve undue argument (given the subjectivity of value), unacceptable transaction costs or (for the vendor) the risk of failing to cover time-based costs such as salaries, rent.
3. Assessing relative contribution within an organisation, ideally as part of a balanced scorecard.

In modern times, further justifications are sometimes given (e.g. managing yourself more effectively, having a defensible record of what was done if sued) but I believe the three points just outlined are the main drivers.

Moreover, although point 1 seems to have been the original motivation behind the introduction of GTR in 1920, it has often proved more aspirational than real. In contrast, points 2 and 3 are very real indeed. These are the points which account for its widespread adoption.

Problematic side-effects

We shape our tools and thereafter they shape us.

Culkin / McLuhan, 1967.

The problem with GTR is that, in practice, it seems to increase risks such as

1. Rewarding inefficiency and disincentivising investment in efficiency. GTR, in short, gets in the way of necessity becoming the mother of invention.
2. Harm to individual health by encouraging over-working.
3. Devaluation of contributions which cannot be immediately monetised.
4. Exacerbation of emotions such as distrust (‘be ready to explain how you’ve been spending your time’), fear (‘will I make my billables target?’) and futility (‘I’m just a billing machine’) — with all that flows from this. Alienation, in short.
5. Significant transaction costs, i.e. time spent reporting, and in some cases analysing, how time was spent. There is a lawyer in-joke about wanting a time code for ‘time spent recording time’.

That’s all very well, but…

the question is, what better method is there?

Reginald Heber Smith, 1940 — see chapter 1 for context

There has for several years now been great interest in the legal world in addressing problem 5 (transaction cost) with technology to ‘passively’ (that is, automatically) capture activity. There is also a theory that passive capture may, in the right hands, be able to help significantly with problem 1 by increasing insight as to what is really being done.There are various tools which seek to help with this, with varying levels of success so far. They track which applications a user is using and attempt to interpret what’s being worked on, using natural language processing and other techniques.
There are two factors which suggest that some caution may be required.
First, the history of the last century of time recording in law suggests that some managers will conclude sooner or later that, if we can auto-capture X more hours of billable time from you per year and also save you Y hours a year reporting your time, then your annual billable hours target should go up by (X + Y) hours.
Secondly, there is evidence from other fields to suggest that granular automatic surveillance has bad side-effects. Highly sophisticated automated performance monitoring has been going on for some years now in ‘unskilled’ and ‘semi-skilled’ work. Its human consequences appear to be quite troubling. Josh Diesza’s recent article is a compelling introduction: The pattern is that enhanced metrics empower managers to squeeze more from workers, and the managers are themselves squeezed from higher up. The fact is, also, that such tracking has been going on for some time in skilled financial services IT work off-shored from western countries to India. An attempt to bring the same software system onshore into the UK financial services industry received negative publicity last month. Margaret Heffernan has argued that the risks of this sort of approach in the ‘white collar’ context include devaluing time spent chatting with colleagues unmediated by software and ‘[t]reating people like robots [with] the very real risk that they will behave like robots.’
Having looked online, the software vendor’s answer to concerns of the type just expressed tends to be that implementation is a matter for the customer’s judgement. Don’t blame us, we only created it. The typical answer from said customer seems to be that the tech now exists and it’s a tough competitive world. If we don’t do it, others will. Don’t blame us, we didn’t create it.
Neither of which is exactly reassuring.

Automating GTR more cautiously

I would suggest that there is a case for applying the precautionary principle before plunging too deeply into automated time recording.

If the automatically collected data is in the full control of the individual and only goes further once approved or edited by them, then one can readily see how it can offer benefits in relation to problem 5 and, possibly problem 1, but it really needs thinking through.\

To give a few examples

• If a timekeeper enters a certain amount of time for task X after concluding that the system either over-measured or under-measured it for some reason, should the original data be maintained (and, if so, for what purpose) or over-written? And should data of this sort be collected and analysed ‘anonymously’ by the organisation, or is that simply going to translate into more pressure of an undesirable type?
• If you have a genuine interest in addressing problem 1, then what are you going to do to help your people think about and sub-divide their tasks proactively and record time to those? Simply generating more structured data is not enough.

I would say that the topic cries out for an incremental approach with careful assessment at each step of the likely benefits and possible unwanted consequences.

A further option – lose the ‘G’ from GTR

This option may appear eccentric because it involves questioning the orthodoxy that more granular, structured automatically collected data must be beneficial.

But imagine a law firm in which the following practices are applied.

• Nobody records hours and minutes. Instead, each week people certify an approximate % of time spent on the affairs of each client and on major categories of non-client activity (e.g. management, know-how, sales). Not the absolute hours, just the %.This is aggregated over a reasonable period (e.g. quarterly and yearly) to understand broadly what people are doing. Non-granular time recording, in other words.
• Contribution is assessed without regard to recorded time. Other industries manage.
• Training is provided in project management and the use of appropriate tools, and an effort is made to introduce a continuous improvement culture.
• Pricing and associated cost risks are handled in three ways.

(1) Predictable elements of work to be priced based on delivered units of value i.e. outputs. More effort than at present would be put into defining and delivering these efficiently, by automation or highly defined processes.

(2) Portfolios of matters with unpredictable work elements are priced on a per matter basis in a way that will average out, with a suitable escape clause for the occasional case that blows up into something radically more substantial.

(3) The third way — and the trickiest — is that pricing of genuinely complex, unpredictable, substantial work should be fixed in advance based either on a fixed scope (where possible) or (where not) on a good faith commitment of an approximate fragment of a defined person’s or group’s activity over a fixed or recurring period, with agreed assumptions as to the scope of activity. The idea is to allow for genuinely complex, unpredictable work to be handled in a flexible way while covering costs. If the matter lasts longer than initially expected, the period is extended by agreement. Some mutual trust is required. A relationship, not just a transaction.

In theory (i.e. leaving out the human element), you might think that there is no reason why these three ways of pricing cannot be optimised even with GTR in force. In practice, however, with GTR in place, time-based billing is such an easy way out that abandoning the ‘G’ of GTR may be necessary for real progress, as a sort of Ulysses pact.
And that’s quite apart from the potential human benefits of dialling down the ‘G’ in GTR.

Obstacles

• For an existing law firm of any size, the cultural change will be tough to understand for many individual lawyers, leading to potential stability challenges.
• Some clients used to the traditional approach may tend to cherry pick. That is, to welcome more predictable pricing while undermining it by demanding ‘shadow billing’ of hours spent. My experience of shadow billing suggests to me that it incentivises work practices which show the client a ‘discount’. There has always been an element of co-dependency about GTR, hence its resilience.
• The grip of time recording has become so strong that it has even inveigled itself into court procedures (e.g. in England & Wales) for assessing the reasonableness of costs.
• That sort of entrenchment will take time to challenge: there has at least been a recognition in recent years in England & Wales of the need for restriction of budgets in litigation, though the rest of the system has not yet been adjusted satisfactorily to make this work properly.
• Cost bases (e.g. salaries, rent) are still largely time-based and still have to be covered. Creating a divergence between revenue and cost models is in some ways attractive but imposes pressures of its own.

Concluding thoughts

I don’t have a magic solution to offer (sorry).

However, the potential benefits of liberation from the accidental tyranny of time recording are significant enough to make it worth at least thinking about answers which go beyond simply removing time entry frictions.

In reality, I suspect that GTR will be with us for a while and that it will be automated in ways that go too far. Departing from it will be really hard, perhaps practically impossible where it’s already established.

That said, for a new, or at any rate smaller, law firm or a non-partnership legal services provider, it should be easier to build a business not dependent on GTR. Such an approach could differentiate the firm in a positive way, for both clients and team.

And for more established law firms, maybe that’s a motivating, disruptive threat worth at least reflecting on.

Human nature is such that no system will ever be perfect, but I offer this article as a contribution. My aim has been to recognise the very real needs which time recording seeks to address as well its problems, and to discuss the options with a degree of nuance lacking from the all-or-nothing approaches to the question of time recording.