Search
Close this search box.

LFaaS (Law Firm as a Service) III – AI and the shape of law to come

This blog develops some of the ideas in our May 2014 and September 2014 LFaaS blogs

So where are we – where is your firm – on the spectrum between ‘business as usual’ and ‘the
end of lawyers’?  As the market enjoys its best spell for a while, what will be the impact – the how, when and how much – of the new technology changes we’re hearing so much about? And how best to use the sunny intervals ahead of the next squall?

The financial crisis ended the seller’s market for commercial legal services.  In the current buyer’s market, technology is contributing to a shift in the rules of the game away from the historic law firm economic model, driving productivity up and costs down.  Even in the sunny spells, this is making demand more volatile: on demand provision (ODP) and legal process outsourcing (LPO) are approaching scale and starting to disrupt established buying patterns.

That’s the ‘now’.  Artificial intelligence (AI) is the next significant technology change on the horizon, and we will be hearing more about firms ‘getting an AI’ in months to come.  AI has four main components: first, a natural language user interface so everyone can type or speak their question in plain English (‘what’s the earliest date I can terminate this contract’?) Second, awareness of the context of the question so it can better analyse large datasets (like legislation, cases and in-house and online know-how). Third, generation of plain English evidence-based responses ranked from most to least likely.  Fourth, and most importantly, AI is cognitive – meaning dynamic: it learns from user feedback so next time it is asked the same question its answer will be better.

All the tech majors are working on AI – Apple has Siri, Google has Google Now and Microsoft has Cortana.  IBM has put $1bn behind Watson, its cognitive computing platform, and the best known illustration of legal AI at the moment is the University of Toronto’s IBM Watson-powered ROSS ‘digital legal advisor’, a runner up in IBM’s Watson competition earlier this year.

Legal AI falls squarely into the category of overestimating change short term and underestimating it longer term.  In due course, it will add significantly to the tools available for advisory work (research, report writing and compliance) and document review (corporate and finance due diligence, property title reporting), crunching (litigation discovery) and assembly (contract drafting and automation).  At the moment, you probably won’t get much change from six figures for an enterprise AI if you can get one at all, and it will be 2017 and 2018 before it starts to be adopted widely but in say 5 to 7 years’ time the price will have come down and everyone will have one.

So, what will the practical impact be?  It is worth considering this in the context of the historic law firm economic model, where fee income divides into thirds between direct costs (lawyers’ salaries), indirect (all other) costs, and equity profit. Within the model, property costs for a central London firm can be 8 to 10% of fee income and IT around 3 to 5%, so that total lawyers’ salary, property and IT costs can be about half total fee income, or between two-thirds and three-quarters of total costs.  It’s these costs that AI is likely to impact most, but in different ways.

A key feature of direct costs, especially in the mid-market, is the historic ‘qualification premium’.  For NQs, a typical starting salary is £60,000 and hourly rate is around £200. At PQE6 level, these figures are around £90,000 and £400.  These salary figures are derived from the 2015 Edwards Gibson salary survey at http://www.edwardsgibson.com/salaries.html shown in table 1 below.

Table 1

[table id=1 /]

But if the NQ salary is calculated rateably off the difference in NQ and PQE6 hourly rates, it would be £45,000, or 25% less than current £60,000 market rate.  The premium gets smaller each year up to PQE5 but the total premium on salaries for a group of associates at yearly PQ increments between NQ and PQE 6 on this basis works out at just over 10% as shown in table 2 below.

Table2

[table id=2 /]

So in a firm where all associates are at these levels, the aggregate premium works out at 10% of direct costs, or 10% of profits on the historic economic model.

This mismatch has helped open the door to ODPs and LPOs as they have lower resource costs and can charge lower rates for market acceptable service while still making an acceptable margin.  It is also one reason why the traditional pyramid is getting more pointy – more Shard, less Cheops.  AI will accentuate this trend.  When legal AI takes off, lawyer productivity will rise significantly.  This is likely in the short term to improve demand for associates nearer to qualification as AI will significantly enhance productivity.  Longer term, it is likely to reduce partner: associate leverage, sharpening the pyramid further and putting more pressure on the qualification premium.

Average office lease lengths are around 6 years at the moment (see for example the IPD/Strutt and Parker Lease Event Report 2014 – http://www.struttandparker.com/publications-and-research/publications/lease-events-2014/) so firms buying or extending a lease should review what these changes are likely to mean for them.  Law firms have got used to buying more space than they need to cater for anticipated growth.  But the leverage implications of AI and other changes like greater homeworking challenge that thinking.  Getting it wrong is costly – a law firm paying 10% of total fee income in property costs on a lease of six years or longer could have a current financial commitment equal to two years’ profits, as well as be paying for more space than it needs.

Getting an AI will add to firms’ IT costs.  In addition to AI itself, the associated planning, management and implementation costs will reduce profits in the first year – pushing up total IT costs to the higher end of the 3-5% range.

When law firm AI use has become widespread, the way it developed will look obvious with hindsight. Looking ahead, it is anything but. Yet it is reasonably foreseeable that AI will trigger significant change in the next 5 to 7 years, so it’s worth firms factoring this in to their longer range thinking now.  Even with the sun shining, demand is becoming more volatile.  When the next recession comes around (whenever that may be) and demand stops rising or falls, the impact of the combination of cyclical and AI/technology-driven structural change will increase pressure on the one third of total fee income that is left as margin.

Share:

More Posts

Send Us A Message