Hacker News new | comments | show | ask | jobs | submit login
SimpleLegal (YC S13) Reduces Legal Bills With Machine Learning (techcrunch.com)
138 points by outericky 1571 days ago | hide | past | web | favorite | 45 comments

Many may be unfamiliar with how legal billing practices have evolved over the past years in the U.S. At the risk of boring people, I'll try to sum it up as I understand it. It does have some interesting aspects.

In the days when law was more a "profession" than a business, blue-chip clients would pay the bills of blue-chip firms based on, e.g., a $50K invoice that said "For services rendered." In a typical case, no further explanation given and none requested. This probably was a vestige of the "old boy's network" by which top lawyers and top executives came from a similar elite pedigree and it was regarded as undignified to question professionals about the mysteries of how they worked their craft. (This goes back before my time but I have it on good authority from senior partners with whom I worked circa 1980 in prestige firms).

That represented billing at the elite level of law. In the mundane, everyday practice of law, the common practice in billing was indeed that of fixed-fee services. In pre-Silicon Valley days of Santa Clara County in the 1950s, for example, sole practitioners would bill for basic services based on a fee schedule put out by the local bar association (e.g., "Preparation of will - $400"). Those were the days when local bar associations held much more sway than they do today and, in order to practice, a local lawyer had to do such things as sign up to represent criminal defendants in murder cases as part of an obligatory process (these were the days before the Supreme Court rulings made the right to counsel in capital cases a much more involved process than it is was back then). In any case, it was the courts that caused time-based billing to be foisted on clients as a matter of routine when they struck down the local bar fee schedules as illegal price-fixing.

The pattern above is pretty evident: in all cases, lawyers as professionals were seen as part of a closed network by which they themselves set the rules that everyone else simply had to accept. In billing, that meant that lawyers billed as they deemed fit without much scrutiny and without much explanation.

The scale of law was also very different then: back in that day, a prestige law firm in San Francisco, for instance, might have 20 to 25 lawyers at the most. The work it did was sophisticated but a typical high-end practice might have included heavy doses of such things as personal injury, maritime practice, construction, and the like in addition to sophisticated commercial and contract work.

As litigation exploded in the 1960s and 1970s, as law came to fill social needs previously not considered within its province (e.g., non-discrimination laws), and as mechanisms were adopted to promote large-scale legal actions (e.g., class actions), law morphed from having been primarily a "profession" to being primarily a business. The McCutchen (now Bingham) firm that I worked for in the early 80s had about 250 lawyers then and was considered elite. In 1965, it has 20 lawyers, a level to which it had grown from its formation in the 1880s. Today, it has well over 1,000 lawyers and is part of a conglomerate firm with branches in many cities throughout the world. Along with this growth came the rise of hourly billing practices that were typically "aggressive." Demand was such that large companies sopped up the services and paid dearly for the privilege: high rates, team-based billing practices with multiple attorneys and paralegals typically staffing a matter while each billing every minute of all their interactions with one another in meetings, chats, intra-office communications, etc.

In the beginning, as this explosion in growth occurred, clients were reticent to question the resultant high billings. By the mid-1980s, however, the remnants of the old-style "profession" were rapidly fading and law was a big business. Firms such as Finley Kumble were suddenly formed out of nothing by defecting big-firm partners who saw large and fast profits as the primary goal of their business and the billing practices quickly became abusive (Finley became a poster child for this sort of thing and fell apart owing to client revulsion at such abuses).

Thereafter, large companies began to question billings much more closely. The rise of in-house lawyers was in part a reaction to the high cost of outside services but, in time, those lawyers were also used routinely to "manage" the outside firms by keeping their billings in line, among other things. After a while, as the go-go years of the 1980s and 1990s culminated eventually in the tech bubble of 2000, a funny pattern emerged by which a big-firm billing was almost in the nature of an "opening offer." It would routinely come in very high, the in-house people would demand it be reduced, and, after some back-and-forth, "adjustments" would be made and it would be paid.

The pattern lasted until the bust of 2008 and 2009, which bust sent the big firms reeling and has caused a further re-evaluation of billing practices. The reason for the further change was simply one of supply and demand. Pre-bust, no matter what the abuses, the law firms held ultimate sway because even the largest companies with the most sophisticated in-house staffs would be wary of switching firms easily or of wanting to alienate their main outside firms in any way. That changed with the bust. Suddenly, the large firms looked bloated, overstaffed, and inefficient. And clients, who came to hold the leverage, used it to demand billing concessions: lower rates, caps, fixed-fee arrangements, discounts, and the like. And the law firms had to adjust.

In the meantime, quite apart from the big firms, technology has changed law firm billing practices generally. Owing to the widespread availability of information today, clients are much more savvy about how legal billing works and are not hesitant to ask for arrangements that make sense for the client even while being fair to the lawyers providing the services. Thus, scrutiny of legal billing is taken as a given and we have indeed come around 180 degrees from where it all used to be.

What SimpleLegal is doing is thus very timely and interesting. Honest and fair billing should be the goal of both lawyers and clients and this seems like an excellent model for helping to promote this. I wish them well in their endeavors.

Excellent write up!

A couple of additions:

First, there are a few practice areas where the old model still reigns, generally those where the courts approve bills rather than a client. You still see bankruptcy conferences where each creditor committee and the trustee bring two partners and three associates -- all billing back to the estate for a total of five figures an hour.

Second, another big change was a series of rule changes and ethics panel decisions which allowed for contingency fees in many areas of the law. Traditionally such arrangements were considered akin to barratry. While most people imagine that contingency fee cases are all small law firms doing personal injury, there are some types of very large cases that are done that way (i.e. qui tam litigation).

Great write-up. My big concern is that legal billing goes the way of medical billing where the whole thing becomes a game. Our goal is to get to clean and fair bill where the client is happy to pay for services rendered and the law firm feels well compensated for a job well done.

Not sure if this link is behind a paywall, but this visualization of Profit per Partner for the top 200 law firms show why people may feel legal billing has gotten out of control: http://www.americanlawyer.com/PubArticleTAL.jsp?id=120260182...

I'm not sure what the visualization is supposed to demonstrate. PPP is a function of supply, demand, and leverage.

As an economic matter, billing methodology cannot have an impact on the overall profits of law firms as a group. At the end of the day, the clients pay for a service--hours are just an imprecise way to allocate "scope of work" risk between the client and the service provider.

If your software says that one firm is charging 10% too many hours relative to the other comparable firm that also does work for you, that can save you money in that instance. But it cannot save you money if it says every firm is charging 10% "too much." Then that's just the market price of the service--if you cut down those hours, hourly rates will just increase to achieve the same price, at least in the long-term equilibrium.

In economic terms, the real value of your service is reducing transaction cost inefficiencies by offering better information and allowing clients to better allocate "scope of work" risk between themselves and outside counsel. But real reductions in legal bills, not achieved through finding efficiencies in the billing process, must be rooted in supply and demand: either increasing effective supply by considering a broader range of law firms for work, or reducing demand by bringing work in house.

The reference to medical billing is interesting, but I don't think your explanation is economically sound there either. Medical bills are not high because hospitals charge $200 for an asprin. That's an artifact, not a root cause. Medical bills are high because supply is limited and demand is high and inelastic. Software that checked medical bills for high-priced pills would not reduce bills--the prices would just rise in other parts of the bill. Because at the end of the day the customer is not paying for the pill or for a stent, he's paying for an overall result and the price of that result is dictated by economics, not the specific costs of specific billing items.

Instead of charging by the month or by the total percent billed, it'd be cool to charge a subset of the amount saved.

I've seen this work with other companies. For example, http://www.sourceconsulting.com/ which monitors FedEx and UPS bills for errors and refunds.

There's also a legal company in New York that courier companies use to fight all of their traffic tickets (which really add up). Whenever they lose the appeal, the legal company gets nothing, but if they win, the courier company pays something like half of the ticket price to the legal company. It really makes it a no-brainer to sign up.

Co-founder here. Charing a percent of savings is an interesting model. We approached customers with it and they pull back from it fairly quickly. It sets up a conflict of interest in that there is an incentive to reduce more than is appropriate. Our goal is not simply "less." We believe customers should pay what they owe, but not more.

We also believe that the relationship with your law firm is important. Just cutting a bill doesn't help that relationship.

Law firms have an incentive to support a product that makes billing more predictable. Right now, they engage in a lot of ad-hoc discounting and writing off of hours to meet ad-hoc budget targets, and at the end of the day clients don't pay a substantial fraction of their bills (15% or so industry-wide).

Say the general counsel at a F500 agrees with his outside counsel that a particular motion needs to be filed in a litigation. Right now, he has an idea in his head of how much it should cost, say $75k. Or he solicits an estimate from his outside counsel how much it should cost. But that company has probably filed hundreds of such motions in litigations over the years. There is a lot of data to be used to make better guesses. If the software says: this motion will cost $75k with a standard deviation of $15k, based on that company's history with similar firms, then the company benefits from predictability, and the firm benefits by having a realistic budget and some objective reference they can point to when the bill comes due and they ask: "please pay this in full."

I like your thinking! We believe that billing should be fixed, not just reduced.

We're starting on the side where we have the most experience and have been able to get the most traction. We see providing evidence-based billing as an important part of the move to flat fee billing (or alternative fee arrangement).

Our end goal is to fix the process. We want to make sure legal billing doesn't become the joke that is medical billing.

Thanks for the feedback. We have evaluated this type of pricing model, but past experiences in larger settings make it unattractive. Basically the incentive is to reach the threshold, and then scale back just a tad. We don't want to create a bad relationship with a company and a law firm, but we do want to make sure they are paying the right amount. Firms serve a great purpose, and they should be compensated; the correct amount.

How does this interact with lawyers' terms of engagement? Under terms of engagement you'll be incurring fees and by the nature of an invoice you can't necessarily contest this after the fact.

Unless of course you agreed with your lawyers that any invoice would be subject to review by SimpleLegal although I would suspect there would be some opposition to this. Although of course doesn't take away from the value of the service in allowing a company to monitor legal expenditure and to gain greater insight into this (possibly with a view to obtaining leverage for future fee discussions).

On a final point, your privacy policy already assumes that you will use data to improve your service and to provide external reporting. This would seem to imply a degree of aggregation. Although as stated this would be on an anonymous basis, the very fact it happens may be off-putting to certain companies.

Right now it's a buyer's market for legal services. Regardless of what the retainer says, if you call up the relationship partner and point out that you were just billed $1500 by a second year associate for photocopying he will work with you.

Beyond that larger clients are negotiating (imposing really) explicit conditions into thier agreements to prevent abusive billing. Things like minimum detail requirements (no more "5hr - lgl resrh"), no first year associates, no clerical work, etc. Then there's capped and flat fee billing.

Definitely a time of big changes.

Good points. It would definitely be a good feature if explicit retainer conditions could be fed into the system to analyse bills, and certainly a client is in a strong position where associates are photocopying.

The more interesting point would be like in the example provided in the article (billing half an hour for mailing) where the system considers the charge for the stated work to be excessive but the client may not be in as strong a position to question it as with the clear cut example you gave.

Definitely interesting times though!

Most large companies (banks, insurance companies, large retailers) do have strict billing guidelines as to what firms can charge for.

We (SimpleLegal) will work with those guidelines and train our models accordingly.

Example: Walmart's Outside Counsel billing guidelines http://www.acc.com/advocacy/valuechallenge/toolkit/loader.cf...

You know your way around a legal bill!

As bradleyjg points out, some charges make their way onto a bill when they shouldn't. It's usually not malicious.

The problem is that legal bills don't get enough attention from the law firm. They rely on the client to tell them when something is a problem. We make that process easier. And hopefully everybody, even the law firm, is happier.

As for the privacy concerns, we use data to train our machine learning models. The outcome of those models includes categorization by activity (internal meetings, phone calls, legal filings, etc.).

The "shared" component is for training purposes. At no point do we share invoice content. The key to convert a numeric vector back into a partial view of the original text changes every day. Security and confidentiality of customer data is a big area of concern for us. We also provide our customers with a named list of people with access to the content.

There is a lot of this sort of routine business communication the review of which can be automated. I'm thinking of invoices, etc, of all sorts. I wonder about their privacy story: if I'm CC-ing my bills to SimpleLegal, how are they protecting that data? Are they aggregating between customers, etc? Sensitive information about legal strategy can appear in bills, so it's important to have a good story for potential customers of this service.

Also, there might be a business opportunity on the flip side. One of the things law firms worry about is collection rate: the percentage of bills that are paid. Industry-wide, the rate hovers in the mid 80% range. Could this technology be used to get clients and firms on the same page and boost collection rates (even if it resulted in discounts elsewhere)?

Co-Founder here - Security and privacy are one of our highest priorities. Our encryption, storage and transfer policies are as good as you'd expect with any other provider dealing with sensitive information (banking, contracts, etc).

CC'ing the bills is a convenience we provide since most companies already receive their invoices by PDF via email (or by snail mail), we also support direct upload by the firm or by the customer. No human eyes, except the customers, see the data unless absolutely necessary, and that access is limited to select, named, screened individuals.

We currently do not aggregate data or share data in any way. There may come a time where we provide insights into billing practices or fraudulent activities in aggregate, but that is a ways out.

While we agree that incremental billing (in 6 minute increments) [edit] is a pain for both lawyers and customers[/edit], it's not a model that can be wholly abandoned. However, we do plan to be in the forefront of fixing the headache and process.

rayiner makes a great point about (and possibly created customer demand for) the security of your service being a selling point. I'm wondering if billing firms could fall under client-attorney confidentiality, but I doubt it.

On the other side, your firm has placed itself in the unique position of being able to price-compare different law firms, and even different customers of the same firm. You could pivot/expand into a price-shopping service (kayak for law firms), a law-firm analytics provider (giving a/b testing results and other demographic info to law firms, leading to better price stratifying), or a service for law firms to gain pricing information about their competitors.

Co-founder here. Legal bills are not considered "work product" so the client can share them if they choose.

The content of legal bills is very important to protect. We have technical safeguards as well as management/process safeguards. The most important of which is limiting the number of people with access to actual content.

We have data and can price compare across geographies and law firms. But it is important to remember that each transaction is unique. That's why we don't release any of that information currently. If we find the right way and the right venue to share that aggregated information, we will do so if we think it is in the best interest of our customers.

Great concept, but not sure about the claim that legal bills aren't "work product" - it depends on what they say. And it's quite probable that the most effective legal bills for data analysis purposes (i.e. detailed ones) would contain work product or attorney-client privileged information. When I was a practicing lawyer, I always redacted bills before filing them in court to support fee petitions.

Separate issue whether disclosure of the bill to you as vendor, with a documented expectation of confidentiality, would constitute waiver either of WP or A/C privilege. No obvious reason that disclosure to your service would be different than disclosure to TyMetrix, Serengeti, or Sky Analytics, assuming your T&Cs are well-crafted. But obviously check with your lawyers on that.

Also, purely from a business standpoint, you can bet that some lawyers will raise the confidentiality objection to discourage/resist adoption.

All is to say, the issue is an important one. Glad to see you're taking it seriously. Good luck!

I think SimpleLegal is my favorite startup out of the batch; I'm an absolute sucker for ideas that boil down to "apply common sense and technology to verticals that are usually ignored by those things."

Really excited to see how things evolve for them.

Co-founder here. Thanks!

We just saw a ton of money being thrown at a problem that looked like it needed some help being solved. We're seeing growth in insurance, banking, patent heavy companies (hardware), and startups raising money. Even had a VC ask about becoming a customer!

Q: How do I get attention from VCs?

A: Make something that the VC, personally, himself, wants to use.

Love it.

If a client can't understand what he is being billed for by his lawyers, the problem won't be fixed by machine learning -- he needs to call his lawyer and ask what he is being billed for. Lawyers are generally required to explain their bills and itemize if requested by clients in nearly (but not quite all) 50 states.

Indeed, the larger problem faced by SL, and one that the legal industry as a whole is now facing, is the shift from per-hour/itemized billing to flat-fee for service arrangements which are becoming a significantly larger portion of the market across all fields (and which already dominate legal markets such as immigration, criminal, estate planning, and tax).

They need to move up-market as quickly as possible. If you're a Fortune 500, your legal bills can be extremely voluminous and be very painstaking to review. There are already products to help review them, but they're probably crap. At the same time, because of that volume and all those repeat plays, the machine learning algorithms probably have a lot more to work with. But I don't think $600/year for a business makes sense for such customers. If the service is really useful, and if they can scale it up to be useful to a big company, they need to charge a lot more.

Co-founder here. Great points about bigger businesses. I have worked for 10 years with insurance companies, captive finance firms, and other financial institutions with their legal data. Our pricing should be only slightly better than incumbents. For companies >$50k/yr in legal bills, we charge 1% of billed.

We're piloting now with a company with $1.1B/yr in legal bills. They're not so interested in the dashboard, but in access to our algorithms.

And yes, the existing products to review legal bills are crap (note: I'm waaaaay biased, obviously.)

If you're a Fortune 500 company, you have the legal staff to handle legal billing--or even the legal work itself directly.

Otherwise, I agree--going upmarket is their most likely pivot. They're also likely to switch to a %-saved fee model, which is what existing bill analysis companies already do, quite successfully. (SimpleLegal isn't the first company to tackle this problem, and it won't be the last.)

My advice to SL? Since they're clearly going the website-based growth route rather than the human interaction growth route, they need to work out partnerships with other legal tech providers, i.e., Clio and various legal app developers for iOS/Android.

Legal is a cost center at F500 companies; if you have a product that can make a several- percentage- points dent in an 8-9 figure cost center, you should market it accordingly.

Think of it this way: F500's also uniformly have IT departments staffed with highly-paid IT professionals. How good a job do you think most F500's do with IT cost management?

I know I'm arriving very late to this thread, but F500 companies overwhelmingly turn to outside counsel for their legal needs. There is inhouse legal staff that review the work done by outside counsel, approve bills, and negotiate retainer agreements, but by far the majority work is done by outside counsel.

At the enterprise level, legal spend analysis is a huge deal, and yes the existing tools for electronic billing are really clunky (I work in this space).

%-saved is really hard to accurately calculate on a legal bill. Often inhouse counsel will negotiate a "discount" with an outside firm (yay savings!), but then that firm will overbill to make up the difference, and may do so in ways that aren't really detectable even by "machine learning". For example, they can pad their hours or add extra staff to the matter, or add some mysterious disbursements, or hell just raise their base rates next year to make up the difference. Who is to say that "REVIEW AND DISCUSS LEASE CONDITIONS AND DISCUSS SAME WITH CLIENT" should have taken 2 hours instead of 4? (A real example I just pulled from a real LEDES file).

I did a report for a F500 company a few years ago that showed that after the switch from paper to electronic billing, savings increased (in the form of discounts and auto-adjustments made by our rules engine), but overall spend increased anyway (in the form of increased fees and hours billed). So I'm not really sure that this is a problem that can be fixed by technology. People are gaming the system on both sides of the transaction.

> If you're a Fortune 500 company, you have the legal staff to handle legal billing--or even the legal work itself directly.

I'd think that the general counsel of such a company is a prime target for this product. Carefully reviewing millions of dollars worth of charges billed six minutes at a time is extremely labor intensive. Having a product that does a first pass to find anomalies is a great way to cut down on the effort required.

Actually you are right but for the wrong reasons. The general counsel at a F500 company is not the one carefully reviewing the invoices, although his approval authority may be required for only the most significant bills.

But general counsel do want to keep overall legal spend under control, and are decision makers when it comes to purchasing several million dollars worth of legal software. For the purposes of closing enterprise sales, these are the guys you want to be taking out to fancy dinners.

Do you guys have a service for law firms so they can run their bill through your algorithm and get a list of bill entries clients are likely to flag? Could be useful for firms looking to maximize revenue and reduce client complaints.

Partners typically edit bills substantially before sending them out to clients, and I know a partner in my current job spends hours doing this every week for our practice group; I don't think he has much love for the process, so a way of automating it would be amazing, I'm sure.

Yes! The alternative is often some very busy partner spending hours of unbillable time every week making sure the bills match the client's required formatting, often because whatever crappy software the client currently uses gets confused when it sees commas instead of semicolons separating entries or whatever.

Billing partners hate reviewing bills. First year associates write some interesting narratives in their time entries.

We are working on the best way to provide this service. Most likely working through a partner. We like the idea of an "Approved by" label. But we want to avoid any conflict of interest.

Reading the homepage, I have no idea how it works other than it uses "machine learning" on the invoices you upload.

In regards to how it works...

From a customers point of view... we provide a system by which you can receive and review your legal bills in one place (instead of dealing with PDF's and paper invoices). Then you can approve them, mark them up, and pay them.

In the background we provide categorization of line items, insights into what they mean, if the they trigger any guideline flags / questionable charges. As far as how that happens technically... well, that's our secret sauce. Can't divulge it.

I dont know who the target is for this. But if it's big enterprises, then this might not be as appealing as it sounds. I used to work for Datacert, a legal apps vendor for large companies(walmart, jpmc, novartis etc). They use e-billing with a well defined format(ledes) all items on an invoice is code with a specific code(UTBMS). We used a rules engine(Drools) to ensure proper billing. A proper use for machine learning in legal space would be simulations for different pricing models(Alternate fee arrangements(AFA) in legal parlance). If machine learning could evaluate different pricing models for a given case and provide inputs, that would be a big selling point.

You'll get 80% of the benefit just by looking at word frequency, highlighting outliers and then a weight based on factors such as length and secret-sauce weighting.

Bonus points if you're using multiple categorizations (using different weights for different industries).

NLP / statistical stuff is fun ;)

Are you scanning / OCRing the documents? I never managed to get the OCR to be good enough for invoicing, there always had to be a manual process to fix the (machine-learning-flagged) errors.

Or don't you need accurate-to-the-cent invoices?

Word frequency is in use at many larger insurance companies today. You can certainly find problematic bills with word frequency and the hours billed, but you end up with a lot of false positives so you still have to manually review everything. We get in deeper than word frequency.

And, yes! NLP + statistics is fun!

How long till we can run a similar system over the invoices from IT consultants and programmers? ;-)

Implemented contact us form in PHP: 6 hours, $720 .. BZZZZZZT!!

> Although the business wasn’t sexy, it was profitable.

To these guys at TC it seems that sexy is the priority. Glad to see that for pg and co. sexy = profitable too. That's what business is about, please remind crunchie folks.

I don't follow the example given "the system flagged a line item where a professional billed a half hour for mailing." - that doesn't sound like an outlier to me, possibly the smallest billable time is half hour, or maybe the one page mailing had other complexities, like needed to be over-night delivery, required trip to the PO etc etc. I would need to see the system detect something more demonstrative of being useful.

The actual example was .5 hours to prepare a 1-page form and .5 hours to mail the form. It probably did require return receipt.

That's one expensive stamp.

The lawyer probably didn't intend to bill that much, but filing out a time card at the end of the week results in problems like this.

I'd love to see law firms publish total hours billed by lawyer. I think that would solve a lot of over-billing if they did.

I've personally felt this pain. Even as an engineer, I never thought to solve it myself. Massive kudos!

They should use a Summarization API: https://www.mashape.com/mojojolo/textteaser#!documentation

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | DMCA | Apply to YC | Contact