A puzzle hangs over legal tech: Why isn’t there more of it, particularly the sort that helps ordinary people?
Conditions have long seemed ripe for a legal tech renaissance. While there remains debate about where to assign blame for the justice gap, its magnitude is undeniable. A parade of “legal needs” surveys in recent decades shows that a hundred million or more Americans experience one or more civil justice problems each year yet receive no help in resolving them.Footnote 1 Take this decades-long crisis of immense, unmet civil legal need and add in rapid advances in technology, especially artificial intelligence (AI), and one would expect a robust “direct-to-consumer” (DTC) legal tech sector that provides affordable, tech-enabled solutions to fill the justice gap.
And yet, investment in DTC legal tech – just one measure of an industry’s health and potential – has languished compared to lawyer-driven software applications, such as Harvey or Casetext. Beyond LegalZoom and RocketLawyer, DTC legal tech is a footnote in a market dominated by law firms and corporate buyers.Footnote 2 And investment in DTC legal tech is a pittance compared to “fintech.” Think here of Robinhood, a commission-free stock trading tool that has received a hefty $6 billion in funding,Footnote 3 more than ten times even LegalZoom.Footnote 4 Becky Sandefur, the dean of access-to-justice scholars,Footnote 5 concluded in a 2019 report that legal tech tools serving “non-lawyers” are “restricted and limited” in functionality and “only partly match the types of justice problems most commonly reported by Americans.”Footnote 6 A more recent report from Duke University reached the same conclusion: “justice tech”Footnote 7 has struggled to overcome “system barriers.”Footnote 8 As one “justice tech” founder recounted, “I had a big dream and I knew it would be hard, but I had no idea.”Footnote 9
So why hasn’t a robust DTC legal tech market emerged? At least three explanations have so far dominated the debate. The first (and most common) blames restrictive rules that say that only lawyers can practice law (i.e., prohibitions on “unauthorized practice of law,” or UPL) or own law firms (i.e., Rule 5.4’s bar on fee-sharing).Footnote 10 The former limits functionality, consigning DTC legal tech tools to form-filling and “document assembly,” thus providing only limited help to those in need. The latter bar DTC providers that are not fully lawyer-owned – like LegalZoom – from employing lawyers to supplement their tech-based services. A second prominent theory is insufficiently potent technology: Automated tools, even sophisticated AI-based ones, cannot perform the all-important translational work – cutting through legalese and explaining options and outcomes in plain language – that is necessary when serving unsophisticated lay clients.Footnote 11 While not a problem for legal tech directed at lawyers, it’s a major hurdle for tools that aim to serve people who need actionable advice in clear, understandable, and bite-sized forms. A third explanation focuses on courts’ technology systems: Looking across some 14,000 court jurisdictions in the United States, one sees a checkerboard of technology and data infrastructures. The resulting technological Babel makes impossible the scale necessary to induce tech providers to invest time and capital in developing and then maintaining robust, user-friendly systems that serve customers with very limited ability to pay.
This chapter argues that undergirding each of these explanations is a deeper and potentially intractable problem. Though barriers such as restrictive rules, weak tech, or the court technology checkerboard are commonly invoked, they fail to account for the limited scale and longevity of most DTC legal tech. The better explanation for anemic legal tech – or at least a core component of a complete one – is the uniquely challenging market economics of providing low- and medium-income Americans with legal services. Facing already narrow margins, DTC legal tech is locked in an unwinnable race to attract and retain customers who only need their services once – and, worse, may not realize they need them at all. Lowering other barriers – say, by lifting the lawyer’s monopoly, or leveraging newly potent forms of generative AI – might give DTC legal tech room to breathe, but doing so cannot alone solve the underlying problem: To build sustainable models, DTC legal tech needs passive income streams and enterprise-level customers.
That conclusion, withering as it is, is crucial to thinking about the future of legal services and access to justice. For instance, it is possible that “regulatory reform” advocates have it wrong, at least as to technology-based legal services. If liberalized legal services rules are unlikely to yield a flowering of DTC legal tech anytime soon, then perhaps the access-to-justice movement should be investing its limited capital, both political and real, in efforts to build out human-centered models of legal services delivery, such as the community justice worker movement.Footnote 12 By extension, perhaps legal tech entrepreneurs should focus on devising tools that extend the reach of the human providers, beleaguered and overwhelmed, that already dot the civil justice landscape. Even more bracing, perhaps legal tech developers should be creating tools that serve courts, not self-represented litigants.
As we note below, courts in the coming years will face a “make or buy” choice. “Make” means court-connected, court-hosted systems, including court portals, remote hearings, online dispute resolution (ODR), digital self-help, or e-filing, whereas “buy” means courts structuring themselves, particularly their technology and data systems, to be more accessible to intermediaries, especially DTC legal tech providers, who serve self-represented litigants. Make and buy require very different infrastructures, and it’s not at all obvious that courts, perennially underfunded, have the money to do even one, let alone both. But if DTC legal tech is doomed, and budget-constrained courts must choose, then “make” – taxpayer-financed, “public option” legal tech – may be the wiser investment.
We develop these ideas using a mix of theory and evidence. The first section below reviews the “usual suspects” – three common explanations for anemic DTC legal tech. The next section sets forth our market- and business-centered theory, thus building a fuller account of the challenges facing DTC legal tech. A third section reports our research, including interviews, with six DTC legal tech providers, chosen to reflect a cross-section of the current market landscape, and then maps them to the full set of “barrier” theories. A final section considers the implications of a DTC legal tech sector that may never get fully off the ground.
A few orienting notes on scope: First, our inquiry focuses on DTC legal tech providers: client-facing providers that use technology to deliver legal services. We thus bracket legal tech tools that instead automate law practice via a “technology assist”Footnote 13 to lawyers and legal workers.Footnote 14 Moreover, we focus on legal tech providers that offer actual legal services, defined broadly as “help with a justiciable issue.”Footnote 15 As such, we exclude story-telling tools, reminder tools, and digital signature tools which, while client-facing, do not provide a legal service; we also exclude “legal marketplaces,” which merely connect consumers to an (often traditional) legal services provider.Footnote 16 Finally, we include DTC legal tech providers whose end users are individuals and/or small businesses, but not lawyers or large corporations. In so doing, we employ a broader definition of “consumer” than previous studies.Footnote 17 By defining consumers to include small businesses in addition to individuals, we focus on tools within the so-called PeopleLaw segment of the legal services industry, as distinct from corporate-facing BigLaw or legal aid.Footnote 18 In serving ordinary individuals and not organizational clients, PeopleLaw sits at the cusp of the justice gap: individuals and businesses who do not qualify for legal aid but who also cannot afford traditional legal services.
8.1 The Usual Suspects: “Barrier” Theories
Three explanations have dominated the debate about anemic “justice tech.” Each, however, raises hard questions about where the real problem lies. This section reviews each explanation to set the stage for our own market- and business-focused account of why DTC legal tech has languished.
8.1.1 Restrictive Rules and Regulatory Reform
The most common explanation for anemic legal tech is regulatory constraints on law practice. In the US context, two types of restrictions are most salient. The first is the prohibition on UPL. The second is Rule 5.4’s bar on fee-sharing, which limits investment in law firms to lawyers and prevents investors who are not lawyers and organizations that are not law firms to own, invest in, and profit from the sale of legal services. Notably, numerous states have relaxed, or are considering relaxing, these regulations.Footnote 19 But as it stands, these rules continue to constrain the services that legal tech providers can offer.Footnote 20
However, whether liberalizing reforms can unleash new entity- and technology-based modes of legal services delivery remains unproven. In the United Kingdom, where UPL was already more relaxed, loosening fee-sharing rules to permit nonlawyer investment in legal services providers has had an underwhelming impact on innovation. While the majority of such providers in the United Kingdom serve individual consumers or small businesses,Footnote 21 a comparative study of nonlawyer ownership in Canada, Australia, England, and Wales concluded that outside investment’s impacts on access to justice remain limited outside of certain practice areas, such as personal injury.Footnote 22 Moreover, it’s unclear whether the benefits of these economies of scale are consistently passed on to consumers. While greater nonlawyer investment generally has promoted technological innovation and the development of new legal services,Footnote 23 evidence is mixed on whether reforms have moved the needle on price.Footnote 24
8.1.2 Weak AI
A second common explanation for anemic DTC legal tech is that AI is simply not potent enough to productize PeopleLaw. One theory is that, unlike legal tech for BigLaw, legal tech for PeopleLaw must interface directly with individuals and small business owners who require lay translation of legal concepts,Footnote 25 something natural language processing (NLP) technology, powerful as it is, has struggled to do.Footnote 26
The “weak AI” explanation for languishing legal tech is an important one, but it is hardly decisive. After an “AI winter” of relative stasis, AI is once more undergoing staggering leaps in potency with the arrival of powerful “large language models,” or LLMs – a branch of “generative AI” that is capable of producing clear, plain language text. Champions see a two-fold promise. First, new AI systems are plainly better than earlier generations of NLP at translating plain language into legalese and vice versa.Footnote 27 As already noted, lay-to-legal translation is essential if you want to build usable tools. Second, increased translational capacities might mean that new generative AI systems can take lay narratives – an individual’s plain language explanation of a problem – and then map them to legal ontologies, pathways, and even likely outcomes.Footnote 28 For the first time, AI might be potent enough to plausibly offer actionable advice to self-represented litigants on options and outcomes.
Current prognostications, however, may miss some important points. For starters, generative AI may not move the needle as much as some think. Much of existing DTC legal tech does not rely on cutting-edge technologies: Document assembly, for instance, is relatively low-tech, using simple automation engines to help self-represented litigants answer complaints and avoid default judgments.Footnote 29 The value-add of sophisticated generative AI systems may not be as robust as some assume.
Moreover, to be both reliable and cost-effective, AI must adapt to an ever-shifting constellation of substantive and procedural rules across jurisdictions. Direct-to-consumer legal tech providers have to this point relied solely on human lawyers to manually update their tools as laws change or new rules and regulations are issued. Here, too, questions remain about how soon, or even whether, newly potent forms of AI can handle subtly and ever-changing laws and regulations without human intervention.
Time will tell whether continuing AI advances can make meaningful progress on these problems. But, for the moment, weak AI does not offer anything approaching a fully satisfactory theory for the current state of DTC legal tech. For some applications, current tech capabilities seem more than ample. For others, continued AI advances may hold the key.
8.1.3 The Court Technology Checkerboard
A third common explanation for anemic DTC legal tech extends from American federalism. In a nutshell, most PeopleLaw clients have a limited ability to pay. So for providers, scale is king. However, in addition to the variations in substantive and procedural law just noted, a checkerboard of technology systems, particularly e-filing systems, and data infrastructures across thousands of local court jurisdictions make scaling services nearly impossible.
More specifically, DTC legal tech is limited by two types of jurisdictional complexity.Footnote 30 First, patchwork systems create jurisdictional differences in what forms litigants need to e-file, what requirements they need to meet, and what system – both front-end and back-end – litigants e-file through. Filing requirements are a key pain point that increases the complexity of delivering clients filing-ready documents; DTC legal tech often requires full-time lawyers to manually track and update requirements across the jurisdictions they serve.Footnote 31 Second, even if technologists find a way to track and code around jurisdictional differences, self-represented litigants cannot e-file at all in many statesFootnote 32 and, even where they can, they can only do so in limited case types. Filing fees – and e-filing can often introduce additional fees – pose an additional burden; there are no consistent digital payment systems, so some jurisdictions require litigants to pay in person.Footnote 33
As with the other commonly invoked barriers to robust DTC legal tech, the state and local checkerboard of technology systems and data infrastructures might improve, and efforts are underway to develop common, jurisdiction-spanning data standards.Footnote 34 Unless automated approaches can replace brute-force monitoring and updating, the checkerboard problem – complex, incongruous, and confusing requirements that vary state to state and even courthouse to courthouse – will remain a threat to low-cost delivery models. At the same time, the technology checkerboard plainly has less to say about federal-level innovation, not to mention large states with relatively unified court systems. As with the other explanations, the checkerboard is, at best, a partial explanation for anemic legal tech.
* * *
In short, even if rules were liberalized, AI continued to vault ahead, and the checkerboard made more uniform, DTC legal tech still faces substantial barriers. The next section begins to sketch them.
8.2 A New Take: Market Barriers to Robust DTC Legal Tech
What else might explain anemic legal tech? This section develops an alternative, market-based theory focused on legal tech providers as businesses – and, very often, startups – and the shape of the markets they seek to serve. To flesh out its particulars, it draws on theory and industry analogues to propose a three-step, market-based explanation: Legal tech lacks a sustainable customer base because (i) limited customer demand and (ii) high customer-acquisition costs produce (iii) an acute “lifetime value trap.” The rest of this section elaborates on each of these challenges.
PeopleLaw – the market that DTC legal tech is best positioned to serve – has been flooded with lawyers for decades,Footnote 35 yet it remains notorious for its lack of profitability.Footnote 36 The vast majority of PeopleLaw is bespoke and labor-intensive, and there isn’t much room to lower prices when the average solo practitioner already earns less than $60,000 (and may be facing significant law school debt).Footnote 37 And while legal tech has often served those priced out of traditional PeopleLaw, that market likely cannot sustain the scale necessary for a tech startup to grow, particularly where legal tech providers typically face stiff competition to attract individual customers.Footnote 38
With the ever-increasing cost of digital marketing, including the rising ticket price of search engine optimization (SEO), legal tech must dramatically increase the dollar value of each client to break even, let alone turn a profit. While this problem may be addressed by building a steady base of loyal, returning customers, demand in PeopleLaw is hardly consistent: Customers are less likely to need another divorce than they are a replacement pair of their favorite brand-named sneakers. To solve this problem, other consumer services segments – health tech and platform tech – have redefined the customer: Instead of individual users, they serve enterprises that come with big-ticket contracts and reliable revenue. But there is no clear enterprise customer, at least not yet, for legal tech to tap into.
8.2.1 Limited Customer Demand
In general, legal markets are hindered by informational asymmetries that artificially curb demand: People do not understand their true legal needs, and do not seek appropriate representation. One way to combat low demand is to target services to specific markets: Think here of personal injury lawyers’ billboards, which make the nature of their services crystal clear. But when providers tailor their offerings to a legal services niche particularized enough to passively attract customers – such as sole business proprietors or divorcees – that niche often has too few clients to justify the expense of serving it.
8.2.2 High Customer-Acquisition Costs
Even if a legal tech provider can crack the demand problem and generate interest in their services within their target demographic, founders note that “it still [takes] a lot of work to convert that interest into a paying customer.”Footnote 39 This is because customer-acquisition costs (CAC) are incredibly high. Traditional, low-tech methods don’t seem to work in the legal tech space: Finding consumers by word of mouth can be “challenging” and take “more time than they originally planned.”Footnote 40 In general, overcrowding in DTC markets has driven up the price of digital advertising,Footnote 41 eliminated any arbitrage, and made the customer-acquisition math much more challenging.Footnote 42 For legal tech, even in markets where users understand their specific legal need, intense competition between providers can render legal services financially inaccessible. Take immigration, for example: Provider overcrowding has made SEO on immigration-related terms so expensive that many for-profit providers charge fees in the thousands to recoup advertising costs.Footnote 43
8.2.3 The Lifetime Value Trap
A DTC brand’s success hinges on how much headroom there is between customer lifetime value (LTV) and CAC.Footnote 44 Now that CAC has skyrocketed, it is “widely accepted” that firms should direct more effort into retaining existing customers than attracting new ones.Footnote 45 Yet there are only two ways to boost LTV: repeat customers, or customer cross-sell. DTC legal tech is set up for failure on both.
Where consumer demand is steady – for example, in consumer goods – the key to repeat customers is loyalty: getting your customer to stick with you over your competitors. Customer-centric, interaction-based marketing models are the current gold standard for increasing customer loyalty and, thus, LTV.Footnote 46 Consistent touchpoints allow companies to gather nuanced customer data that can predict and even generate demand. Allbirds, a DTC shoe brand, can use the data of existing customers to estimate how often a particular customer will need to replace their shoes, and to build customer profiles that allow it to maximize each dollar spent on CAC. The same relatively straightforward calculation is possible for razors, laundry detergent, and all manner of consumer goods that are used up and regularly repurchased. And in consumer goods, cross-sell opportunities are often obvious: A company can easily expand into selling shaving cream alongside razors.
A common strategy among DTC companies is to build customer loyalty by expanding beyond their initial product offerings to create a complementary ecosystem of products and services. Peloton, for instance, initially offered big-ticket exercise bikes but quickly expanded to offer digital exercise class subscriptions, exercise apparel, and wellness programming.Footnote 47 It thus transformed from a bike retailer to a holistic exercise company with the ability to collect and use customer data to drive growth.
Direct-to-consumer companies that provide services, not goods, also leverage consumer data to drive LTV.Footnote 48 By interfacing directly with consumers, DTC brands can generate huge troves of data about usage preferences. TurboTax, for example, sells customer data to generate advertisementsFootnote 49 and targets users with “offers”Footnote 50 – that is, upsells them.Footnote 51 And perhaps most important of all, the relentlessness of annual taxes means that TurboTax enjoys repeat customers – and, even better, customers with increasingly complex needs over a lifetime – with little investment required on their part.
Where building loyalty among individual customers is impracticable, DTC service providers often turn to enterprise-level customers: companies, not individuals, who purchase big-ticket and renewable contracts. Subscriptions sold to enterprises, not individuals, are also generally more profitable: While Amazon’s marketplace generates more revenue, its cloud solutions – sold only to enterprises – account for the lion’s share of Amazon’s profits.Footnote 52 Similarly, many health tech players have made insurers and employers their core customers.Footnote 53 More complex healthcare services use a B2B subscription model: Livongo, a diabetes management platform, is free to the individual user but paid for by employers, health plans, or health providers.Footnote 54
Most importantly, enterprise-level contracts serve as vital cash infusions that give startups with limited equity – like most legal tech – the capital they need to scale and grow.Footnote 55 Biotechnology startups often get early “revenues” from contracts with large pharmaceutical companies, providing both cash flow and visibility into on-hand capital for longer-term investments.Footnote 56 This ability to forecast is particularly important in maturing startups looking to scale.Footnote 57 Major cash infusions are especially vital for legal services startups that, because of Rule 5.4, cannot raise outside investment using asset-backed guarantees and do not have outside equity to pad their cash flows.
Structural forces prevent legal tech from implementing each of these traditional strategies. First and foremost, legal tech cannot rely on predictable demand to increase LTV. Unlike shoes, people do not always need legal services. And the more specialized the legal service is, the rarer the potential repeat customers or cross-sell opportunities become. This creates a tension between increasing specialization of legal tech – the very unbundling of legal services that some view as a necessary precursor to lower-cost legal servicesFootnote 58 – and a DTC legal tech provider’s ability to generate sustainable LTV.
Second, traditional enterprise customers do not exist for legal services, likely due in part to the lack of legal insurance. In the United States, there is little legal insurance outside of medical malpractice for doctors.Footnote 59 Employers, reliable enterprise customers, are unlikely to offer legal insurance on a large scale: Unlike healthcare, where both employer and employee have a vested interest in the employee’s health, employees often have legal needs that are actually adverse to their employers – for instance, issues around benefits or even claims against the employer itself.
Third, enterprises unique to legal services are poorly positioned to support and scale DTC legal tech. BigLaw seems the obvious choice: Because firms can invest in DTC legal tech without running afoul of UPL or Rule 5.4, they can give small startups the credibility needed to win customer trust. But the economics of BigLaw are not designed to withstand the major up-front investment required to stand up a legal tech startup. A new DTC legal tech provider must spend at least six months to a year fine-tuning SEO to hit the right searches and convert leads to revenueFootnote 60 – an expensive process even without the years required to create the document assembly tool in the first place.Footnote 61 Cash-strapped courts could, in theory, be a promising enterprise customer for established providers, supplying them with a steady stream of litigant-clients, but they face high hurdles in supporting startups.
Finally, the nature of legal services creates barriers to legal tech companies building holistic product suites for which customers have ongoing needs. For instance, Hello Divorce – a DTC legal tech company described in more detail below – followed the “Peleton-model” and introduced nonlegal services that extend the lifetime of its divorce clients. But the introduction of those services required careful, resource-intensive navigation around strict UPL and Rule 5.4 prohibitions. Even then, the lack of perpetual demand for legal services creates issues. While Peloton customers will likely only buy one bike – just as Hello Divorce customers will likely only get one divorce – exercise and wellness are ongoing needs; divorce counseling is not. Direct-to-consumer legal tech must not only stave off competition; it must also contend with a more existential question: If I solve my client’s legal problem, will they ever need me again?
8.3 The Barriers in Operation: Case Studies in Legal Tech
This section puts data on what, to this point, has been mostly open-ended theorizing about the workings of the legal tech marketplace. We do so by running the various theories – both the “usual suspects” and our new, market-based way of thinking – through a select set of legal tech providers, as summarized in Table 8.1.Footnote 62 These legal tech offerings represent a rough cross-section of the nascent DTC industry in terms of each entity’s services type (from online software wizards that generate ready-to-file legal documents to legal advice from humans), the subject-matter area of those services, the entity’s organizational structure, its fee structure (including its for-profit or nonprofit status and its sources of revenue), and its geographic scope. Taken together, these companies offer an ideal workbench to explore how the different possible explanations for the industry’s anemic status operate and intersect across different segments of the market. They also make plain our central thesis, as summarized in Figure 8.1. While conventional explanations for anemic DTC legal tech – restrictive lawyer rules, weak AI, and the jurisdiction-by-jurisdiction technology checkerboard – might explain languishing legal tech as to some subset of DTC providers, only our market-based theory can fully explain lagging legal tech across the board.
Table 8.1Long description
The table summarizes the work of 6 case study subjects in column 1, followed by the scope, structure, 3 legal services namely, document assembly, legal advice, and attorney-client matching, and 4 fee structures namely, one-time customer transaction, ongoing customer subscription, attorney referral fees, and grants or individual donations, from left to right in order.
The work summary for the 6 case study subjects are as follows.
Hello Divorce. Generates and files divorce documents for clients and connects clients with lawyers at its sister firm, Levine Law.
Atticus. A legal marketplace that performs client intake and connects users with vetted outside lawyers, focused on social security and disability.
Tenant Power Toolkit. Generates and files documents for tenants facing eviction in California with a focus on the L A courts.
Legal Zoom offers document production, filing, and legal advice for small business owners and solo proprietors.
Upsolve generates but does not file Chapter 7 bankruptcy documents.
SixFifty is a SaaS company that offers employers platform-level employment and privacy solutions with one-off document assembly available on their website.
The data for the 3 legal services and 4 fee structures are represented in tick marks signifying that a particular column is checked for a legal tech provider. The following are the data for the scope, structure, the 3 legal services, and the 4 fee structures in order, from left to right for the 6 legal tech providers.
Hello Divorce, multi state, dual entity, all columns checked except for one-time customer transaction fees and grand or individual donation.
Atticus, national, dual entity, only attorney-client matching and attorney referral fees are checked.
Tenant Power Toolkit, C A, stand-alone, only document-assembly and grant or individual donation are checked.
Legal Zoom, national, stand-alone, all except attorney-client matching and grand or individual donations are checked.
Upsolve, national, stand-alone, only documents assembly, attorney referral fees, and grants or individual donation are checked.
SixFifty, national, wholly owned subsidiary of Wilson Sonsini, all except attorney referral fees and grant or individual donation are checked.
Figure 8.1 Impact: degree to which a barrier restricts scale.
Figure 8.1Long description
The table compares the degree to which 4 different barriers namely,Restrictive Rules, Weak A I, The Checkerboard Problem, and the Lifetime Value Trap, impact 6 legal tech companies’ ability to scale. There are 3 impact degrees. A black circle indicates that the barrier has a significant and negative effect on the provider such that lifting the barrier would enable the provider to scale, grow, or otherwise offer better services to their clients; a gray circle indicates that the barrier limits scalability or profitability in minor ways; a white circle indicates that the barrier has no or minimal impact.
For Hello Divorce, restrictive Rules, The Checkerboard Problem, and the Lifetime Value Trap, are black circles, and Weak A I is a white circle.
For Atticus, Weak A I and the Checkerboard Problem are white circles; Restrictive Rules is a grey circle, and the lifetime value trap is a black circle
For the Tenant Power Toolkit, Restrictive Rules is a white circle, weak A I and the Lifetime Value Trap are grey circles, and the checkerboard problem is a black circle.
For Legal Zoom, restrictive rules and the lifetime value trap are black circles, Weak A I is a white circle, the checkerboard problem is a grey circle.
For Upsolve, Weak A I, the Checkerboard Problem, and the Lifetime Trap are grey circles, and restrictive rules is a white circle.
For SixFifty, the Lifetime Value Trap is a black circle, and Restrictive Rules, Weak A I, and the Checkerboard Problem are white circles.
8.3.1 Restrictive Rules: Why Hello Divorce Struggles When Atticus Does Not
Restrictive rules have shaped what services American legal tech can offer and how their businesses are structured. But just how much explanatory power do they carry, and how, exactly, do rule restrictions limit available business models when trained on particular providers?
Key to answering those questions is to see the combined effects of UPL and Rule 5.4, not one or the other operating alone. For instance, LegalZoom’s core business focuses on document assembly, which is not affected by UPL rules or changes to them. Rule 5.4’s constraint on nonlawyer ownership, however, adds a further layer of complexity by limiting provision of legal advice even by full-fledged lawyers. In the United States, LegalZoom’s nonlawyer ownership means it cannot hire attorneys to provide direct legal advice to its customers – a major disadvantage when competing against traditional law firms. But in the UK and Arizona, which permit nonlawyer ownership in some circumstances, LegalZoom employs licensed attorneys to provide permitted legal services. And it makes a mint in doing so – those legal services are a major profit driver, increasing transaction volume and allowing LegalZoom to charge a premium over its tech-only services.Footnote 63
But workarounds of Rule 5.4 in the United States are few. New ventures must either forego outside funding or adopt convoluted dual-entity structures to allow the tech side to receive funding while the law firm itself does not. Hello Divorce, for instance, operates both Hello Divorce, a technology company, and the Levine Family Law Group, a full-service law firm, as separate entities through a complicated “dual-entity” model in order to skirt both UPL and Rule 5.4 restrictions. And doing so is costly.
To see why, consider first that Hello Divorce provides digital DIY divorce products across document assembly, e-filing, credit repair, and other financial services.Footnote 64 Its online platform also offers access to expert help from in-house mediators and lawyers at Levine Law.Footnote 65 Users pay a monthly subscription fee of $99 to access the DIY web-based tool and additional fees for financial coaching or legal services.Footnote 66 The Levine Family Law Group, a long-standing family law practice, provides those legal services,Footnote 67 which are used by approximately 50 percent of Hello Divorce’s clients.Footnote 68 Those clients sign a fee agreement with Levine Law, not Hello Divorce.Footnote 69 But in order to provide legal services through the Hello Divorce app – which hosts all scheduling, correspondence, and documents – Levine Law pays Hello Divorce a monthly administrative fee of $4,750.Footnote 70
These crisscrossing contracts are a direct product of restrictive regulation. Lest it violate UPL, Hello Divorce the tech company cannot hire lawyers to provide legal services; its technology cannot provide legal services, nor can its paralegals offer legal advice.Footnote 71 But under Rule 5.4, Levine Law cannot split legal fees with Hello Divorce. And since Rule 5.4 bans nonlawyer investment, Hello Divorce cannot collect legal fees directly from customers or own an in-house law firm. Nor can Levine Law support Hello Divorce as a wholly owned subsidiary: It is unlikely any law firm outside of resource-rich BigLaw can afford to shoulder the costs of scaling a startup, especially since Rule 5.4 cuts off access to loans and outside capital.
This convoluted structure, while end-running UPL and Rule 5.4, brings substantial operational complexity, making the business more expensive, harder to run, and less attractive to investors. For starters, Hello Divorce cannot enjoy basic economies of scale such as shared software subscriptions, malpractice insurance, workers comp, and cybersecurity.Footnote 72
The dual-entity workaround also restricts the flow of capital between Levine Law and Hello Divorce. Venture capital and private equity can only be directed to Hello Divorce,Footnote 73 while legal services revenue must flow into Levine Law. Thus, venture funding cannot be used to pay attorney salaries – Levine Law’s biggest cost bucket.Footnote 74 The dual-entity structure’s mismatched cash flows also split value between the two companies, making Hello Divorce appear less profitable – and less attractive to investors – on paper.Footnote 75
In stark contrast to Hello Divorce, AtticusFootnote 76 – an “attorney-client matching” service – has set up a dual-entity model that works simply because there is no UPL risk to mitigate. Atticus does not provide direct legal services, or anything approaching legal services, and no money exchanges hands between Atticus and individual users. Atticus’ “front-facing” entity, a law firm licensed in California, acquires clients, screens them, and matches them 1:1 with a vetted attorney in the Atticus network.Footnote 77 Those network providers then pay Atticus a referral fee in compliance with Rule 1.5e and applicable state-level regulations. While Hello Divorce’s “front facing” entity is its technology platform,Footnote 78 Atticus’ “tech” entity serves as its back-office: a venture capital-backed Delaware C corporation that owns the tech platform and manages spending across both entities.Footnote 79 Thus, Atticus is subject to far less regulatory scrutiny,Footnote 80 especially because its services do not compete directly with attorneys.Footnote 81 Atticus actually makes the existing legal monopoly work better by solving a mismatch between demand and supply. Rather than competing over the same client pool, Atticus helps existing PeopleLaw attorneys scale their services by finding them clients – including clients who may not otherwise pursue legal help at all.
In short, restrictive rules have plenty of explanatory value when it comes to key segments of the legal tech industry. Just ask Levine Law and Hello Divorce. But, as Atticus’ success shows, they hardly explain anemic legal tech across the board.
8.3.2 Weak AI: The Upside of Generative AI for the Tenant Power Toolkit
Consider another “usual suspect:” that underpowered AI has prevented DTC legal tech from building the tools necessary to scale PeopleLaw solutions. But even “full-powered” AI is not positioned to address DTC legal tech’s – and particularly, DTC document assembly tools’, most pressing issues. As noted previously, much document assembly is technologically simple: Providers plug complicated legal expertise into a technologically basic software wizard. Instead, the most labor- and capital-intensive phase is hiring lawyers to (1) create a plain language logic tree and (2) track ever-changing procedural and substantive requirements.Footnote 82
While these pain points constrain scale, generative AI is hardly a panacea. AI may never be capable of replacing lawyer translations, especially in more complex areas of the law. Plain language translations require precision and difficult judgment calls: Translating law into plain language is uniquely challenging because it carries inherent risks of over- or under-specification, creating a trade-off between making legal language understandable and retaining its original meaning.Footnote 83 Given this complexity, it may be some time before NLP can take the reins. And as discussed in Section 8.3.3, so long as technology cannot itself brute-force through the court technology checkerboard, scaling across jurisdiction remains prohibitively costly. In short, while some blame “weak” AI for “weak” DTC legal tech, “stronger” AI is unlikely to have the capabilities necessary to meaningfully jumpstart more effective DTC document assembly tools.
Still, as highlighted by the Tenant Power Toolkit, some providers may nonetheless benefit from next-generation AI. Currently, the toolkit asks a series of plain language questions to gauge eligibility and generate legal documents – such as fee waivers, proof of service, or even answers – for California tenants who have been served eviction papers.Footnote 84 While the toolkit can e-serve and e-file documents via One Legal, it can do so only in California.Footnote 85 Its scale and scope is limited by the pace at which human lawyers can design new plain language logic trees or translate existing ones, and the number of jurisdictions it can actually file in. Could generative AI enable the toolkit to reach more users by translating and simplifying its existing services?
Perhaps. First, generative AI can help providers like the Tenant Power Toolkit serve non-English speaking clients by instantly translating existing document assembly tools into languages beyond English. Often, the people Tenant Power Toolkit seeks to serve require a tool in their native language.Footnote 86 This time-consuming translation work is currently done by humans. As AI improves, even if it cannot create the logic trees themselves, it may offer a reliable and lower-cost solution to extending the reach of human-created tools.Footnote 87
Second, generative AI can make existing tools more accessible by reducing the number of questions users must answer. Branching logic trees are time-consuming. The Tenant Power Toolkit finds that users often do not complete the process, having underestimated how long and complex the tool is.Footnote 88 Upsolve, another tool catered toward low-income folks, has cited similar issues.Footnote 89 Here too, generative AI may present a novel solution: Users could type in a plain language problem that AI would then map onto legal questions and outcomes with no need for elaborate Q&A-like branching.
Even so, while AI could simplify logic trees – asking fewer questions and instead using technology trained on lawyer-made datasets to sift through the information the client provides – for most DTC legal tech providers, this is a nice-to-have and not a need-to-have: Document assembly tools have worked effectively as they are.
8.3.3 The Court Technology Checkerboard: A Problem LegalZoom Monetizes and Upsolve Avoids
Turning to the last of the conventional explanations, to what extent does the state- and local-level checkerboard of laws and court technology systems affect Table 8.1’s menu of legal tech providers? For many, the combined complexity of state laws and filing requirements makes expanding to new geographies expensive and labor-intensive. Return, briefly, to Hello Divorce. Each day, several of its clients’ forms are rejected not because they are incorrect but because the filing requirements have changed since the online tool prepared their documents.Footnote 90 Hello Divorce must employ a full-time lawyer to track and update filing requirements down to the county level, as courthouses often change their requirements with little or no notice.Footnote 91 Because the cost of filing alone makes serving additional states highly expensive,Footnote 92 the court technology checkerboard directly limits geographic scalability: Solutions built for one jurisdiction often cannot solve the same problem for clients in a different location.Footnote 93
Two strategies have been successful in overcoming the checkerboard problem: (1) finding an area where the law is uniform, but filing is not, as LegalZoom has or (2) limiting services to document preparation only, like Upsolve.
LegalZoom operates across 50 states and more than 3,000 counties in the United States; its core business is generating and then filing documents for small businesses.Footnote 94 LegalZoom has achieved such scale by monetizing the checkerboard problem. While LLC formation implicates different laws and regulations at the local, state, and even federal levels,Footnote 95 the basic steps for setting up an LLC are consistent across all states.Footnote 96 By choosing a relatively uniform area of the law, LegalZoom was instead able to focus its resources on tracking and coding the different filing requirements of each jurisdiction.
Perhaps counter-intuitively, because its solution actually solves filing complexity, simplification of state-by-state systems poses a major risk: LegalZoom’s investor materials highlight that “if U.S. state agencies increase their offerings for free and easy-to-use business formation services such as … filing portals to the public, it could have a significant adverse effect on our business.”Footnote 97
But for Upsolve – which, as a reminder, assembles but does not file chapter 7 bankruptcy forms – such filing portals would make its services more accessible and effective. An estimated 40 percent of Upsolve users drop off at filing because courts are inconsistent in their treatment of self-represented litigants.Footnote 98 Generally, e-filing remains too complex for Upsolve’s online tool to integrate, especially as requirements continue to change.Footnote 99 Despite expanding its scope in other ways – offering information and referrals to help customers with other types of bankruptcy, piloting consumer debt relief advocates in New York City, and building a new online tool for immigration – Upsolve’s technology-enabled services remain narrow: They assemble, but do not file, legal forms.
Here again, the checkerboard argument is complicated. Refraction through a more market-based model suggests that inter-jurisdictional variation is at once a threat and an important source of value.
8.3.4 The Lifetime Value Trap: A Challenge for All For-Profit Providers
Turn, now, to our market-based theory of anemic legal tech, which proceeds from the premise that low demand and high advertising costs have made customer acquisition expensive, so DTC legal tech providers must find a way to make each customer, once acquired, more profitable. As reflected in Figure 8.1, for-profit providers universally struggle to escape the LTV trap. Only for nonprofit providers – Upsolve and the Tenant Power Toolkit – do the economics of customer acquisition and retention have only minor impacts.
8.3.4.1 Customer Acquisition
At first blush, one obvious solution to the LTV trap is for legal tech to partner with law firms, allowing them to tap into an established client base. Our case studies, however, demonstrate that referral rates are low. SixFifty is a wholly owned subsidiary of Silicon Valley law firm Wilson Sonsini that provides platform solutions that customers, usually small businesses, use as part of their daily operations.Footnote 100 These solutions focus on streamlining common document assembly processes, not just fulfilling one-off document needs, and are marketed as legal automation software.Footnote 101 One of the very few legal tech players with direct ties to a BigLaw law firm, SixFifty nonetheless maintains a careful distance from Wilson Sonsini, using standard UPL disclaimers despite relaxed regulations in its home state of Utah.
While SixFifty and Wilson Sonsini do cross-refer customers – SixFifty even offers Wilson Sonsini clients a “friends and family” discount – those referrals almost never result in new business for either entity.Footnote 102 Poor referral rates are largely attributable to low customer overlap: Customers go to SixFifty and Wilson Sonsini for entirely different needs.Footnote 103 We see similar mismatch between customers of Hello Divorce and Levine Law.Footnote 104 Law firms make “chandeliers” – they handle multimillion-dollar deals and disputes – while document assembly tools standardize “lightbulb” solutions for simpler problems.
Instead of relying on referrals, most DTC legal tech providers turn to digital advertising to attract new clients. For some, it’s an easier task. People in need of a divorce, for instance, know they have a legal need and often make predictable searches, making it easy for Hello Divorce to target potential clients.Footnote 105 In other areas – particularly in large and diffuse markets – SEO is expensive and imprecise. LegalZoom, for example, generally targets small business owners; sales and marketing is their largest operating expense, amounting to nearly one-third of total revenue.Footnote 106 But even if a company has resources to spend on SEO, it may not be an effective distribution channel. In immigration, for instance, the high volume of both general and specific searches (e.g., “immigration help” or Violence Against Women Act (VAWA) application”) and providers’ need to target outreach based on their sub-specialty (e.g., Deferred Action for Childhood Arrivals (DACA) work permits) makes the SEO strategy impracticable.Footnote 107
8.3.4.2 Customer Retention
To increase LTV, many providers prioritize cross-sell. As previously noted, because no one plans to get divorced more than once, Hello Divorce has shifted its focus to more holistic lawyer-alternatives such as financial counseling and marriage mediators to extend the lifetime of each client.Footnote 108 LegalZoom also focuses on driving additional purchases and cross-sell for its customers,Footnote 109 including by upselling its customers to higher margin solutions.Footnote 110 For example, LegalZoom plans to expand credentialled-professional-assisted offeringsFootnote 111 similar to those offered by HelloDivorce.
Another way to boost customer retention is through subscriptions.Footnote 112 Despite being a key component of LegalZoom’s strategy to enhance LTV – in fact, its “business depends substantially on [] subscribers renewing their subscriptions”Footnote 113 – growth in one-time transactions continues to outpace growth in subscriptions.Footnote 114 And an all-in subscription strategy comes with a major pitfall: Deferred revenue as a result of subscription units can make cash flows dependent on the size, timing, and terms of subscription agreements.Footnote 115
SixFifty, for its part, has successfully combined cross-sell and subscriptions by building a suite of complementary products that allows it to naturally sell existing customers additional products (e.g., selling an employee handbook customer a subscription to its severance paperwork tool). This is partially attributable to the substantive legal areas it serves: Every business enterprise that interfaces with consumers must remain compliant with ever-changing privacy and employment regulations. Bundling multiple types of compliance under one vendor simplifies the process for small HR and legal departments; for example, they only need to be trained on and use one platform solution. SixFifty also occupies a niche where legal need is ongoing: Year after year, employers will need to hire, onboard, and let go of employees.
8.3.4.3 Escaping the Lifetime Value Trap
The LTV-to-CAC ratio, which compares the value of a customer over their lifetime to the cost of acquiring them, is the primary measure of DTC profitability.Footnote 116 For ecommerce, a ratio greater than 3:1 is considered “good.”Footnote 117 In SaaS,Footnote 118 industry benchmarks place the average LTV-to-CAC ratio at 3:1 for business services and 6:1 for financial services.Footnote 119 DTC legal tech, which straddles ecommerce and SaaS, likely sees much lower – and thus unsustainable – LTV-to-CAC ratios.
LegalZoom, for its part, “aims to achieve” a ratio of 3:1 within twenty-five months.Footnote 120 Because LegalZoom does not publish (and in fact claims not to know)Footnote 121 its customer churn rate, it’s difficult to calculate its exact LTV-to-CAC ratio based on publicly available data. That its goal is a ratio of 3:1 suggests that LegalZoom’s actual ratio is lower, and thus, below industry benchmarks.
By contrast, SixFifty achieves a 3:1 LTV-to-CAC ratio,Footnote 122 putting it on par with other business services but below analogous financial services. But SixFifty is likely unique among its legal tech peers; its primary customers are businesses with large annual contracts and thus higher average LTV.Footnote 123 By contrast, LegalZoom must retain a subscription customer for ninety days to recoup acquisition costs.Footnote 124 Despite enjoying higher LTV, SixFifty’s CAC is likely to remain high because it uses a traditional sales motion reliant on SEO. Even this legal tech success story is doomed to chase LTV through cross-selling or by retaining subscription customers for as long as possible.
* * *
More so than the “usual suspects,” market-based barriers to growth explain anemic legal tech. Through these case studies, we can trace the universal profitability struggle to a single, relatively simple culprit: the customer. Scholars envision DTC legal tech serving individuals in need of unbundled, specialized legal services.Footnote 125 But demand for such services is uniquely fickle: Customers rarely understand their own legal need and, unlike annual check-ups or taxes, specialized legal need is rarely ongoing. By serving their clients, providers often render themselves moot. And given high CAC costs, one-off clients cannot sustain DTC legal tech. If we want to move the needle on access to justice, we need to accept the reality that leaving legal tech growth up to the “market” might mean that effective, scalable legal tech for ordinary people remains out of reach.
8.4 Implications
To this point, we have identified and tested an alternative, market-based theory for lagging DTC legal tech that provides a fuller, industry-spanning explanation than the standard arguments around restrictive rules, weak AI, or the court technology checkerboard. This section builds out some of the implications of our findings for the future of legal services and access to justice. Two loom largest. First, the structural challenges facing DTC legal tech hold important lessons for current “regulatory reform” efforts, and they may even tip the scales away from liberalization strategies that aim to foster a robust, competitive legal tech marketplace. Second, market-based barriers to a robust DTC legal tech ecosystem may have fundamental implications for how courts should see their role, and build out their own technology and data infrastuctures, as the civil justice system enters its next chapter.
8.4.1 Is Regulatory Reform a Wrong Turn?
Our market-based theory offers a cautionary note, perhaps even waves bright red flags, for theories of regulatory reform that aim to foster, or otherwise depend upon, a robust DTC legal tech market as a way to serve the millions who currently face civil justice problems without meaningful legal help.
Regulatory reform, while a complex amalgam of reform impulses, can be distilled into three basic flavors. Entity-level reform efforts aim to relax UPL, Rule 5.4, or both in order to permit entities – from already-existing ones, such as law firms or LegalZoom, to new tech startups – to devise new, entity-level legal services delivery models. Direct-to-consumer legal tech finds its footing in this entity-level flavor. Role-based reforms call for new, top-down licensure schemes to mint new classes of legal professionals – think physician assistants or nurse practitioners for law – with entry requirements, ongoing educational and reporting requirements, and regulatory oversight patterned after current regulation of the legal profession.Footnote 126 But a third flavor, “community justice workers,” is gaining steam.Footnote 127 Drawing from both entity – and role-based approaches, this new reform model calls for deputizing certain trusted legal services providers – for instance, legal aid groups – to hire, train, and oversee new types of nonlawyer service providers, rather than top-down licensure, in order to extend their reach. This trio of reform flavors is currently jockeying for position across dozens of states as access-to-justice concerns move more squarely onto legal and political radars.
The deep, market-based challenges facing DTC legal tech plainly present a significant challenge to the first, entity-level approach – and may even tip the scales in favor of the other models. Indeed, if one-to-many legal services are unlikely to flourish, and if sandbox reforms, as in Utah, or ABS systems, as in Arizona, cannot yield a robust, competitive marketplace of DTC legal tech providers, then perhaps reformers should put their eggs in other baskets. Seeking political change is not costless. Reformers must be careful to spend their limited real and political capital wisely, and in ways that will achieve movement ends, rather than seeking change for change’s sake.
To be sure, this chapter’s anatomy of the market-based challenges facing DTC legal tech does not perform a full-dress analysis of the relative economic viability of all three reform models. Role-based reforms suffer from economic challenges of their own: The quality of services and the fees charged by newly minted paraprofessionals will almost certainly be directly proportional to entry and training costs and the stringency of regulatory oversight. And even a significant discount off the fees charged by lawyers – currently averaging $300 per hour – will still price most Americans out of the market. The community justice worker approach has problems of its own, as it depends on legal aid and other providers who are perennially funding-strapped and may not be inclined to take on significant new oversight duties in support of a model that, in effect, cannibalizes their own service provision.
Here, then, lies a more bracing possibility: Perhaps regulatory reform is itself a wrong turn – a market-based move that cannot solve the core market failure that is fueling reform calls in the first place. Perhaps, in other words, we should be seeking other ways to narrow the justice gap and serve the millions who must navigate a complex civil justice system with little or no help. That leads directly to a second implication: the role of courts, and the technology choices they make, as they prepare for the next chapter of the civil justice system.
8.4.2 Courts’ Make-or-Buy Choice
A second, and related, implication of our findings goes to how courts can best leverage potent new legal technologies, particularly with the advent of generative AI, for self-represented court users.
In the current court tech landscape, we see the challenge facing courts as a version of “make or buy.” Courts can “make” new legal services by creating virtual self-help centers, from simple chatbots to a full-scale, end-to-end digital courthouse front door where court users can come, input a plain language description of a problem, and get back actionable information and advice, be routed to other appropriate forms of legal help, or access needed form-filing and e-filing tools. Courts can also offer court-hosted online dispute resolution (ODR) platforms. The first ODR platforms at eBay and PayPal were simple gathering places where disputants could gather, usually asynchronously, and bargain their way to a settlement.Footnote 128 But with generative AI, we can glimpse more advanced versions – call it ODR 2.0 – that primes the disputants with information, including, perhaps, a prediction as to how the case would come out if litigated to a judgment in court.Footnote 129
Alternatively, courts can “buy” assistance for self-represented litigants by improving accessibility and inter-connectivity – for instance, through application programming interfaces (APIs), or by adopting common data standards – and by taking other measures, such as lowering e-filing barriers, to make it easier for legal tech providers to provide fully integrated services to self-represented litigants.Footnote 130
Neither “make” nor “buy” is without deep challenges. As already noted at length, “buy” depends on a robust legal tech marketplace that may not materialize even with substantial relaxation of regulatory constraints on legal services and more standardized, less checkerboard-like court technology and data infrastructures. The uniquely challenging market economics of serving moderate and low-income Americans may simply be too much to support a truly robust marketplace of legal services providers.
“Make” can seem no less daunting. Above all, it requires significant technical capacity that few courts currently have. Indeed, courts are woefully lacking in IT expertise, never mind AI. And, even if they partner with one of the purveyors of foundational, open-source LLMs in order to harness the newest and most potent forms of generative AI, significant in-house work will need to be done to customize and improve the accuracy of the models. As a result, partnering may be necessary in the age of generative AI, but partnering only partially solves the problem of limited court technical capacity.
Still, a reasonable conclusion is that “make” may be the better course, particularly if budget-strapped courts can only choose one. More bracingly, it’s possible that “make” is the only plausible path forward, particularly in the parts of the civil justice system where access concerns are most acute – the parade of debt collection actions, evictions, and family law matters that now dominate state court dockets. In that part of the civil justice system, as the average litigant’s ability to pay moves toward zero and profit margins thin, the market-based critique advanced in this chapter, and the uniquely challenging market economics of serving low- and even medium-income Americans, gains even greater purchase, leaving courts and policymakers with few other levers. Indeed, if a truly robust DTC legal tech marketplace is doomed from the start, then court-hosted, taxpayer-financed, “public option” legal tech – not selective, privately offered services serving the relatively better off – may be the wiser and fairer use of limited public funds.
8.5 Conclusion
This chapter has squarely asked the question that has long hung over legal tech: Why has DTC legal tech that serves individuals and small businesses directly not risen up amidst massive unmet civil legal needs? Our answer is a disheartening one: Inclusive legal tech may never materialize, at least not at a scale that can more than dent the problem of access to justice, because of the uniquely challenging market economics of serving ordinary people. But there is power in that recognition, for it can push us to consider what other measures might be necessary to make the civil justice system more open and fairer for all.
Open access