Lessons from founding a venture-backed health equity startup

LingoHealth started with a mission — to create an inclusive solution for patients who were often left behind by the US healthcare system. My cofounder and I recognized this opportunity in our professional work — we both had backgrounds working in healthcare — and in our personal lives as the children of immigrants. Beyond our first-hand experience, public health research and news emphasized the need to reduce inequities for those who face language and cultural barriers.

Like other now-defunct startups, LingoHealth couldn’t get to product-market fit fast enough. The ultimate product we pursued — patient engagement and care navigation for limited English proficient Medicare Advantage members — may not be truly solvable by a venture-backed software company. It was incredibly difficult to make the decision to wind down. It is heart-wrenching to say goodbye when we know there is still a significant problem left to solve. We believe that there can and should be better care for families like ours.

Fundamentally, LingoHealth was trying to solve a “people problem” related to finding a qualified multilingual labor force to activate some of the hardest-to-engage patient populations. We hope our learnings will help whoever comes after us develop a sustainable solution to this enormous problem in a scalable way.

Our key learnings (explained in full below) are:

1. Beware of a tech-enabled services model with specialized talent
2. Build trust during user acquisition
3. Know how to fit into the broader ecosystem supporting your mission
4. Find your customer early (ideally before anything else)
5. Avoid making “chocolate-covered broccoli”
6. Spend less time on the problems you can’t solve

Before I dive into our learnings, let me take you on our journey, though feel free to jump to the last section directly for the detailed takeaways.

A Big & Personal Problem

On the surface, it seemed like language was the biggest challenge for limited English proficient (LEP) patients. However, as we talked to more patients and family caregivers, we learned that the gaps omnipresent throughout the broken US healthcare system — cost concerns, mistrust, and lack of access — were exacerbated by language and cultural barriers.

With the spotlight on health disparities in 2020, health equity was top-of-mind for health plans and providers. They were hungry for a solution that could bridge the gap for vulnerable, hard-to-reach patients. Our personal experiences highlighted the need for a more inclusive solution to serve aging immigrant patients than what the traditional US healthcare system offered. Demographic shifts emphasized the economic urgency to this problem: non-white seniors will be 45% of the senior population by 2060, with estimated Medicare spending of $503+ billion.

We set off to fully understand this problem. In interviews with older LEP adults, they often mentioned a lack of trust due to language barriers with their providers. Many LEP adults feared unknown costs and lacked understanding of how to manage their health. Especially during acute health events, LEP patients often felt they were treated like “second-class citizens.” We heard countless heartbreaking anecdotes of how language barriers contributed to misdiagnoses or even more tragic outcomes. Depending on the age, socioeconomic status, and technology proficiency, there could be a range of other social determinants of health that complicated their healthcare journey in and beyond the doctor’s office. Oftentimes, they turned to English-fluent family or community members for support, while trying to balance not burdening others.

Multicultural, multigenerational families, where the younger generation feels a strong sense of duty to care for elders, were critical to the healthcare navigation for LEP older adults. Through surveys, we found that young adults from families that did not speak English at home were three times as likely to be involved in the healthcare of their loved ones, versus their counterparts from English-speaking families. While culturally competent care navigators or community health workers could fill a similar role, we wondered if better support and activation of informal caregivers might improve patient outcomes and alleviate the burden on the healthcare system. These younger, more English fluent and tech savvy family caregivers were often a safety net for their family, and many worried about “what if” scenarios as their loved ones age. These concerns echo the sentiments felt by the “Sandwich Generation”: feeling pressure to care for aging parents and their own growing families. Many immigrant children referenced financial, mental health, career, and personal relationship implications from these caregiving pressures. Some immigrant children became involved during a parent’s health scare, but felt helpless when they didn’t have the full picture or an understanding of the best next step. Others wanted to balance ensuring their parents were adhering to lifestyle changes and medications, while not being a “nag”. We often heard statements like, “I’ll get disowned if I ever send my loved ones to a nursing home, but I don’t know if they are actually staying healthy.” How might we offer these essential, informal care navigators peace of mind and help their LEP family members get better care?

A screenshot from one of our many pitch decks…

The Roller Coaster Ride of Building

We built LingoHealth to meld our mission with the promise of technology. To summarize our product journey: we started with a culturally inclusive family caregiver solution (think Wellthy for adult children of immigrants), then pivoted to a senior support solution focused on specific immigrant populations (think Papa for different cultural groups of LEP seniors), and finally landed on a patient engagement / care navigation solution for LEP older adults (think Accolade for different cultural groups of LEP patients). For a detailed graphical overview of our journey, click here.

We initially focused on addressing the care needs we had heard from LEP patients and their family caregivers through a digital solution. We launched a family caregiver-oriented app and our “Early Access Program” in English, Spanish, traditional and simplified Chinese. After receiving feedback, it was obvious that we had struck a chord with our beta users, but most did not have pressing care needs in their families.

We segmented these users into two personas:

• “Dutiful and Unsupported Danielle” — the immigrant kid who was actively helping and advocating for a loved one with health or aging-related needs, with limited support
• “Busy but Obliged Bianca” — the immigrant kid who was not currently supporting a loved one but knew they may be called upon in the near future

We ultimately onboarded more “Biancas,” who wanted help with monitoring their aging loved ones from afar or better engaging them with preemptive discussions on aging. These “Bianca” concerns were rife with thorny and nuanced interpersonal dynamics that felt tricky for a startup to solve, so we, instead, tried to acquire more “Danielles.” Yet, we learned that typical channels for finding “Danielles”, like Facebook groups for family caregivers, often lacked enough members with LEP loved ones. Wanting to stay true to our original mission, we went back to the drawing board.

We knew that we would need to work with payers, such as health plans or risk-bearing providers, to monetize our product. We committed to using a B2C2B strategy and found our initial users in direct-to-consumer channels, given the success of other digital health companies. We spoke with several potential Medicare Advantage partners to gauge their appetite for a pilot. Although they could logically understand how a family caregiver solution like our MVP might benefit their vulnerable older members, they couldn’t justify the ROI on this type of product since the impact on patient outcomes felt more indirect. We aren’t alone in this finding. Most B2B family caregiver solutions sell into employers of caregivers to help with productivity and retention of employees who have family responsibilities.

So we pivoted!

We narrowed our focus to the LEP older adult. Family caregivers would be part of the final solution, but we wanted to focus on serving the older adult first. We launched a series of different prototypes around Medicare enrollment, care navigation, and patient engagement, focusing on Chinese seniors to start given the dearth of existing resources in-language.

We ran into two existential issues with this product. First, distribution channels for LEP older adults were unexpectedly gatekept and made user testing challenging. When we reached out to leaders at community based organizations (CBOs) and social groups for this population, they were happy to connect but were not comfortable with sharing our product with their members, largely out of concerns about user confusion since we were in “product feedback” mode. We then recruited “community ambassadors” through our networks (e.g., our parents’ friends of friends), but most ambassadors felt disengaged and didn’t actively recruit others to join. Without a path to acquire LEP older adult users for product testing, it was difficult to validate our user hypotheses.

Second, we saw that our value proposition relied on a level of trust and cultural sensitivity that couldn’t be achieved just through software or text-based communication. We would need to build a multilingual staff, which would be a significant lift to recruit and train. One of our biggest barriers was what we called the “call center” problem — we knew we needed some human-enabled member services, but it became more and more likely that our perceived value proposition would be predicated upon this multilingual, high-touch service. While our founding team could support up to a certain number of individuals, it was unclear how we would grow without more bilingual human capital or financial capital from pilot partners.

Amid these concerns, we continued to talk with potential pilot partners. We found a Goldilocks problem: payers in areas that had less dense LEP patient populations were less interested in spending more resources on solving this problem, while payers in areas that had dense LEP patient populations had already invested in in-house resources, like culturally-specific programming or bilingual staffing, to better address these needs. As we broached partnerships, we noticed that our product was frequently pigeonholed as a “translation service”. To many large payers, translation / interpretation services were a “check the box” function and a race to the bottom on pricing. Our challenge was securing enough partners with a willingness-to-pay for culturally inclusive engagement features or services, such as family caregiver involvement, beyond translation or interpretation. Although we eventually connected with organizations interested in our envisioned culturally inclusive platform, the operational lift required to serve a pilot for thousands of members in multiple languages would require more resources than we could offer or they could pay.

With these pilot challenges and deteriorating market conditions, we didn’t have confidence that we had the resources to build a product to effectively address our mission. We made the decision to not move ahead.

What We Learned

1. Beware of a tech-enabled services model with specialized talent: We realized as we moved further into exploring patient engagement and care navigation that our product would require a significant operational lift to offer in-language call-based support. From a team perspective, we needed a dedicated person for translation and in-language customer service. And we knew from all of the patient stories that interpretation and translation are difficult to do well (machine translation helps but is far from good enough for trust building and would require substantial quality assurance), and talent for this type of work is scarce both in the US and abroad.
   Our audience also wanted to engage through WeChat, LINE, WhatsApp, and calls, which made the copywriting difficult to scale as it needed to be built across multiple channels, in multiple languages. This meant we would need to not only have bilingual content in our digital platform but also multimodal interactions, such as short videos or phone call alternatives to filling out an online onboarding form, to ensure members could use our product. All of this would require costs that would throw our unit economics out of whack.
   We spoke with a number of other executives who needed multilingual customer service teams, and bilingual talent shortages consistently appeared as the hardest problem. Building and scaling bilingual teams will likely continue to be a problem as organizations navigate language inclusivity.
2. Build trust during user acquisition: User acquisition is extremely difficult in the space where we landed (Mandarin-speaking older adults, or even more broadly, LEP older adults). Despite our B2B2C strategy, user acquisition was immediately necessary for user research, which was near impossible. User acquisition should be prioritized for its importance in validating and building the product. Since we intentionally targeted a group of users who mistrusted the “system”, we were especially hindered by the fact that our team did not “look and feel” like the end user (an LEP older adult). We then faced a “chicken and egg problem” where we needed user acquisition to refine our value proposition but couldn’t acquire users without a defined value proposition. Compared to working with family caregiver users where we were successful in building trust through our shared background and empathy as children of immigrants, it became obvious that we couldn’t readily garner the trust of LEP older adults without expanding our team.
   Moreover, building trust and designing for inclusivity risks seemingly starting “narrow”. To avoid this, we purposely started with Spanish and Chinese but then confirmed these two non-overlapping markets required meaningfully different user acquisition strategies. After shifting to Chinese only, the divide between Cantonese, Mandarin, and other dialects raised questions as to whether we could build trust without speaking the same dialect as users. There were so many different cultures that we ultimately would need to build trust with — each with its nuances. We talked to a potential Bay Area partner who would’ve been more interested if we had Cantonese services, while a potential Los Angeles partner only wanted support in Korean and Vietnamese. These fragmented audiences made it difficult to establish a beachhead and iterate more on building trust with users.
3. Know how to fit into the broader ecosystem supporting your mission: In our experience, impact-oriented startups will face the question “why is this not a non-profit?” more often than other startups. There’s certainly some bias in that statement (e.g., “you can’t run a business and care about underserved communities at the same time”). But it raises questions about the broader ecosystem. Why would someone use a digital product, instead of visiting their local nonprofit community organization or using an existing government service (e.g., SHIP)? How are we able to collaborate with CBOs in a way where it’s a win-win for both? Getting exact answers to these questions ASAP can help differentiate your value proposition better and build referral channels earlier.
   By fitting into this broader ecosystem, you can also better avoid point solution fatigue. In digital health, users are now inundated with different apps and portals. We had to contend with users (and partners) feeling point solution fatigue. A common question we encountered as a patient engagement / care navigation product was how we would be different from a patient portal or how we would work in tandem with it. Patients are increasingly juggling multiple portals with their health information, and many, understandably, are turned off by this experience.
4. Find your customer early (ideally before anything else): In B2B2C, customers and users are often two distinct stakeholders with different needs. This means solving the Venn diagram between early adopters and people with a hair-on-fire problem twice. In hindsight, we would’ve built partnerships earlier, even before considering efforts to reach out directly to users. There were multiple places we could’ve fit into a patient’s journey but where we should start differed depending on the payers we talked to and where it would provide the most value to them. Without securing an “early adopter” partner, it was near impossible to crystallize a specific value proposition. Finally, an experienced (and collaborative) partner can help establish solid compliance protocols, such as SOC2 and HITRUST that will likely be needed beyond HIPAA.
5. Avoid making “chocolate-covered broccoli”: It’s hard to avoid confronting behavior change when developing patient engagement products, which often entail increasing use of preventative care. For example, trying to get more patients to an annual wellness visit through reminders and rewards may work to a degree, but it doesn’t fix the underlying reason most folks don’t go: because they do not understand the value of these visits. As a result, solutions that lower surface-level barriers to entry may only work for a subsegment of users and may not be a sustainable, scalable approach. We learned that there were too many barriers that we couldn’t solve: financial pressure, poor prior experiences, and “I-just-don’t-want-to”, to name a few.
   Furthermore, we saw that direct-to-consumer healthcare has higher engagement and perceived value when there is a “transaction” like a telemedicine appointment or a mailed prescription, but a “nice to have’’ service like information about Medicare or an on-call care concierge was difficult to garner continued engagement, especially for an audience who was already skeptical. For instance, we’ve seen older Chinese adults get excited to attend free virtual workshops hosted by Chinese doctors to get questions answered on the spot. Without the ability to offer a “transactional” service to address an urgent need, it felt difficult to garner the level of engagement we needed.
6. Spend less time on the problems you can’t solve: This might feel obvious. Nonetheless, we, as a mission-oriented team, often waffled on decisions like: should we be purely non-clinical or clinical? Are we offering anything related to financial support for affordability or price transparency? We didn’t have the resources or expertise in-house to solve all these concerns, but we felt obligated to think through solutions given what we had heard from potential users on why they loathed their care experience. These questions took our focus away from building towards what we could solve and finding pilot partners earlier. These “rabbit holes” escalated the rate of burnout for our team during development. In the end, managing our “emotional runway” was as important as managing our financial runway.

Although LingoHealth is ceasing operations, I remain grateful to have had the opportunity to work on addressing this important health equity issue. I hope the learnings from our journey will empower others in this space to think innovatively about how to improve the care experience for underserved communities.

To our early users, thank you for taking a leap of faith and using our product, for trusting us with your family’s care, and for believing that our small team could make a difference in healthcare. To our team, I’m indebted to each of you for your loyalty and sacrifices on this journey. I’m forever grateful for the insights and support of our brilliant formal and informal advisors. And none of this would have been possible without our investors — thank you for taking a chance on us and being in our corner throughout the highs and lows.

I look forward to finding more opportunities to fight for health equity. My inbox is always open to discussing more about healthcare startups or equitable access — find me on LinkedIn to stay connected.