What the 2010s home-care apps got wrong about families
There is a folder on Susan Brandt's laptop called "Tried It." Inside are bookmarks, downloaded PDFs, and a few exported data files from seven different apps she used between 2013 and 2021 while caring for her father, Gerald, who had Parkinson's disease and congestive heart failure and lived alone in a ranch house outside Columbus, Ohio. CareZone. CaringBridge. Lotsa Helping Hands. A task-coordination app whose name she can no longer remember but whose logo was a green circle. A medication tracker that synced with Gerald's pharmacy for about four months before the pharmacy changed its API and the sync broke and nobody fixed it.
Susan is 57 now. Gerald died in 2022. She kept the folder not out of sentimentality but because she is a project manager by profession and she could not stop herself from documenting the gap between what each tool promised and what it actually delivered. The gap, she says, was almost always the same gap. "They all assumed that the problem was information. That if everyone in the family could see the same calendar, the same medication list, the same task board, the work would distribute itself." She pauses. "The work did not distribute itself."
The decade between roughly 2012 and 2022 produced a remarkable number of technology products aimed at the family caregiving market. Some were built by founders who had cared for a parent and wanted to solve the problem they had lived. Some were built by digital health entrepreneurs who had identified eldercare as an underserved market with favorable demographic tailwinds. Some were built by health systems and insurance companies trying to reduce readmissions and close care gaps. Nearly all of them shared a set of assumptions about what family caregiving was, who did it, and what it needed. Nearly all of those assumptions were wrong in the same ways. Most of the products are gone, or hollowed out, or absorbed into something else. The families are still there, doing the work the same way they were doing it before the apps arrived.
Understanding what went wrong is not an exercise in nostalgia for failed startups. It matters because the next generation of tools — and there is a next generation, already being built — will repeat the same mistakes if nobody names them clearly.
The coordination fallacy
The dominant theory of the 2010s caregiving app was what might be called the coordination fallacy: the idea that the core problem in family caregiving was that people didn't know what needed to be done, or didn't know who was doing it, or couldn't see the same information at the same time. Fix the visibility, the theory went, and the work would follow.
This theory was not entirely wrong. Coordination failures are real. A family where the daughter in Phoenix doesn't know that the son in Pittsburgh already scheduled the cardiology appointment, and so schedules it again, and the father ends up with two appointments on the same day and misses both — that family has a coordination problem. A shared calendar helps.
But coordination was rarely the binding constraint. The binding constraint was almost always labor: there was not enough of it, it was not distributed fairly, and the people who had more of it were not going to get more help from their siblings because a task board told them what was left undone. The task board, in fact, sometimes made things worse. It made visible, in a way that could not be ignored, exactly how much one person was doing and how little everyone else was doing. That visibility did not produce equity. It produced conflict.
Renata Okonkwo, a 49-year-old in Houston who cared for her mother, Adaeze, through the last four years of her life, used a popular caregiving coordination app for about eight months in 2017. She had three siblings — one in London, one in Atlanta, one twenty minutes away in Sugar Land. The app showed, in real time, who had completed which tasks. "My brother in Sugar Land could see every single thing I was doing," she says. "He just didn't do anything about it. The app didn't make him show up. It just gave him a more detailed picture of what he was choosing not to do."
The coordination fallacy reflected a particular kind of technological optimism that was widespread in the 2010s: the belief that friction was the enemy and that removing friction — making it easier to see, easier to communicate, easier to assign — would change behavior. In many domains, that belief was correct. In family caregiving, it ran into something the apps were not designed to handle: the fact that the friction was not logistical. It was relational. It was decades of family history, unresolved resentments, geographic distance that was sometimes chosen, financial arrangements that nobody wanted to revisit, and the fundamental human capacity to look at a task list and decide that someone else will get to it.
The nuclear family that wasn't
The second major mistake was demographic. The apps were designed, almost universally, for a particular family structure: two or three adult children, geographically dispersed, coordinating care for one aging parent who lived alone or with a spouse. The interface assumed a relatively small, relatively legible group of people with roughly equivalent relationships to the care recipient and roughly equivalent standing to make decisions.
This family exists. It is not the majority of caregiving situations.
The actual landscape of family caregiving in the United States is considerably messier. It includes spouses caring for spouses — a situation where "family coordination" means one person doing everything, and an app designed for group task assignment has nothing useful to offer. It includes adult children who are also caring for their own children, sometimes children with disabilities or chronic conditions of their own, so that the caregiving is not a discrete project with a defined team but a continuous, overlapping set of obligations with no clear boundary. It includes blended families where the relationships between potential caregivers and the care recipient are complicated by divorce, remarriage, and the particular tensions of step-relationships. It includes households where the primary caregiver is a grandchild, a niece, a neighbor who became family, a friend who is the only person left.
It includes, very commonly, a situation where one person is doing almost all of the care and the "family" that the app is designed to coordinate is largely theoretical — people who are listed as contacts, who receive updates, who express concern from a distance, but who are not going to fly in to do the laundry.
Marcus Webb is 44 and lives in Detroit. He has been his grandmother Loretta's primary caregiver since 2019, when Loretta's daughter — Marcus's aunt, who had been the primary caregiver before him — had a health crisis of her own. Loretta is 84 and has Type 2 diabetes, moderate dementia, and the kind of stubbornness that her doctors call "preserved personality" and that Marcus calls "she threw a slipper at the home health aide." Marcus has two uncles and a cousin who are theoretically part of the caregiving network. He downloaded a caregiving app in 2020 and invited all of them. His cousin responded to one task. His uncles never logged in. The app sent him weekly summaries of how many tasks were outstanding. He deleted it after three months.
"It kept showing me all the things that weren't getting done," he says. "I already knew what wasn't getting done. I was the one not doing it because I didn't have time."
The apps were designed for a family that was ready to work together and just needed a better system. Most caregiving families are not that family. Most caregiving families have one person who is doing the work and a variable number of people who are not, for reasons that range from geography to denial to their own health problems to the fact that they and the primary caregiver have not spoken civilly in six years. An app cannot fix any of that. An app that assumes it can will frustrate the person who needs help most.
The body that wasn't in the room
The third mistake was about what caregiving actually is. The apps were, almost without exception, information products. They managed calendars, task lists, medication schedules, care journals, document storage. They were good at things that could be represented as data: who has an appointment on Thursday, what medications are due at 8 a.m., where the insurance card is filed.
They had nothing to offer for the things that cannot be represented as data. The physical work of transferring a person from a bed to a wheelchair. The 45 minutes it takes to help someone with Parkinson's get dressed in the morning, not because the task is complicated but because the disease makes every movement slow and the caregiver has learned that rushing makes it worse. The decision, made at 2 a.m., about whether the labored breathing is something to call the doctor about or something to watch until morning. The conversation, repeated every few days, with a person who has dementia and keeps asking where her husband is, and her husband has been dead for eleven years.
None of that is a coordination problem. None of it is solved by a shared calendar. It is physical, emotional, and clinical labor that requires a body in the room, and the apps — designed by people who were mostly thinking about information management — had no framework for it.
This mattered in a specific way: the apps implicitly defined caregiving as the tasks that could be put on a list, which meant they systematically underrepresented the work that couldn't. A caregiver who used an app to track Gerald's medication schedule and doctor's appointments was generating a record that showed Gerald's care as a set of discrete, completable items. The record did not show the four hours a day Susan spent in the house doing the things that don't go on a list: the companionship, the monitoring, the physical assistance, the emotional management of a man who was frightened and sometimes angry about what was happening to his body.
When Susan showed the app's summary to her brother, who lived in Portland and called every Sunday, he looked at the task completion rate and said it seemed like things were under control. Things were not under control. Things were held together by Susan's continuous presence, which the app had no way to represent.
The invisibility of physical and emotional labor in digital caregiving tools is not a minor design flaw. It is the central reason those tools failed to change the distribution of caregiving work. You cannot redistribute labor that the system cannot see.
The market that wasn't there
There is a fourth mistake, less about families and more about economics, but it shaped everything else: the apps were built for a market that did not exist in the form the founders imagined.
The caregiving app market of the 2010s was premised on the idea that family caregivers would pay, directly or through their employers, for tools that made caregiving easier. Some employers did offer caregiving benefits, and some of those benefits included app subscriptions. But the direct-to-consumer market was thin. Family caregivers are, as a group, financially stressed — they are spending money on care, often reducing their own work hours to provide care, and frequently depleting savings. Asking them to pay $9.99 a month for a task-management app was asking them to add a line item to a budget that was already stretched.
The alternative revenue model — selling to health systems, insurers, or employers as a care-coordination tool — required demonstrating outcomes that the apps were not designed to measure and often could not produce. A health system that wanted to reduce 30-day readmissions needed something more than a shared calendar. An insurer that wanted to reduce emergency department utilization needed clinical integration that most of the apps could not deliver. The enterprise market was real but demanding, and most of the consumer-facing apps were not built for it.
The result was a graveyard of products that had genuine users, genuine utility for some of those users, and no sustainable path to revenue. CareZone, which had real traction as a medication and health-record management tool, was acquired by Walmart in 2020 and then effectively shut down, its users given 30 days to export their data. The shutdown left thousands of caregivers scrambling to reconstruct records they had spent years building in a system that turned out to be rented, not owned.
Diane Castellano, a 52-year-old in Albuquerque who had used CareZone for four years to manage her husband's multiple sclerosis medications and her mother's post-stroke care, got the shutdown notice on a Tuesday and spent the following weekend exporting, reformatting, and rebuilding her records in a spreadsheet. "I had put everything in there," she says. "Every medication change, every lab result, every note from every appointment. Four years of information. And then it was just gone, because some company decided it wasn't worth keeping." She now keeps everything in a Google Sheet and a physical binder. She does not use caregiving apps.
What the failure left behind
The failure of the 2010s caregiving apps did not leave the caregiving landscape unchanged. It left it changed in specific ways that are still being felt.
It left caregivers more skeptical of technology products that promise to solve caregiving problems. That skepticism is rational and earned, but it also means that genuinely useful tools face a higher burden of proof than they would in a market that had not been burned before. Caregivers who spent time migrating to a platform that then shut down, or who invited their siblings to an app that their siblings ignored, or who built a medication-tracking system that broke when the pharmacy changed its software — those caregivers are not eager to start over.
It left a gap in the middle of the market. The tools that survived are mostly at the extremes: simple, free, general-purpose tools like shared Google calendars and group texts that caregivers have adapted for caregiving use, and expensive, clinically integrated platforms designed for health systems and care management organizations that are not accessible to individual families. The middle — purpose-built, affordable, genuinely useful tools designed for the complexity of real caregiving situations — is mostly empty.
And it left a set of lessons that the next generation of builders would do well to learn, because the demographic pressure that drove the 2010s app wave has not gone away. The oldest baby boomers are in their late seventies. The number of Americans over 80 will roughly double between now and 2040. The caregiving workforce shortage is worsening, not improving. The families who are doing the work are not going to have more time, more money, or more siblings who suddenly decide to show up. The need for tools that actually help is real and growing.
What a tool that actually helps would look like
The lesson of the 2010s is not that technology cannot help family caregivers. It is that technology designed around the wrong model of caregiving will not help, regardless of how well it is executed.
A tool designed around the right model would start with a different assumption: that caregiving is not a coordination problem among equals, but a labor problem concentrated in one or two people who need practical support, not a better way to ask their relatives for help. That means tools designed for the primary caregiver first — tools that reduce the administrative burden on the person doing the most work, rather than tools that distribute visibility to people who are not doing the work.
It would take seriously the physical and temporal reality of caregiving. A medication reminder that fires at 8 a.m. is useful. A system that helps a caregiver track whether a new medication is causing the fatigue that started two weeks ago, and formats that observation in a way that can be communicated to the prescribing physician, is more useful. The difference is between a tool that manages information and a tool that supports clinical judgment — the kind of judgment that caregivers are making constantly, without training, and often without anyone to consult.
It would be designed for the actual family structures that exist, not the idealized ones. That means being useful to a solo caregiver with no family network. It means being useful in a blended family where the relationships are complicated. It means not requiring a group of engaged, cooperative participants in order to function.
And it would be built with a sustainable economic model that does not depend on caregivers paying out of pocket for something they can barely afford, and does not depend on health system contracts that require clinical integration the tool cannot deliver. The most promising models are probably employer-based caregiving benefits — which are expanding, slowly, as employers recognize the cost of caregiver turnover and absenteeism — and integration with the Medicaid HCBS programs that fund home-based care for the most complex patients, where care coordination tools have a clear clinical value proposition.
Susan Brandt still has the folder on her laptop. She is not bitter about the apps that didn't work. She is a project manager; she understands that most products fail, and she understands why these ones did. What she is, still, is tired — tired in the way that four years of primary caregiving leaves a person, tired in a way that no app was ever going to fix, but tired also of the particular exhaustion of being promised a solution and handed a task board.
"What I needed," she says, "was someone to come to the house. Or money to pay someone to come to the house. Or a system that would have made it easier to find someone to come to the house and figure out how to pay for them." She pauses. "None of the apps did any of that. They all assumed the problem was that I didn't have enough information. I had plenty of information. I had too much information and not enough help."
That is the sentence the next generation of builders needs to read before they write a single line of code. Too much information and not enough help. The families are still there. The work is still there. The tools that will actually matter are the ones built around that sentence, not around the fantasy of a family that coordinates itself if only someone gives it the right software.