Soon after its launch in 2016, Betterhalf.ai, a pioneering matrimony app in India, matched a male graduate from the Indian Institute of Management Ahmedabad, a leading business school, with a female graduate from the same school. The match was based on results produced by its compatibility quiz. The artificial intelligence–backed algorithm declared the match as 89 percent compatible. Shortly after the first date, the male user wrote negative feedback: She wasn’t even able to understand my jokes. How can you say that she was 89 percent intellectually compatible?
Rahul Namdev, the co-founder and chief technology officer of the app, stood by the match. The man and the woman had both graduated from one of the best colleges in India, a fact that Namdev said “in itself is the accuracy of my algorithm.” Still, Namdev was reluctant to dismiss a user’s feedback. “You can only give users a vague idea of how compatible they are with each other,” Namdev concluded. Since then, the app has replaced percentages with graphical charts to represent user compatibility through multiple quantitative variables.
In India, marriage has long been the province of intelligence: the mind over the heart, sound reasoning exercised by the leadership of families to maintain their stability as well as their social and economic status. In 2021, the BBC reported that some 90 percent of marriages in India are still arranged, a triumph of tradition over change. At the same time, technology has been enacting changes in the practice of that tradition. The driving principle is still intelligence. Only now it’s artificial.
Betterhalf.ai is just one of India’s more than 1,500 matrimony portals, an industry valued in the hundreds of millions of dollars. According to The Economist, Matrimony.com, the biggest publicly listed such platform, had a market value of around $160 million (US) and annual revenue of $57 million (US), 850,000 subscribers, and more than 100,000 successful matches as of 2021. Today, these ever-influential sites have customer bases in the millions. In fact, the COVID-19 pandemic caused their stronghold in the country to only grow, with many Indians turning to the web to find marriage partners in the absence of in-person and organic connections. Both Matrimony.com and Shaadi.com claim to have grown by 30 to 50 percent during the pandemic, according to the New Indian Express.
Never before has a generation of Indians been able to choose from such a wide variety of partner-hunting methods. For the country’s youth, a majority of whom count marriage as a life goal, these technologically advanced services serve as an alternative to multifarious methods of finding a spouse.
Now, some of the most prominent market holders in the enterprise, including Betterhalf.ai, Jeevansathi.com, Matrimony.com, and Shaadi.com, are embracing AI as the next step in innovation. From predicting partner preferences to automating safety procedures and streamlining customer experience, the virtual-matrimony industry in India is striving to optimize the country’s generations-old custom of arranged matchmaking.
Traditional Indian matchmaking as we know it—a reliance on word-of-mouth referrals, family and friend networks, matrimonial brokers, and the belief that marriages are made between families, not individuals—stands at a moment of inflection. AI-driven matrimony attempts to democratize the process—promising that “perfect” life partners may be a few clicks and taps away. India’s millennials and Gen Z are awkwardly straddling the old and the new—resorting to age-old customs like “biodatas” (marriage résumés) and astrological charts while also vulnerable to contemporary technology–induced phenomena like online matrimonial fraud cases, swipe fatigue, and catfishing. Can AI do a better job of finding them a partner?
Mumbai-based journalist Pratiksha Thayil started to hunt for a spouse at twenty-six by subscribing to Shaadi, Jeevansathi, and Betterhalf. She wanted to find the kind of love her parents have—just through a different mechanism. Sitting in Mumbai’s Zen Cafe, Thayil laments her inability to find “someone who can fill in my dad’s shoes,” she says.
Thayil warily and almost nonchalantly swiped through profiles, navigating lazily written biographies. Her only superficial criteria are her future partner’s height and the city he lives in. She’s petite and would like her husband to be a little taller than her. “Now that I have a choice, I’m going to make a choice,” Thayil says. “My logic is simple. If it’s a love marriage, maybe you don’t think about these things. You just fall in love. But with an arranged marriage, I have the liberty to think of these things practically.”
Sociologist and academic Parul Bhandari, in her 2020 book Matchmaking in Middle Class India, calls this type of calculated bargaining, becoming increasingly common in India, “negotiated” marriages. “The modern is indeed about remakings, adjustments, and changes and is far from a neat category,” Bhandari writes. She illuminates the anxieties of a generation raised after liberalization: the need to settle professionally first, maritally second—the desire to achieve imagined notions of love and sex and to lead a lifestyle that matches or surpasses their current one.
Still, a common grievance with these older matrimony hubs is the heavy involvement of parents or siblings. In some regions of India, 40 to 50 percent of the accounts created on Shaadi are managed by parents. “Even if the profile is created by individuals, especially in the case of women, it gets managed by or is influenced by parents quite a bit,” Ekta Checker, Shaadi’s associate director of marketing, told The Print in 2022. Indeed, the practice is baked into the virtual-matrimony marketplace to the extent that websites ask users to specify the account owner—the primary user, their parent, sibling, friend, or relative. Thayil finds direct parental involvement “quite funny.” “I don’t even know if my matches are actually interested in me, because their family members are usually the ones operating the account,” she says. “For all I know, his mom might be the one interested in me.” In this way, India’s virtual-matrimony industry is simultaneously navigating tradition and modernity, clutching onto historical customs while projecting progressive belief systems. “In [Indian] culture, that positive tension existed even fifty years ago,” Betterhalf co-founder Pawan Gupta explains. “Parents want something else. Their kids would want something else.”
But the cost of going online to find a spouse has repercussions beyond the value divide. Even though, theoretically, it has to work only once, the indefinite exercise of swiping for a companion is taking an emotional, physical, and psychological toll, or “swipe fatigue,” on modern matchmakees. “Going through three different apps and so many profiles, not reaching anywhere, gets extremely frustrating,” Thayil says. “It takes a lot of patience.”
She is one of the many working professionals in India who manage a five- or six-day workweek along with the demands of living in a metropolis like Mumbai—commuting, chores, making time to see friends.
It was the app founders’ own comparable journey that led them to start Betterhalf. Gupta and Namdev set out on their partner-hunting ventures only to realize—and attempt to solve—the common fail point they believed made these platforms ineffective: compatibility.
In 2018, one of India’s first AI-backed algorithms emerged in the matrimony sphere. Inspired by their own arduous and anxiety-filled partner-search efforts, Bengaluru-based Gupta and Namdev founded Betterhalf. In 2015, the two had moved to San Francisco after graduating from the Massachusetts Institute of Technology. They regularly exchanged stories of failed conversations with women they met online—Namdev on his community portal, Gupta on Shaadi and Jeevansathi. They realized that these services failed to prioritize compatibility, focusing mainly on caste, religion, income, and physical appearance, says Gupta, seated in his cubicle at Betterhalf’s Bengaluru-based headquarters, where he leads a 150-person team. “We wanted to use mathematics and machine learning to route more and more people towards marriages.”
Betterhalf brands itself as more progressive than its predecessors Shaadi, Jeevansathi, and Matrimony. That’s not just for its use of advanced technology but also its ideology, touting itself as India’s first matrimony app free from familial intervention. The app uses machine learning, an aspect of artificial intelligence in which the algorithm improves its performance over time as it encounters new scenarios. With the appearance and feel of a regular dating app, Betterhalf’s AI-backed algorithm trains on its user’s lifestyle and partner preferences, purportedly generating more accurate and suitable matches over time.
For Thayil, the app has generated superior matches. She receives more satisfactory suggestions there than on other apps. The older websites are often not even able to filter her results based on the basic criteria of location. Thayil’s biographical background is typical of Betterhalf’s users: working professionals between the ages of twenty-five and forty, coming from tier-one or metropolitan cities such as Delhi, Mumbai, Bengaluru, Chennai, Hyderabad, Kolkata, Ahmedabad, and Pune.
Despite the niche demographic, Betterhalf’s numbers indicate fast growth. The app has a current average monthly user count of 1 million, and the founders pride themselves on its high retention rate of about 30 percent for paid subscribers (in comparison, international dating apps like Bumble, Hinge, and Tinder have a retention rate of about 10 percent a month after downloads).
But despite what the numbers might indicate, the nascent AI-matrimony industry in India is rife with problems—from algorithmic bias to data privacy and matrimonial fraud. “I don’t think we’re at a point where we can trust AI completely,” says Garud Iyengar, a professor in the department of industrial engineering and operations research in the engineering school at Columbia University. “AI should be used to increase your choices, not find you a mate.”
Authenticity is one of the biggest priorities for the developers of Betterhalf, according to product manager Shankar Krishnamurthy. The company uses AI-backed technology to verify users’ photographs and IDs. But Betterhalf also has an offline verification team to vet any cases flagged by the system. Currently, the app is set to limit the user’s connection requests per day for a non-verified profile; if one verifies their profile using a government-approved ID, they can send limitless connection requests daily. Now, the team is debating whether to make initial two-step verification mandatory. The decision comes in the wake of surging matrimony fraud cases in the country, from sextortion to Tinder Swindlers and large-scale con schemes. Many of the instances can be traced back to the weak verification systems of dominant matrimonial sites.
In February 2023, a woman extorted about $100,000 from a sixty-five-year-old Mumbai-based widower whom she had met on a matrimonial portal, as reported by the Times of India. The woman threatened to circulate their sex footage on social media and amongst his private contacts. Compelled by shame, the man paid up for months before finally approaching Mumbai’s cyber police. A year before this, a sixty-six-year-old man was caught after swindling twenty-seven women, through marriage, out of up to $16,500 each. This skilled con artist managed to deceive even educated, accomplished professionals—an assistant commandant of the Indo-Tibetan Border Police, a chartered accountant, a teacher, advocates practising in the Supreme Court of India and the Delhi High Court—by posing as a high-ranking health education executive on sites like Jeevansathi and Shaadi. He was caught a mere month after a conman posing as a scientist on a matrimonial site defrauded fourteen women of over $100,000, as reported by the Times of India.
These incidents are a fraction of the actual such cases, according to Radhika Jhalani, a tech policy consultant in India. Many online dating- or matrimony-related cases go unreported, she says, owing to the stigma around finding a partner on an app. Often, the targeted victim is a member of a marginalized or disadvantaged community—older women, widowers, or LGBTQIA individuals.
But the path to security isn’t as straightforward as mandating more documentation. Jhalani, along with Prateek Waghre, an independent policy researcher, and Tejasi Panjiar, former associate counsel of policy at the Internet Freedom Foundation, believes that the collection of confidential user information poses a risk of data breach. Panjiar cites the example of Bumble India’s data breach of 2020, where the app left the location and other private data of 100 million users unsecured. Many of these online portals don’t have end-to-end encryption for messaging, making the user texts secure only at the server level.
Weaknesses in data protection law in India amplify users’ vulnerability. After several years of work, India’s Ministry of Electronics and Information Technology finally released the country’s first data privacy law in 2023, the Digital Personal Data Protection Act. But unlike the 2021 version of the bill, the legislation fails to recognize personal and sensitive data as a separate category. “It’s worded very vaguely,” Waghre says. He believes that the current amendment’s policy of fining users for frivolous complaints or incorrect information is likely to make netizens reluctant to report fraud, insensitive content, or other privacy violations.
For all three cyberspace experts, the solution is a multi-layered one that cannot be implemented solely at the technical level. Jhalani urges law-enforcement agencies not to discriminate between online and offline harassment—the latter normally perceived as the more urgent concern. She says that the police’s attitude will often be dismissive. “They’ll tell you, ‘Do you want us to figure out a murder or do you want us to figure out an [unknown caller]?’” For Waghre, the harder problem to solve is that there are people in society actively looking to cheat or inflict violence on vulnerable users—a “downstream from broader societal issues like lack of trust and misogyny.” The government needs to roll out more digital-literacy campaigns, and matchmaking apps need to hire more women on their trust and safety teams, Jhalani believes.
Even the industry that was born as an antidote to matrimonial fraud in India—private detective agencies investigating pre- and post-marital fraud—has come under scrutiny for its own violations. Private detective agencies in general have seen a rise commensurate with online matrimony, with reports of a 30 percent annual growth in 2018. The organizations—often female led—use tools like high-resolution cameras, night-vision lenses, GPS trackers, call, text, and email records, and phone tapping to offer everything “from checking boy/girl’s job, affair, medical history, litigation cases, disability, debt, education, drinking habits to family’s social reputation, social standing, conduct etc,” according to the website for Sleuths India, a twenty-three-year-old private detective agency with more than 10,000 solved cases. The agency also offers post-marital or divorce-related services, including probing extramarital affairs and child custody issues. According to a 2015 Guardian report, Delhi alone is home to 3,500 such agencies.
In 2018, police in the city of Thane arrested four private detectives for the unlawful use and sale of the call records of targeted individuals. In 2022, a victim of one such agency filed a petition in the Delhi High Court for the review of the 2007 Private Detective Agencies (Regulation) Bill. The bill, which was pending before Parliament and was eventually withdrawn, asked for government bodies to supervise the currently self-regulated agencies. It proposed a government-mandated licence for the practice of private investigations and sought to “safeguard the rights of an individual to privacy and freedom.” But even this proposal came under scrutiny by experts. The bill “needs to be re-evaluated, exacting specifications on the powers of agencies, delineating powers of State and private authorities, and bringing amendments to central legislations to uphold the human right to privacy,” Nishtha Nikhil Gupta wrote in The Leaflet, adding that the current draft was “farcical and illusory.”
Even so, India’s data-regulation framework and the legal community at large might be able to catch up to the evolution of AI, albeit slowly—perhaps following the likes of Europe’s General Data Protection Regulation, often touted as the strongest privacy and security law in the world. The other issue is baked into the nature of AI: algorithmic bias.
Bias in AI algorithms is one of the most pressing concerns of scholars and technologists. For instance, few matrimony services acknowledge non-binary genders or sexuality preferences. Gupta of Betterhalf says that the team “didn’t have time in the initial years to think about that problem” but claims that discrimination on the grounds of gender, sexuality, or community is against their values as a company. He adds that offering matchmaking services for non-binary and LGBTQIA users is a possibility for the product at some point in the future.
For now, not only does Betterhalf exclude LGBTQIA people but it also makes predictions about users’ religion, caste, gender, and language based only on their first and last names. The patented technology, called “single-click matchmaking,” is an attempt to shorten matchmaking apps’ long onboarding processes that lead to a significant user dropout rate. Iyengar explains that this type of dangerous stereotyping can, ironically, limit users’ options by confining them to a certain, potentially inaccurate, demographic.
Other platforms like Shaadi and Jeevansathi have been called out for discrimination on the basis of caste and colour. With caste-based discrimination illegal in India, Shaadi’s algorithm was accused of bigotry in 2020, as it didn’t automatically include matches from “lower” castes for “higher”-caste individuals, unless they changed their settings to categorically include those communities in their results. The platform refused to accept responsibility and issued a statement that such a filter “works as an important proxy to determine lifestyle fitment.” The same year, one user started a petition to urge Shaadi to remove its filter for skin colour, arguing that it perpetuates colourism in the South Asian community. “The notion that fair skin makes a better bride/husband is still of significance. Whilst completely ignoring the personality, experience of life and the ability to make a good partner and son/daughter in law,” the petition read. The movement, which quickly gained traction, prompted the site to remove the filter.
Shaadi denied accountability. “Since we do not collect or capture this information on our platform, one cannot filter profiles using this. Hence, this search filter has no implications on matchmaking. That said, the search filter was blind spot from our side and we have removed it,” the company stated on Twitter in response to complaints. Jeevansathi soon followed, removing its own “complexion” filter in response to user backlash.
Traditional matchmaking still thrives in India. Sanjay Kirtania is one of the country’s most prominent traditional matchmakers, particularly in affluent Marwari communities. He grew up watching his father make matches and advise families. In 1996, shortly after his father retired, Kirtania founded Subhlagan, a matrimony service with offices in several major Indian cities.
The concept of dowry, or a version of it, still flourishes in traditional matchmaking methods, including Subhlagan. Many clients are interested in connecting with others in their financial bracket, and clients are sorted into ten to fifteen categories and matched based in part on their wealth and financial standing.
For Kirtania, Indian matchmaking is the art of understanding families. “Matchmaking is a deeply personal and physical task. It can’t be done arbitrarily. People should make eye contact and gauge body language to get a sense of their prospective partner,” he says. He scoffs and adds proudly, “Also, Shaadi.com’s founder Anupam Mittal—his cousin came to me to find a wife.”
“Relationship managers” at Subhlagan organize intense brainstorming sessions for the candidate and their family in their luxurious office space, where prospective profiles are projected on a big screen and every proposal elicits an in-depth group discussion.
Kirtania recounts an episode where the mother of a Mumbai-based matchmakee was looking for a partner for her daughter. “She was very adamant on the spouse being based in Mumbai,” he recalls. Kirtania had just received the biodata—a marriage résumé that typically features details like the prospective bride’s or groom’s name, age, height, profession, complexion, caste, horoscope, and family background—of a young man from Bengaluru who he intuited would be a good fit for the daughter. He asked the mother to consider the match, but she rejected the proposal based on his location. Kirtania persuaded her to send her daughter for one meeting with the man, who happened to be in Mumbai. “The kids ended up chatting for several hours, after which they eventually got married,” he says.
Kirtania’s example ties into what is called an “exploration-exploitation dilemma” in AI. Every artificially trained data aggregator has the choice to either exploit its existing data collection or to continue looking for new options, which in the context of matchmaking apps might mean generating new results outside of the user’s demographic zone. Human matchmakers are thought to be able to switch more easily and effectively between those choices. “AI can only ease this business, not complete it,” says Raman Sharma, a relationship manager at Subhlagan. “Every family has a unique set of questions driven by a unique set of concerns. Understanding that takes deep knowledge of our culture and the family and their background. It requires emotional intelligence, which AI cannot stand in for.”
Countless marriage brokers like Subhlagan—official and unofficial—exist across the country to uphold the centuries-old tradition of arranging marriages. Vows for Eternity, or VFE, is another curated, offline confidential matrimonial search service in India that caters to affluent families and cosmopolitan individuals, with basic membership costing $2,000 (US) per year. Founder and CEO Anuradha Gupta and her mostly-women team stress that they believe in matching couples “based on value systems, mindsets, and personality fits” rather than biodata. “We believe in arranged introductions, not arranged marriages,” says Gupta, who serves Indian clients around the world while being based in San Francisco. Gupta resists the checkbox methodology of matrimonial websites and traditional matchmakers and is credited with arranging 1,200 marriages in over twelve years of her service.
In contrast to what Betterhalf and the matrimonial websites are doing, VFE asks its clients to complete a 100-part questionnaire, compiled by psychologists, requiring the user to spend a significant amount of time, effort, and emotional energy. The survey is situation based and designed to provide an understanding of one’s value system and personality: “When you get hurt, how do you respond? Do you withdraw or do you mull over things? Do you need your time or do you have an outburst?” After this step, Gupta or a team member has a video call with the client to understand their childhood influences and previous relationships. For the highest-tier members, these calls can last four to five hours. “We act as a sounding board for our members. You can’t substitute these conversations with AI. Matchmaking is as much an art as it is a science,” Gupta says. “Everybody loves movies and travel and music and whatnot, but those things don’t form the basis of a marriage.”
Meanwhile, Namdev and Gupta’s vision of making Betterhalf a “marriage superapp” is slowly but surely coming to life. In March 2023, the duo raised $8.5 million (US) in a Series A round of funding, counting a number of tech-world heavyweights among their investors—Instagram co-founder Mike Krieger, Dropbox co-founder Arash Ferdowsi, former Bumble interim CMO Derek Callow, former Spotify San Francisco product head Brendan O’Driscoll, and former Uber CPO Manik Gupta, among others. During our interview, Gupta explained his five-year plan. He wants Betterhalf to be a one-stop shop that automates every pre- and post-matrimony service—from wedding catering, bridal makeup, and wedding photography to supporting a couple’s first pregnancy. “[We] want to own three to five years of somebody’s marriage journey, from pre-matchmaking to [the birth of] your first baby.”
