What tech can learn from the social sector

August 31, 2023
Temina Madon and Robert On

Since our earliest days, the Agency Fund has partnered with South Park Commons to sponsor a fellowship for social entrepreneurs. Why would we, a nascent anti-poverty movement, partner with a Silicon Valley tech community? As we’ve written before, we think technology plays a role in scaling complex psychosocial interventions – by enabling targeting, personalization, and experimentation at scale. But we want to build technologies that have a positive impact on society, and that's hard to get right. 

So, we have spent some time learning from the tech sector. To hedge against building a product that no one wants, we focus obsessively on metrics of engagement – you can’t have a successful product if people don’t use it. Like startup founders, we also believe in never-ending experimentation and product iteration… Because you can always do better. 

Yet the tech sector tends to focus on shorter-term outcomes – like website views, clicks, and ad revenue. At the Agency Fund, we care about the longer-term impacts of new technologies. We want to learn how an app or messaging service affects people’s health, safety, or prosperity. 

The social sector is deeply focused on these distant, but life-changing, outcomes. Organizations like Rocket Learning and Shamiri Institute create campaigns and services that shape people’s life trajectories, in measurable ways. They view technology as an instrument for improving people's fortunes. But the bar is high: they need to measure and document the change. 

Fortunately, over the last few years, scientists have become adept at rigorously measuring the impacts of social welfare programs. In fact the 2019 Nobel Prize in Economics was awarded to 3 researchers who have used field experiments to isolate the causal effects of different programs. We think that technologists can learn something useful from this: about how to measure the real-world impacts of an innovation... and how to improve the longer-term welfare of their customers. 

Here we outline a few ideas from the social sector that tech founders might want to explore:

Idea #1: Engagement with a service is only one piece of real-world impact.

The tech sector has built sophisticated systems to analyze the interactions between users and digital products. There’s an entire ecosystem of web analytics platforms that will automate your user metrics, cohort analyses, and A/B testing flows. But this success has also created a blindspot for some tech companies: they now have the mentality that measuring anything “offline” is not scalable. 

The reality is that real-world, offline behavior is increasingly more observable, at massive scale. The tech sector may not routinely measure it, but academic researchers do. We have validated instruments for tracking welfare and quality of life using surveys, satellites, mobile data, and photos. For companies that want to make lives better, there are now reliable tools for measuring relevant outcomes. 

For example, the tech non-profit GiveDirectly gives out cash to low-income households. They routinely run randomized experiments, in which they compare the life outcomes of their users to the outcomes of randomly assigned control groups. They’ve looked at a range of outcomes, from  household consumption and assets, to family members’ health and school success. They’ve even used satellite imagery to monitor changes in people's housing quality. Tracking a control group increases their costs of data collection, but it also helps drive conviction that they’re having a positive impact in the world. 

GiveDirectly also looks at the long run: they have followed a panel of their users over 10 years, tracking whether participating households have benefitted materially or emotionally from the services provided. They make follow-up calls with every recipient, to ask about their experiences. Honestly, I can’t imagine Instagram doing this with its users (though it’s an interesting thought experiment!). 

Recipients from the GiveDirectly Project 100 initiative. You can view stories from Project 100 recipients online. 

Giving tech founders a more accurate view of their customer’s life experiences could help with product design and targeting. It could inspire new kinds of services that operate across the virtual and physical worlds – creating a more meaningful metaverse. It might even uncover ways to serve lower-income users, while still making a profit. 

The point is that the world is bigger than what you find in online transactions. Data collected in the virtual world should be complemented with real-world outcomes. Does user engagement actually correlate with user well-being, on the timescale of a monthly or annual subscription? Let’s start observing, measuring, and learning!

A caveat, of course, is that few people will trust internal company research on a product’s welfare benefits—so you may want to find an academic partner to help with data collection and analysis. And, longer-term studies take time. But your patience may be rewarded with the deep insights offered by real-world data. 

Idea #2: For most people, their time is their greatest asset. Let’s figure out how to value it. 

In an increasingly unequal economy, most people don’t have enough money to pay for their digital consumption. So tech companies auction off consumer time and attention to advertisers. They provide services that they claim are valuable to users—simply because people spend time on them. But time spent consuming ads does not necessarily accrue value to the end user. 

Still, we see lots of companies claim that use of a product or service means that the consumer values the service. Time spent “on platform” is used to justify impact: after all, the user has control over how they spend their time. An issue with this argument is that many people actually want alternatives. They want services that value their time, but are still engaging or useful. A mainstream example is BeReal, a social app that provides users with more authentic interactions than its competitors. A more relevant example is SameSame, which offers free, anonymous mental health support to LGBTQ+ youth. Their chatbot is entertaining and design-forward – but they use clinically validated assessments (with consent) to ensure that engaging with the product leaves users less depressed and anxious.

Of course there are risks to giving people exactly “what they want”. Most recommender systems – whether for e-commerce or news feeds – are designed to optimize for user attention or revenue, not for well-being or social welfare. As a result, these algorithms tend to amplify the status quo. They reinforce existing beliefs and preferences, rather than pointing us to new world views. Why not help people challenge their priors and explore novel perspectives, with tools like bridging-based ranking (developed by a member of South Park Commons)? What if we had to prove that our products actually engaged people's conscience, or promoted social cohesion? 

If we are willing to revalue people’s time and attention, and offer them new world views, we might be able to avoid the popular backlash and regulation that often halt innovation. 

Idea #3: Emerging markets hold ample opportunities to do good and do well. 

With all the software and measurement innovation that’s happened over the last decade, we think there are rising costs (and diminishing returns) to further innovation in advanced markets like the US, Europe, and East Asia. Talented founders might find more impact and success building in emerging and developing markets. Why? Perhaps for the same reason that today’s top economists study development economics: because there are more people in need, with more complex problems – and therefore more opportunities to do good.

We are not advocating for a new wave of extractive corporations that prey on less advantaged communities. But there are fast-growing (and young) populations in most low- and middle-income countries. These economies are steadily growing, and there is thirst for innovation. Generative technologies, like large language models (LLMs), will allow us to deliver complex, life-changing services to greater numbers of people, with unimagined levels of impact. For ambitious technologists, why get stuck around a local optimum? There are vast, under-served markets that can generate oversized benefits. Here are a few examples:

  1. In many developing countries, the delivery of public services is weak (think education, healthcare, and even tax collection). This creates opportunities for innovative govtech and civic tech, as well as private sector alternatives to public services.  
  2. Governments and non-profits transfer billions of dollars in cash to households in need each year. But few countries have efficient welfare systems; in many cases, payments are still made via paper checks or cash stuffed in envelopes (fintech, anyone?). Governments also need more effective ways to target the neediest. 
  3. Informal markets are the dominant economic force in developing countries. They’re characterized by lots of small-scale entrepreneurs, and relatively few large firms. As a result, they’re more difficult to organize, and they tend to be inefficient. E-commerce, payments, logistics, and business management software could bring greater performance and profitability to small- and medium-scale enterprises (MSMEs). At scale, this can mean massive profits.
  4. It remains difficult for people in informal markets to find jobs… And there are too few jobs in the first place, because firms are so small. Can generative AI play a role in job creation or entrepreneurship? Can it reduce the frictions in new firm formation, marketing, and management?

Across Africa, Latin America, and South/Southeast Asia, there are early-stage funds, like the Digital Financial Services Lab and the Catalyst Fund, that are eager to provide both capital and expertise to entrepreneurs building for underserved markets. As always, there is also a caveat: if you are not building for people just like you, it is important to have a strategy for getting quality user feedback as you iterate. You need to embed yourself in the problems that you want to solve. You need to live in the communities you want to serve. You may need a co-founder who has deep experience in the markets you’re exploring. Your solutions need to adapt to local market structure, geography, and culture. 

But if you want to change the world, remember that non-profits and academic economists have been innovating for the last 25 years, right alongside the tech industry. And we think there are some valuable lessons for tech entrepreneurs, if you take the time to look closer.