Last year, around 60 countries banned the use of smartphones in schools, including the Netherlands, France and the United States. The restrictions come amid widespread concern that social media can harm children’s mental health. It’s a valid concern. But amid the discussion, other technologies used by children, such as educational technology (edtech), fly under the radar.

As the director of the International Center for EdTech Impact, I believe this is a mistake. Edtech has the power to improve children’s learning, but it needs analysis.

Scientists are building giant “evidence banks” to create policies that actually work

The edtech industry is booming and is expected to be worth a staggering $600 billion globally by 2027. About 500,000 educational apps are available, from literacy apps to math games. In the United States, each school district used about 2,700 such tools last year. And many philanthropic funds are helping low- and middle-income countries (LMICs) roll out edtech in schools.

We’ve seen how educational apps have helped children learn – for example, they’ve improved math and reading skills in Malawi, reducing gender disparities in classrooms. And they are extremely necessary. About 70% of ten-year-old children in LMICs struggle to read and understand simple texts. And up to 70 percent of nine- and ten-year-olds in low-income areas of the United States cannot read at a basic level.

But with little regulation and few standards in place, I worry that governments, schools and families around the world are spending huge amounts of money on ineffective apps that only serve to teach children to play manipulative games. I want researchers, developers and investors to come together to improve the quality of educational tools.

A 2023 report from the United Nations UNESCO reveals that there is little hard data on the value of edtech tools (see go.nature.com/4hf8f7d). Applications are rarely verified or validated by researchers. And reviews of companies’ own products often focus on screen time, not how apps address educational gaps.

Governments should stipulate that apps cannot be sold as “educational” unless rigorous research has shown they improve children’s learning and well-being. National certification schemes based on independent research and catalogs of recommended resources would help teachers decide which apps to buy.

The Great Rewiring: Is Social Media Really Behind an Epidemic of Teen Mental Illness?

Collaboration is crucial. Simply relying on regulations would empower a few big players who have the financial capabilities to follow the rules. A collaborative approach is more likely to focus on children and stimulate innovation among start-ups and local entrepreneurs.

Shallow, unpaid advisory roles are a thing of the past – companies need dedicated specialists to effectively design, evaluate and innovate. So policy makers should provide funding to encourage partnerships. Finland and some US universities, for example, compensate scientists for spending time researching educational tools through university edtech accelerators. A European Union-funded project to bridge academia and the edtech industry is exploring how scientists, through secondments and mentoring, can drive innovative studies.

Incentives are also needed to improve data availability. Scientists need raw data about existing applications; companies collect this information but often keep it private. A common repository to which both scientists and firms can contribute has long been a dream for edtech researchers. The 2025 Toolkit – an international competition for edtech innovation led by a consortium of organizations – will include an award for technologies using open datasets. It is a good first step, but more such initiatives are needed.

Together, researchers and developers must define the educational outcomes each app aims to support—and determine how to measure them. About 30 educational apps are released every month and about 65 frameworks can be used to evaluate their effectiveness. Researchers struggle to compare apps that have been evaluated in distinct ways, and teachers and families can find the mass of data and disparate methods confusing.

Ready or not, AI is coming to science education—and students have opinions

Figuring out how to measure results will not be an easy task. Consider a seemingly simple story-reading app. Such an application will influence language development, reading motivation, critical thinking and more. All of these aspects of learning will be experienced differently by native and non-native speakers, people with learning disabilities, and so on. Randomized controlled trials—considered by some to be the gold standard for edtech evaluation—need to be supplemented with various measures of success that account for this variability. Efforts could involve holding focus groups with teachers and children, ongoing pedagogical evaluations, or following families over time to assess reading motivation, for example.

Taking steps towards collaboration will ensure that more tools are science-based. A bigger challenge will be to convince companies that research is a valuable learning process, not a service to confirm positive impact and drive sales.

With well-designed tools, edtech could improve the prospects of many children. If nothing is done, poorly made tools will reinforce educational gaps by disrupting learning, skewing student achievement, and promoting investments in “snake oil” rather than education.