Regulatory sandboxes are often judged by visible activity: how many firms apply, how quickly products are tested, and whether new providers graduate. For inclusive insurance, the better test is whether a sandbox determines if an innovation can responsibly expand access for underserved customers, and if so, what would need to be in place for it to scale.
Inclusive Insurance poses particular challenges that standard regulatory frameworks weren’t designed to address (i.e., reaching customers in the informal economy via new distribution channels, unfamiliar business models, flexible premium payment options, and product approvals and claims payouts that need to be fast). These conditions can make it difficult for innovators to operate under existing rules — and it is therefore equally difficult for supervisors to assess risk without a controlled space in which to learn – this is where a dedicated sandbox comes into play.
Drawing on CGAP’s review of insurance sandbox approaches across several countries and discussions with supervisors through the CGAP–A2ii Learning Circle, seven lessons for insurance supervisors stand out. These span from setting learning goals and designing the test environment to managing exit, adaptation, and post-test impact.
1. Start with the learning goal and tailor the duration accordingly
Sandbox design should start with what the supervisor wants to learn. In inclusive insurance, that usually means allowing enough time to test the innovation, including setup, uptake, claims, operational viability, and whether the model can serve underserved customers responsibly. For example, Brazil adopted a 36-month period to give firms time to attract policyholders, experience claims, and show viability.
The point is to ask what the authority needs to learn, over what period, and for what purpose. Without that clarity, sandbox design risks becoming procedural rather than strategic.
2. Design based on the innovation the sandbox is meant to support
Sandbox design reflects policy intent. Open models prioritize flexibility: they allow a broad range of innovations to come forward and help supervisors observe what the market is developing. Thematic or targeted models start from a clearer policy priority – they focus the sandbox on specific gaps or objectives and help supervisors direct scarce attention toward innovations that can address them.
Country practice shows the trade-off. Open models, such as those used in Uganda, Brazil, Pakistan, and Mongolia, preserve flexibility and allow a wide range of proposals to emerge. More structured thematic approaches can help focus attention on inclusion gaps and underserved or informal-sector populations. Ghana, for example, adopted a thematic approach specifically focused on expanding access for the informal economy — roughly 80% of the population without insurance coverage
This leads to a more fundamental question than whether a sandbox is open or thematic: is the authority mainly trying to enable innovation in general, or is it using the sandbox to steer innovation toward specific public policy goals? The answer matters because it influences who applies, which proposals are prioritized, and how success should be judged.
3. Account for the support innovators need before testing
In most jurisdictions, sandbox costs are mainly staff time, with small teams coordinating across supervisory functions. There are, however, exceptions. Uganda, for example, links the sandbox to InsureX, a pre-sandbox incubation program that helps innovators build their capacity, refine ideas, understand the insurance sector rules, and build partnerships before testing.
Where applications are weak or poorly prepared, this kind of pre-sandbox support mechanism may be essential to build a viable pipeline, but it also incurs additional costs.
This suggests that the main question is not only how much it costs to run a sandbox, but also how much support innovators need before they are ready for testing.
4. Protect consumers without excluding the innovators the sandbox is meant to learn from
Authorities differ in the financial requirements they impose on sandbox participants, which may include application fees, minimum capital requirements, personal liability insurance, deposits, reinsurance, or proof of financial capacity. Some authorities require applicants to demonstrate financial capacity before testing; others apply capital requirements case by case, align them with lower-risk insurer categories, or rely on nominal fees and reinsurance depending on the scale of the test, or a combination thereof.
Financial requirements should be proportionate to the risks posed by the test and should protect consumers without excluding new and innovative types of providers, services, or models the sandbox is meant to assess and learn from.
These differences reflect distinct supervisory philosophies and market contexts. However, the underlying rationale is broadly similar: financial requirements should be proportionate to the risks posed by the test and should protect consumers without excluding new and innovative types of providers, services, or models the sandbox is meant to assess and learn from.
5. Plan for exit as carefully as entry
Exit strategies vary because sandbox participation can end in different ways – early termination, completion without market entry, or graduation into the regulated market. Termination may result from non-compliance, consumer harm, material changes to the business model, operational failures, voluntary withdrawal, or failure to meet financial or regulatory conditions.
South Africa offers a strong example of ex ante exit planning. Applicants must define upfront how testing would be wound down and how customers would be protected if the project does not transition.
This is particularly important for inclusive insurance. Low-income and underserved customers may have limited ability to understand what happens if a product is discontinued, an intermediary exits the market, or a test fails.
A good exit strategy does more than protect customers if a test fails. It helps supervisors turn sandbox learning into a clear pathway for what comes next, whether that means winding down a test safely, moving a participant into an existing licensing route, clarifying how the current framework applies, adapting requirements, or, where needed, designing a new framework that allows promising models to scale responsibly.
6. Manage the sandbox as an adaptive supervisory process
Once implementation begins, supervisors often discover issues that were not visible at the design stage. These may include coordination challenges in multi-regulator sandboxes, weak application pipelines, repeated ideas, or products that work in a sandbox but struggle to scale commercially.
An adaptive sandbox gives supervisors a way to refine requirements, clarify transition pathways, and identify where the existing framework is sufficient — or where it may need to evolve.
These issues matter because they show whether the problem lies in the innovation, the market, or the regulatory framework. An adaptive sandbox gives supervisors a way to refine requirements, clarify transition pathways, and identify where the existing framework is sufficient — or where it may need to evolve.
7. Judge success by what changes after testing
The most important measure of success is whether the sandbox generates learning, stronger coordination, and decisions that improve the market beyond the pilot itself.
Country experiences suggest that the strongest sandboxes create effects beyond the individual test. In Pakistan, sandbox learning supported a streamlined licensing regime for digital insurers and standalone microinsurers. Brazil illustrates a different pathway: the sandbox served as a supervised proof of concept for technology-driven insurers using on-demand, embedded, and hybrid products, digital onboarding, alternative data, and automated underwriting. Although the Brazil case did not point to a single regulatory clarification in the same way, it showed supervisors that these models could expand access safely, with 70–90% of policyholders reportedly first-time insurance users.
The broader lesson is that sandbox success should be judged by what supervisors learn and what that learning enables — clearer rules, stronger coordination, better supervisory capacity, or confirmation that the existing framework can support responsible innovation.
The value of a sandbox is not the protected space it creates, but the decisions it enables about which innovations deserve a path to scale — and under what conditions.
Add new comment