search-icon

How to Procure Earth Observation and Machine Learning Services Effectively

As satellite data and machine learning (ML) become increasingly central to tackling region-specific challenges—such as monitoring forests, tracking urban expansion, or responding to natural disasters—governments are turning to custom Earth Observation (EO) solutions. However, many of these services are not off-the-shelf products but bespoke systems built to solve unique problems. 

To ensure successful outcomes, public sector procurement agencies must adapt their approaches. Below are seven key strategies to help guide more effective and sustainable procurements in this space. 

KappaZeta has been providing EO-based remote sensing services to government agencies since 2015. One of our first projects involved the Estonian Agricultural Paying Agency, which commissioned a nationwide service to monitor mowing events. It was a bold step, requiring openness to innovation, as Estonia became the first country to conduct subsidy checks using remote sensing.

Following Estonia’s lead, Denmark and the Netherlands adopted similar approaches—further validating the effectiveness of EO-based services. This momentum contributed to a major shift in the EU’s Common Agricultural Policy (CAP): moving from physical checks on 5% of agricultural fields to monitoring 100%—a change made possible only through remote sensing. 

Based on our experience with public procurement of remote sensing services, we’ve compiled a set of key ideas to help government agencies run more effective and sustainable procurements in this field. The main objective is to reduce government expenditure and create long-term efficiency from using remote sensing as a key enabler for services, that would otherwise be done manually. 

1. Involve Earth Observation Experts Early 

Custom earth observation services depend heavily on what is technically feasible with current satellite data. Involving in-house remote sensing specialists—or consulting private-sector experts—helps define realistic expectations and budgets. This early input ensures the service you are requesting can actually be built using today’s data sources and technologies, helping avoid costly missteps later in the project.

2. Procure Services, Not Just the Final Product 

Purchasing the full product and intellectual property (IP) may seem appealing, but it often leads to hidden burdens. Maintaining an EO/ML service in-house requires rare and specialised expertise, from satellite data experts to ML developers, along with infrastructure for data processing and model operations. Without dedicated resources, the product risks being shelved. 

Instead, procure the service itself. Earth observation companies often serve multiple clients using optimised pipelines, enabling lower costs and higher quality through scale. A service-based model also allows these providers to export their expertise, strengthening the local economy. 

Photo by Nisuda Nirmantha on Unsplash
Photo by Nisuda Nirmantha on Unsplash

3. Prioritise Quality with Balanced Evaluation Criteria 

A procurement scored primarily on price will attract low-cost, low-quality bids. This approach is especially risky with earth observation and ML, where quality depends on both data science and domain expertise. 
To avoid disappointment: 

  • Set clear evaluation criteria including the bidder’s relevant experience, team qualifications, and proposed methodology. 
  • Limit the weight of price to no more than 60% of the total score. 

This encourages serious, capable providers to compete and helps ensure the service meets your operational needs. 

4. Identify and Engage Trusted Suppliers 

Look for suppliers with a proven track record, strong compliance culture, and clear communication practices. These partners invest in long-term trust and efficiency—traits that lead to better results and smoother collaboration. 

Reputable earth observation companies are often willing to share their insights early in the process. Engaging them during pre-procurement stages helps shape stronger tenders and builds mutual understanding of needs and capabilities. 

5. Clearly Define the Service’s Purpose and Integration 

Vague procurement documents lead to mismatched expectations. Be explicit about: 

  • What the service will be used for 
  • How it fits into your workflows 
  • What kind of outputs are needed 

This enables bidders to tailor their proposals, offer technically sound solutions, and provide more realistic pricing. It also allows you to assess proposals more effectively and compare them on the right basis.

Read also: Detecting Tillage Intensity from Space

6. Reduce Vendor Lock-In Through Market Awareness and Interoperability

Concerns about vendor lock-in are valid, especially when services become operationally critical. However, the EO and ML market has matured significantly, with a growing number of globally active providers offering comparable capabilities. 

To reduce dependency risks: 

  • Design services around open standards and documented interfaces, so they can be transitioned or re-implemented by other suppliers if needed. 
  • Stay informed about the vendor landscape—there are dozens of reputable EO service providers worldwide with overlapping expertise. 
  • Prioritise providers who demonstrate transparency, interoperability, and a willingness to support transitions if required. 

Vendor lock-in is less of a threat when your service is built on transferable foundations and backed by a competitive supplier ecosystem. 

7. Build Flexibility and Iteration into the Contract

Custom EO and ML services rarely follow a straight path from concept to completion. As data is explored and real-world needs are better understood, small adjustments are often necessary. 
To support this adaptive process: 

  • Structure contracts in clear phases—such as prototyping, piloting, and full delivery. 
  • Allow for minor scope changes, testing periods, and feedback loops without triggering re-tendering. This flexibility enables innovation while maintaining accountability, ensuring the final service is both technically sound and operationally useful. 

Conclusion

Procuring EO and ML services is not business as usual. These are high-value, high-complexity projects that require clarity, collaboration, and commitment to quality. By focusing on services over products, engaging experts, and setting thoughtful evaluation criteria, public agencies can unlock more successful outcomes and help grow a robust ecosystem of EO innovation. 


Considering launching a procurement for your Earth Observation need? Get in touch with KappaZeta’s leading industry experts who have years of experience working with a number of countries and procurement models. Get in touch with us on kappazeta.ee

RELATED