Being blinded by the status quo or brand power can trick us into believing that we are stuck, but things are rarely as permanent as they appear. If a new coach uses an old playbook, she may end up at the bottom of the log. We may think that Google will last forever, until they don’t.
People who care enough develop their own playbooks. Look at Rassie Erasmus. He took the rugby rulebook, studied it, and created his own playbook. The rest is history.
A rule book is a cage. A playbook is a compass. Rule books demand compliance. Playbooks invite contribution.
A rule book feels safe. It promises certainty. It whispers, “Do this. Don’t do that. Follow me, and everything will turn out fine.” In the industry, rulebooks are often written after success. They are the artefacts of nostalgia—the fossilised remains of what worked yesterday.
Playbooks aren’t documents from a single source or company, but frequently codified strategic approaches broadly adopted across industries.
The idea behind this blog is not to create a new rulebook or playbook, but to provide coaches, captains and players in the Earth observation (EO) industry with guiding questions to help them write their own playbooks.
1) Mission and value: defining the supreme problem
This is the foundational question of any Earth observation strategy because everything—data choices, architecture, partnerships, pricing, service-level agreements (SLAs), infrastructure, and even AI governance flows from it. Without clarity here, the entire EO system becomes a collection of capabilities rather than a coherent solution.
- Which decisions (not datasets) will our customers make with our platform in the loop?
- For each decision, what latency and assurance (P50/P95 time to decision, accuracy, uptime) are truly required?
- What’s our thin wedge outcome (e.g., building level flood depth, dark vessel risk) that we can deliver with SLAs in <120 days?
- How will we price outcomes (per area of interest (AOI), per event, per decision) rather than per scene?
- What is our North Star metric (e.g., cost per actionable, time to decision) and which submetrics roll into it?
2) Data stewardship: prioritising trust over volume
This section is essential because, in Earth observation, data volume is no longer the bottleneck, trust is. Modern EO systems must be designed around reliability, provenance, interoperability, and customer assurance rather than raw imagery.
- Which modalities (optical, SAR, hyperspectral, multispectral, thermal, RF) and temporal cadences do we need to guarantee the outcomes?
- How do we enforce pre-validation (georegistration, cloud/ice masks, radiometric QC) before anything enters the catalogue?
- Do we ship analysis-ready data by default (COG, STAC, OGC APIs), with lineage and cryptographic provenance embedded?
- What’s our policy for chain of custody (hashes, signatures, attestations) from sensor to user?
- How do we handle licensing/data rights (sovereignty, redistribution, retention), and can customers “lift & shift” their data if they leave?
- What is our stance on synthetic data (when permitted, how labelled, how governed)?
3) Architecture: API-first and system interoperability
This is one of the most critical pillars an EO playbook, as it determines whether your EO system becomes a flexible decision engine or a brittle data-delivery product. Your decision will determine interoperability and architecture as prerequisites for reliability, resilience, and customer trust.
- Are our interfaces API first and standards aligned (STAC/OGC) to avoid lock-in for customers and for us?
- Do we deliver programmable subscriptions with SLAs (latency, revisit, denial resilience), not just file downloads?
- What’s our event model (alerts, digests, features) and how does it integrate with customer systems (GIS, mission tools, data warehouses)?
- Which artefacts do we deliver to the right runtime-tiles for GIS, feature vectors for ML, event digests for operations?
- Can the platform run multi-cloud/on-prem/air gapped, and what’s our data egress story?
4) Edge computing: processing data at the source
This pillar is essential because modern Earth Observation systems cannot meet customer expectations; speed, bandwidth efficiency, resilience, and assured outcomes; if all processing occurs on the ground. Your playbook must make this explicit by repeatedly asking which parts of the pipeline must run on-orbit or at the edge to meet latency, bandwidth, and SLA requirements.
- Which parts of the pipeline run on orbit or at the edge (pre-processing, triage, inference) to reduce latency and bandwidth?
- What is our edge packaging (containers, model bundles, A/B updates, rollback) and how do we validate it for radiation/clock skew/faults?
- In degraded comms scenarios, what’s the minimum viable event packet we can deliver to keep operations going?
- How do we prioritize downlink (events > features > imagery) under constrained bandwidth?
5) Sovereignty and security: policy-encoded plumbing
This pillar exists because trust is now the most valuable—and most regulated—resource in Earth Observation. It is no longer sufficient to secure systems at the surface; security and compliance must be embedded, automated, measurable, and enforceable at every layer of the EO pipeline.
- Do we implement zero trust by default (identity, encryption, least privilege) across tasking, storage, and delivery?
- Can customers set sovereign controls: data residency, key ownership, classification/redaction, and policy-based routing?
- What export control/sanctions checks are embedded in tasking and access workflows?
- How do we audit every decision artefact (who saw what, when, and why) without breaking performance?
6) Operational resilience: engineering for the worst day
This pillar exists because Earth observation systems are not judged on how they perform on a normal day; they are judged on how they perform on the worst possible day, when multiple failures, spikes in demand, or external disruptions occur simultaneously.
- What are our P95 latency and uptime targets during peak events, and how do we measure and publish them?
- What is our multi-orbit connectivity plan (LEO/MEO/GEO/surface networks) and failover under jamming or outages?
- Do we exercise space weather and debris playbooks (autonomous avoidance, throttled pipelines, safe mode inference)?
- What’s our MTTR for pipeline incidents, and do we run chaos/resilience drills quarterly?
- How do we guarantee continuity of service if a primary sensor or provider goes offline (multi-vendor, multi-modal)?
7) AI governance: explainability and verification
This pillar is essential because artificial intelligence now sits at the centre of modern Earth observation; from detection to prediction to decision support. Without strong governance, AI becomes a black box, making EO outputs untrusted, unauditable, and unusable in high-stakes environments.
- For each model, what is the model card, training data lineage, and intended use/limits?
- How do we monitor drift (data, concept, performance) and trigger human-in-the-loop reviews for high-stakes outputs?
- Can customers see confidence intervals and evidence panels (inputs, features, residuals) with every decision?
- What is our benchmark suite (shared test sets, third-party validations) and how often do we re-certify?
- How are bias and fairness evaluated for civilian vs. defence use cases?
8) Ethics and compliance: your license to operate
This pillar exists because Earth observation is no longer a niche technical field; it is an industry closely intertwined with regulation, geopolitics, environmental stewardship, dual-use risks, and societal impact.
- Do our outputs meet audit grade standards for ESG, insurance, sanctions, or legal proceedings?
- What is our dual use policy (misuse prevention, access tiers, transparency reports)?
- How do we mitigate astronomy interference (orbit choices, brightness mitigation, scheduling) and demonstrate sustainability (passivation, deorbit)
- How do we communicate limitations and uncertainties so customers don’t over-trust the outputs?
9) Commercial ecosystems: durable and independent
This pillar exists because no Earth observation provider (regardless of size, modality, or funding) can win alone. The EO value chain is now too complex, interconnected, and volatile for any single organisation to control end-to-end. Partnerships and commercial structures determine whether your ecosystem becomes scalable and resilient or fragile and dependent.
- Where do we buy vs. build (sensors, ground, analytics, edge compute), and what are the switching costs?
- Which suppliers carry systemic risk (single points of failure), and what redundancy is in place?
- Are supplier contracts outcome aligned (latency, revisit, denial resilience) and survivable (data escrow, termination assistance)?
- Do we support a marketplace of third-party analytics with transparent leaderboards and revenue sharing?
10) Culture and process: repeatable excellence
This pillar matters because even the best architecture, sensors, AI models, and partnerships fail without the right people, operational processes, and culture to sustain them.
- Do we have SRE for data (observability, runbooks, error budgets) and mission control for live operations?
- Are product, science, and ops aligned on SLA definitions and post-incident reviews?
- Is documentation (APIs, schemas, governance) public by default where possible?
- How do we train customer teams on interpretation and escalation to prevent misuse?
11) Review cadence: measuring what matters for growth
This pillar is essential because Earth observation systems are complex, multimodal, high-stakes infrastructures. Without disciplined measurement and recurring review cycles, even the strongest architectures, the best AI models, and the most resilient operations will drift, degrade, or misalign with customer needs over time.
- Which leading indicators predict customer value (tasking to event time, % edge validated, model drift rate)?
- Which financials matter per decision (gross margin per actionable, cost of data per outcome, support load per alert)?
- How often do we run independent audits (security, privacy, model performance, provenance) and share summaries with customers?
Conclusion
Write your own playbook; the gains are specific and compounding across every corner of EO.
- Satellite operators translate “pixels” into SLA-backed outcomes (latency, revisit, uptime) that de-risk revenue and justify premium pricing when a primary sensor goes dark.
- Data platforms and aggregators turn catalogues into programmable subscriptions with provenance and sovereignty controls, reducing churn while increasing ARPU through event-driven delivery.
- Analytics and AI providers ship explainable, benchmarked models with drift monitoring and evidence panels, unlocking regulated markets and outcome-based contracts.
- Edge and on-orbit teams prioritise compute-to-data pipelines, winning on bandwidth, time-to-decision, and resilience under degraded comms.
- System integrators and marketplace curators formalise onboarding, leaderboards, and revenue-share rules, scaling partner ecosystems without sacrificing trust.
- Solution providers (public sector, insurance, energy, agri-water) codify contracts for auditable results—so decisions, not scenes, become the unit of value
Put simply: your playbook ties decisions, SLAs, provenance, and resilience into one operating system for growth—and that’s how each provider moves from data exhaust to indispensable, on-the-worst-day decisions.
| Abbreviation | Expansion | Context / Meaning |
| AI | Artificial Intelligence | Models and inference pipelines used to derive EO outcomes; governed via model cards, drift monitoring, and explainability. |
| AI/ML | Artificial Intelligence / Machine Learning | Collective reference to analytics models powering detections, classifications, and decision support. |
| AOI | Area of Interest | Spatial unit for tasking, pricing, and outcome delivery (e.g., event-based billing per AOI). |
| API | Application Programming Interface | Programmable access to data/events; supports subscriptions with SLAs and standards-aligned schemas. |
| ARPU | Average Revenue Per User | Commercial metric: increases via event-driven delivery and programmable subscriptions. |
| COG | Cloud‑Optimised GeoTIFF | Analysis-ready raster format enabling efficient, cloud‑native access to imagery. |
| EO | Earth Observation | End-to-end industry focus of the playbook (from sensors to actionable decisions). |
| ESG | Environmental, Social & Governance | Audit‑grade outputs and compliance requirements for regulated use cases. |
| GEO | Geostationary Earth Orbit | Part of multi‑orbit connectivity/failover strategy alongside LEO/MEO. |
| GIS | Geographic Information System | Consumer runtime for tiles/layers; one of the target artefacts for delivery. |
| LEO | Low Earth Orbit | Primary orbit for many EO constellations; included in multi-orbit resilience planning. |
| MEO | Medium Earth Orbit | Complementary orbit layer in multi-path connectivity and dissemination planning. |
| ML | Machine Learning | Model-driven analytics (e.g., classification, inference) feeding features/events to users. |
| MTTR | Mean Time To Recovery/Repair | Operational resilience metric for pipeline incidents and restoration targets. |
| OGC | Open Geospatial Consortium | Interoperability standards (with STAC/COG) for analysis-ready data and APIs. |
| P50 / P95 | 50th / 95th Percentile | Assurance metrics for time‑to‑decision, latency, and accuracy SLAs. |
| QC | Quality Control | Pre-validation gates (geo‑registration, cloud/ice masks, radiometry) before data enters catalogues. |
| RF | Radio Frequency | One sensing modality is considered for outcome assurance. |
| SAR | Synthetic Aperture Radar | All-weather sensing modality used in multi-modal EO outcomes. |
| SLA / SLAs | Service Level Agreement(s) | Contracted performance targets (latency, revisit, uptime, assurance tiers) tied to outcomes. |
| SRE | Site Reliability Engineering | Practices for data/ops observability, error budgets, and mission control. |
| STAC | SpatioTemporal Asset Catalog | Open spec for indexing/discovering EO data; core to standards-aligned delivery. |
