Last updated: May 2026
BeeSwarmly generates swarm risk assessments using statistical models trained on historical sensor data from thousands of hive-seasons. These predictions are probabilistic estimates, not certainties. A high-confidence swarm alert means the model has identified patterns consistent with pre-swarm behavior — it does not guarantee that a swarm will occur. Similarly, the absence of an alert does not guarantee that a swarm will not occur.
Our published accuracy rate of 91% is based on controlled field validations across multiple climates, seasons, and hive configurations. Actual accuracy in your specific environment may vary based on factors including colony genetics, local forage conditions, hive management practices, sensor placement, environmental interference, and data transmission reliability. The accuracy rate reflects performance under optimal conditions and should be interpreted as a benchmark, not a guarantee.
Sensor readings are subject to hardware tolerances, environmental noise, and transmission artifacts. Temperature measurements are accurate to ±0.5°C under normal operating conditions. Weight measurements are accurate to ±0.01kg on level surfaces. Acoustic analysis depends on ambient noise conditions and may be less reliable in high-wind or high-traffic environments. Sensor performance may degrade over time due to propolis buildup, moisture exposure, or battery degradation.
BeeSwarmly is a decision-support tool designed to complement — not replace — regular hive inspections and professional beekeeping judgment. Sensor data and predictions should be used alongside visual inspections, local knowledge, and established apicultural best practices. Do not rely solely on BeeSwarmly alerts to make colony management decisions.
Colony behavior is influenced by many factors that fall outside the scope of hive-mounted sensors, including nearby pesticide application, predator presence, disease vectors, queen quality, and regional nectar flows. BeeSwarmly cannot account for all external variables that may affect swarm timing or colony health.
Our prediction models are periodically updated as new training data becomes available. Model updates may change the sensitivity, specificity, or lead time of swarm alerts. We communicate significant model changes through platform notifications and email updates. Historical predictions are not retroactively adjusted when models are updated.
Sensor data is transmitted at regular intervals (default: every 15 minutes). Network conditions, gateway availability, and server processing times may introduce latency. Dashboard readings may not reflect real-time conditions. In critical situations, direct hive inspection is always recommended.
BeeSwarmly assumes no liability for colony losses, missed swarms, false alerts, economic damages, or any other harm arising from the use of or reliance on our predictions, sensor data, or platform. By using the Service, you acknowledge the inherent limitations of predictive modeling and accept full responsibility for colony management decisions. For complete liability terms, see our Terms of Service.
If you have questions about data accuracy, model methodology, or the limitations described in this disclaimer, contact us at app@beeswarmly.com. We are committed to transparency about what our technology can and cannot do.