With the ever-growing number of disparate, siloed, black-box, and bottlenecked digital and data tools that are on the market offering farmers a plethora of proposed opportunities for ROI and savings through disparate Intelligence, Insight, and Interface subscriptions, the on-farm decision-makers need a better roadmap to understand the quality and usefulness of their data and how it can be leveraged to maximum benefit through independent business intelligence; why it’s used, of what quality, and what outcomes back to their farm level it has.
Data ownership is separate from the current generation of on-farm tools producing data. As we focus on animal welfare equitable and sustainable livestock food systems there will need to be accountability over the quality of data produced by farms’ digital tools, and when farms are required to meet regulatory or audit standards, we need to be able to differentiate between a farm-level and a tech data issue or discrepancy.
Within the current volatile economic climate, farmers are operating to where each cent is assigned to a cost while recognizing there isn’t an infinite milk check to cover the ever-growing number of subscriptions for the disparate digital tools and products available to farms, which are not providing or enabling adequate data ownership and integration models, that enhance flexible independent intelligence for the on-farm decision-makers and their teams.
A model in which farmers own their data pre- and post-algorithmic processing holds new value for farms if utilized appropriately, for farm-independent business intelligence. To better understand the value of the data, we have previously compared the data with rare-earth minerals, for which their market value depends on extraction, refinement, and transformation. Likewise, the value of an individual point of data is low but having a system in place to use millions or even trillions of data points has significant value, both on and off the farm—data and analytics form part of the value chain beginning with the farm business while being highly desired by others within the sector: AgriTech, academia, and supply chain, for research, product development and new value.
Farmers should not be expected to hand their data over to their vendors for “pizza and a beer” (Cooley, 2022) for intelligence, where others derive value from the farm’s anonymized data, which we need to pivot from and stop with this current generation of AgriTech tools that are on the market. Farmers and farming are no different from other digitized sectors in taking maximum advantage of their data assets. The risk is someone else having their data assets that creates a bottleneck for developing new insights or value from their farm’s data. Farmers own their data pre- and post-algorithmic processing, where most farmers and non-farmers agree that the farmer owns the raw data produced on a farm (Fadul-Pacheco et al, 2022), so new business models and infrastructure must be offered and provided to this new asset on the farm.
With the rise of the “citizen data scientist” and engineers near or at the farm, with their ability to use the data for their reporting or management needs at the farm level, without third-party support, where standardized and agile user experience can be deployed, are important as we transition to the big data and internet of things (IoT) on the farm. Where citizen data scientists, should be able to use their own tests and dataset to develop their own Artificial Intelligence (AI) and Machine Learning (ML), tweaking them to their own livestock food system, even if that is low code. Right now, farmers and data owners need to look at how to best integrate the highest value tools and data points to give the best benefit in business intelligence, for example, reduce herd turnover rates, reduce calf mortality, and age at first calving, while understanding the limitations and bottlenecks with AI/ ML black boxes, so that as the data owner you have clarity on where you are getting your on-farm data, both from data quality and analytical standpoints.
Data quality, mapped editing, the removal of data bottlenecks, and current black box algorithms and data tools are critical in effective digital adoption, trust, and literacy, which all impact the value of the data and its value chain on-farm. Disparate, un-consolidated data, with different user experiences at the farm level that itself becomes a bottleneck to rapid adoption and tackling some of the livestock sector’s largest challenges, show greater ROI, transparency, accountability, and understanding of sustainability and welfare in food livestock farming systems. Recent research indicates that only 14% of 129 wearable sensors on the market have been validated (Styger et al, 2021). Suppose we don’t know what the data quality is, while it is being maintained disparately without passing into a farm-central intelligence platform. In that case, this will hinder getting too sensitive and specific ML and AI of value (Garbage in = Garbage out). What are the raw data, and varieties, what volume is currently achievable, and what is aspirational will become a more important part of the data value chain on and off the farm?
Farms must have the ability for independent business intelligence while having access and ability to choose more complex and specific ML and AI, that has been extensively validated, shows an ROI, and can be supplied by the service sector from their areas of specific expertise (nutrition, health, etc.), this allows for a competitive market on the intelligence, insight interface products that will benefit farmers, while regionality and farm style can be taken into account in the algorithm design, and that eliminates farmer lock-in into digital products. The greater validation of AI and ML that is applied to farm data, with transparency will generate greater ROI for farms while allowing new business value to emerge for farmers and the service sector.
The AgriTech sector, including farmers, needs to come together and define a set of standards on data quality and output for critical welfare and sustainability measurement and management metrics from AgriTech tools, which we don’t currently have. Accountability with data is a priority and must be defined as whether it’s a farm issue or a technology issue, which will become more important in the future for regulatory, audits and meeting contractual standards that add value to the farm business and throughout the supply chain.
A Common Data Framework for Livestock (CDF-L), between AgriTech providers and the tools they supply that represents the data, the schema, and structure of the data, as well as characteristics and definition of the data while carrying the metadata with the data through its transition and use by handling data formats like JSON, Parquet, ORC, AVRO, XML, CSV, etc. Historically, the work to build an app has been tightly tied with data integration, but with CDF-L and the platforms and applications that support it or something similar, the two can happen independently, where App makers and/or developers whether these users leverage code-based platforms or a low-code/no-code platforms such as Power Apps or Power BI, they need to store and manage data for their apps, while a CDF-L also allows data integrators as users are responsible for bringing data from a variety of systems to make it accessible for apps to use and data ownership to be maintained and governed. This approach aids in how we use data from the farm to inform the consumer and build trust with where their goods come from, this is an important part of separating insights from the data. The better the quality and volume of data the farm has the better and more advanced AI and ML can be applied, generating value and a valuable dataset that supports the intended message, welfare, and sustainability in livestock food systems.
With the need for new AgriTech business models, enabling farm data ownership, pre-, and post-algorithmic processing, and CDF-L, will give the farms and their teams access to independent business intelligence. A central data infrastructure needs to enable agile and continuously evolving intelligence, insight, and interface products in all the different areas of expertise that are needed on and off the farm, taking to account the many variabilities on each farm, that give regionally agnostic tools, that can be more specific to US dairy farms, then the generally applied AgriTech data tools on the market today.
The greater emergence of citizen data scientists and engineers on the farm will increase the deployment and development of new ML and AI that can be applied to their own centralized datasets, eliminate bottlenecks, and black-box algorithms, and be able to assess and have accountability on data flows. This emergence will also increase the need for a central data infrastructure to manage farm data as an asset and assess its quality, and will enable new independent business intelligence to emerge, driving new value to the farm owner.
- Cooley, 2022, Don’t Sell Your Data for a Pizza and a Beer. Farm & Ranch Data Ecosystems – AgriGates.io (agproud.com)
- Styger et al, 2021, Frontiers | A Systematic Review on Commercially Available and Validated Sensor Technologies for Welfare Assessment of Dairy Cattle (frontiersin.org)
- Fadul-Pacheco et al, 2022 Animals | Free Full-Text | Addressing Data Bottlenecks in the Dairy Farm Industry (mdpi.com)
- Data as an Asset for Livestock Farming – AgriGates.io
- Farm & Ranch Data Consolidation – AgriGates.io
- Farmers Own Their Data! – AgriGates.io
- Farm & Ranch Data Ecosystems – AgriGates.io
- A Common Data Framework for Livestock Agriculture – AgriGates.io