If a technology improves efficiency and saves money, most business people would likely adopt that technology should its benefits be made clear.
In the case of agricultural big data, however, some experts see the inability of developers to articulate value — specifically net savings and return on investment numbers — as one of the major barriers to adoption by farmers. The severity and impact of this knowledge gap is made more stark when combined with less-than-ideal big-data business models.
Why it matters: Big data tech refers to the massive data sets collected by on-board computers, sensors and digital information services offered to the agricultural industry. It has been touted as a game-changer for agriculture, but farmer adoption across Canada, the United States, and Europe has not met expectations. Some believe this is because developers have not been able to adequately illustrate the practical value of big data to farmers, nor create mutually beneficial business models. Solutions, however, are possible.
Measuring value of big data is possible
“Data is raw material for information. Information is what can be used to make decisions, but information has value only if it influences decisions,” said James Lowenberg-DeBoer, Elizabeth Creek chair of agri-tech economics at Harper Adams University, an agricultural institution in the United Kingdom.
Speaking to big-data developers and researchers at a recent conference in Houston, Texas, Lowenberg-DeBoer said his research and extension work, as well as his own experience farming in Iowa, reveals how quickly technologies can be adopted if the economic value is clear. Precision technologies like GPS and autosteer, for example, have a comparatively immediate and obvious economic return associated with their use, such as more efficient variable-rate application, elimination of overlaps and ease of use. Consequently, they became the new norm within a decade of introduction.
Measuring big-data tools like multi-layered soil maps are more complicated, though Lowneberg-DeBoer says it’s still possible to communicate value. Articulating costs associated with the time to collect information, the equipment required, software subscriptions, managing and archiving data, and analysis is the first step.
Three examples of value measurement were given during his presentation:
1) Hybrid variety choice
Information becomes available that a different hybrid corn variety produces five bushels more per acre than the hybrid currently being planted by a farmer. The information cost could, therefore, be the potential loss of those additional bushels plus an additional production costs. The gross value of that information is the value of the extra corn minus the cost of information and extra production costs. This is an example of a clear, short-term value.
2) Cropping system choice
A farmer has an unusual sandy subsoil in one section of her farm. She usually plants full season across the board, including in the often drought-stressed sandy subsoil section. The right information — generated from data pooled from other producers in her area with similar soil types — can shed light on what varieties work better in such growing conditions, whether early-season planting or transitioning the area to winter crops are options. This is an example of a more long-term value.
3) Responding to unforeseen growing-season events
Farmers in a given region participate in an equipment data pooling program, including yield monitoring data. Those farmers suffer an early frost that kills many soybeans and reduces grain quality. Upon harvest, that general yield data allows participants to make a business decision to sell those soybeans more quickly in an attempt to cut losses, assuming the grain in question does not store well. This is a more nuanced value example, but one with a clear decision option.
Proprietary data systems discourage wider (and much-needed) farmer participation
A complicating factor in these examples is the reliance on open, pooled data. In the current landscape, most big-data systems are managed by private companies operating proprietary systems. This has many farmers concerned that farm data can be used to harm their business.
“Another scenario is the company shares that data with a grain buyer that drops the buying price for inventory beans and the farms lose,” says Lowenberg-DeBoer. “This is one of the things that farmers fear, that by sharing their farm data it can be used to, in the end, hurt their business.”
Other reasons for reluctance also exist. Lowenberg-DeBoer says some farmers fear data-sharing means a loss of competitive advantage. This could take the form of, for example, a neighbour purposely outbidding them for rented land when they know the other person experienced a poor growing year, or a situation where their input suppliers take advantage of agronomic data for a stronger price negotiating position. The issue of property rights and whether a farmer should be compensated for others using their farm data is also a factor.
“This is accentuated by the feeling that much of the value of farm data is beyond the farmgate,” says Lowenberg-DeBoer. “There are few solid examples of farm benefits for sharing [information].”
Solutions lay in more refined data-sharing business models
Still, Lowenberg-DeBoer says there are data-sharing business models that could work past these difficulties. However, individual farmer views will ultimately determine participation, so any all-encompassing solution will have to take a broader approach.
To that end, he identified a number of opportunities:
1) Data systems developed with public private-partnerships. This would include governments, universities, farm associations, and ag tech businesses — though in the case of the latter, not those also selling inputs because the conflict of interest would not build farmer confidence.
2) Developing up-front compensation mechanisms for data collection costs. Compensating farmers for the effort of gathering and pooling data if it is used by others, or sold, would dispel some ownership issues.
3) Facilitate farmer benefits for access downstream, or developing ways farmers can more easily use pooled data to make their own on-farm business decisions.
4) Make data collection easier. Implementing RFID tags on seed bags and automating sensor calibration, for example, would encourage greater participation since less time and energy is required to do so.
5) Transparent, third party evaluation of data applications. Testing the actual effectiveness of data systems through randomized control tryouts — where some farmers get access and others don’t — could provide more accurate numbers for return on investment.
6) Clarify ownership issues. This could be done through legislation.
Regardless of current difficulties, though, Lowneberg-DeBoer believes big-data systems will continue to be adopted on a wider basis.
“Precision ag is being adopted when it makes economic sense. I would argue ag data will follow the same pattern,” he said. “Agricultural big-data has the potential for substantial on-farm benefits and probably even greater agribusiness profits. A well-developed big-data system may even have advantage in international competitiveness.”
“The number one obstacle is the lack of business model for farmers to share data.”