Are Ag Inputs More like Wine or Diamonds?

This article is a guest post by Dan Northrup of Galvanize Climate Solutions, LLC.

Introduction – Wine, Diamonds, and Inputs

Wine and diamonds have major differences in how they are priced and in consumer buying motivations.  For wine, 85% of people buy based on the label[1], with less consideration of the quality of the bottle’s content.  Diamonds are priced based on the 4 C's – clarity, cut, carats, and color[2] – quantifiable quality attributes[3].  

Although ag input decisions appear to be diamond like transactions, purchasing is nuanced, and decisions vary by input category (fertilizers, seeds, crop protection)[4]. Price is a primary reason for choosing fertilizers. Fertilizers are tested for nutrient composition and unbiased public sector testing ensures product quality. The correlation between crop yield and fertilizers is reasonably well characterized and supports price-based decisions.

For seeds and crop protection performance is reported as a key reason for purchase, and this attribute takes multiple forms: yield response, weed control, pest control, etc.  Unfortunately, the complexity of these responses and the difficulty in making accurate comparisons between products makes it challenging to achieve the same degree of rigor in ROI calculations at the time of purchase. This is a challenge to an unproven product and is important for product developers because observability of outcomes is one of the key drivers of adoption[5].

The datasets that are available to farmers at the point of decision making also vary by input category. Plant breeders benefit from extensive testing networks. Commercial varieties and hybrids are typically tested for 4 or more years in 100s of locations before they are advanced to the market. The result of this testing is a small set of geographically targeted products characterized by extensive datasets that support diamond like, ROI based decision making. 

The 3^rd major category of agronomic inputs – chemistries, biologics, stimulants – has a more limited infrastructure for generating performance data. Unlike seeds, these inputs are expected to perform across wider geographic ranges and conditions even though this may not be a valid assumption. Without datasets to characterize the performance variation farmers and agronomists do not have the information they need to assess a product’s ROI and this can discourage the adoption of new products and product categories.

Variable Outcomes Challenge ROI based Decision Making and Product Pricing

The challenge of observing returns on a new product is caused by the significant variability in performance which is driven by many interacting parameters (genetics, environment, management). Prediction accuracy requires observation, and the number of samples needed to achieve a given degree of statistical certainty increases with the variation. Thus, highly variable outcomes – particularly yield and ROI - require large amounts information to gain statistical confidence of the potential outcomes. 

In addition to the cost, another unfortunate consequence of a high degree of variability is that assessing ROI requires time. While it is important to prioritize performance-based purchasing and advantageous to price products based on the full value they create, the data gathering and observation challenge to support this value calculation is at odds with a business imperative to advance products quickly.

On Farm Testing to Accelerate and Reduce the Cost of Data Collection

It is my opinion that solving this data generation and product proof challenge is important to ensure that agriculture is an attractive category for investment and innovation. In medicine, drug companies are aiming to use “real world evidence[6]” to augment traditional data collection methods and reduce the cost of drug development. Applying a similar concept to agriculture, several groups[7] have used precision equipment to perform on-farm research. Expanding this capability so that more farmers test new products in the production setting will produce highly relevant datasets to estimate a product’s ROI. Equipped correctly, growers can generate critical data to place and price new ag inputs and create a new dynamic between the customer and the product developer. 

There are multiple tools to expand on-farm product testing and apply the results to product development. Using yield monitors and statistically optimized field trial design, growers can develop higher resolution datasets with multiple tests per field. A critical function of these tools – particularly field design – is to minimize the tradeoffs between the value of the data on product performance and yield peril if a new product does not perform as expected. 

To get to quicker conclusions, it is important to make in-season observations of intermediate performance metrics. While acknowledging that weigh scales are the only determinant of ROI, scales are a blunt instrument. Like a gambler only tracking wins or losses, it is hard to determine a performance edge only tracking yield. Conclusions about a product’s value will be improved by direct observation of in-season outcomes such as crop vigor, nutrient content, predation, etc. The availability of drones, autonomous robotics, and vision enabled implements can be used to measure these attributes without significant additional work.

To generate rigorous estimates of ROI with confidence intervals, testing requires careful tracking of uncertainty. All instruments are imprecise and quantifying the errors that they introduce is necessary to measure averages and standard deviations. Modern data analytics need to be combined with informed views of how to combine datasets and when they should be considered separately. For example, the performance of a product in rainfed or irrigated systems or across soil types may be so different that combining datasets masks real effects.

Statistical consideration of ROI will consider the range of possible performance conditions and outcomes.  Rather than coming to a single number, a product’s ROI is a sum of a probabilistic set of outcomes similar to the expected value of a game of chance. It is important to be patient. Regardless of the degree of testing, there is no avoiding Cromwell's Rule[8], and products will assuredly perform differently than they did in trials – i.e. The Ferdinand Effect / Winner’s Curse [9]. 

There is a critical role for extension and the research community to help growers design and implement trials, analyze results, and adjust practices. Through this collaboration, growers will make adjustments and adopt new inputs without over-reactions to either positive or negative results. Over time, highly favorable and unfavorable seasons will contribute to ROI calculations that converge to an expected value and an accurate price. With data and sophistication in adoption, ag input purchasing can move from wine to diamonds.

Conclusion

Getting great products means being a great customer. Great customers inspire market entry with product developers who can target performance metrics.  When metrics are ill defined, other sales features like labels and bundled services favor incumbents. Worst still, snake oil salesmen are always present. Without strong data it’s difficult / impossible to rank effective products and remove ineffective products from the marketplace. 

Farmers need innovation to thrive in adverse and volatile circumstances. By developing a new product development paradigm and datasets to back ROI based decision making, farmers can drive innovation and encourage new entrants to invest in product development relevant to their pressing challenges. In assessing ROI, companies can pay growers to generate the data, and in the common pricing scheme in agriculture, 1/3 of the value to the company, 2/3 to the grower[10] the deal works for both sides. 

Other Upstream Ag Insights Guest Posts from Dan Northrup:

1. Agriculture – On the cusp of a rapid evolution - November 2023

2. Agricultural Transitions for a Net Zero Economy, special feature - irrigation - November 2024

[1] The Drinks Business, Four in five people have purchased wine based on what the label looks like

[2] American Gem Society, What Are the 4Cs of Diamonds?

[3] Roughly, wine and diamonds have the characteristics of B2C and B2B transactions: sales led by the label and bundled services (B2C) or priced on performance metrics such as return on investment (ROI, B2B). 

[4] Purdue, Farmers’ Purchasing Behavior and Implications for Suppliers’ Go-To-Market Strategies

[5] Upstream, Theory of Innovation Adoption in Agriculture: An Application

[6] JAMA, Modernizing the Data Infrastructure for Clinical Research to Meet Evolving Demands for Evidence

[7] E.g. Indigo, FBN, Iowa Nitrogen Initiative, DIFM, EAGIC, Iowa Soy, UNL, Wine Australia

[8] Wikipedia, Cromwell’s Rule

[9] Ferdinand, Munro Leaf –Ferdinand is a peaceful bull who gets stung by a bee. Because he looks like a wild fighter, he’s brought to the bull fights, only to sit in the ring and return to his normal demeanor. Unfortunately, things that look extra-ordinary, are often ordinary things on an odd day.

[10] While it deserves treatment in a longer article – products advanced through this pipeline will graduate from cost-based to value-based pricing founded on datasets that demonstrate the product’s effects.