To guarantee qualitative rigor, in this case, defined by the points at which no new themes are generated from the interviews, three open-ended questions were asked for each objective, adding up to a total of nine questions in addition to basic biographical questions. The interview questions, which are provided in the Appendix, were standardized with the same nine questions were asked for every participant. In addition, following semi-structured interviewing best practices, we asked participants to elaborate on answers and we dedicated more time to certain topics depending on interviewee expertise. These additional questions and structural flexibility enhanced feasibility and accommodated participant preferences.The first step of thematic analysis was interview coding, in which the interview transcripts were run through the software ATLAS.ti , an AI-enabled qualitative data analysis software. An open coding method was used within the framework of grounded theory in which the textual data is used to uncover the responses of individuals to changing conditions and the subsequent consequences . A procedural characteristic of grounded theory is that data analysis must occur at the same time as data collection; the coding and analysis of the first interview should incorporate details of all potentially relevant information into the design of the following interviews . The similarities between the first interviews confirmed the efficacy of the interview questions.
For the first research objective about compatible motivations, vertical growing systems the coding followed a simple coding process: when reviewing the first few participant transcripts, the data from the first set of three questions was cross-checked to identify recurring codes . Then, each additional transcript analysis added to existing codes, as well as potentially generating new ones. Detecting, classifying, and counting the presence of motivation-related codes quantified qualitative data, providing a more rigid content analysis . On the other hand, a thematic framework was developed and the data was indexed against the framework for the second and third objectives . This inductive thematic framework enabled comprehensive indexing and comparative analysis between the interviewees, allowing for the mapping of patterns. Because this study is about the complexity of stakeholder relations and how subjective perceptions leave tangible impacts, we used a structured yet non-mathematical approach for multi-level cognitive maps to answer the second research objective about stakeholder interactions. Cognitive maps are an example of soft organizational research , representing mental models of stakeholders and the processes by which they gather information and make informed decisions that help them reach personal goals . Cognitive mapping has been employed in many fields, including policy development and healthcare, to examine organizational decision making . Precision weeding is a suitable technology for this methodology because its stakeholders have varying, complex assumptions about weed management issues and the role of agtech as a solution. First, individual stakeholder maps were created from the individual interviews . Then, the individual cognitive maps were combined for each stakeholder group .
To combine individual stakeholder maps to create stakeholder group maps, similar themes were overlaid, links were added between themes that individual interviewees contributed, and clustering was identified in the stakeholder group maps . To answer the third research objective about grower user journey, we employed a similar soft OR approach to code concepts and present them in a “swim lanes” format. “Swim lanes” are a common industry method to map out the customer experience: how customers learn, interact, and respond pre-purchase to post-purchase .The key results of this research revealed the motivations behind adopting precision weeding technologies, the financial, R&D, and social exchanges and collaborations between the stakeholders, and the user journey for growers using the products and/or services of precision weeding startups. The key results included the varied motivations between the stakeholders and thus varied understandings of precision weeding’s value, the controversial role of government in accelerating precision weeding technologies, and the user journey of growers adopting precision weeding technologies.For the first objective about compatible motivations, the questions the interviewees answered differed slightly based on their stakeholder group. The motivations answered why precision weeding technologies should be within the future of weed management: why startups are developing, why growers are adopting, why VCs and CVCs are investing, and why government agencies are supporting precision weeding technologies.
Eleven indexed motivations were found in responses to the first objective. Fig. 1 shows the frequencies of these 11 indexed motivations per stakeholder group. The most common motivator was labor concerns, which was cited by 13 out of 17 interviewees . Three participants added additional details about the labor pressures of organic farming in California, five participants spoke about the competitive labor market, three stakeholders mentioned budget difficulties due to California’s increasing minimum wage, and two participants addressed the role of precision weeding technologies in increasing the efficiency of labor. The interviewee Gr3 said that “as you lose your herbicide, you got [sic] to rely more on hand labor [and] mechanical labor.” While they did not think that precision weeding technologies will ever completely replace hand labor, they speculate that growers will be able “to do a lot of heavy lifting with these newer mechanical weeders.” Following labor concerns, the second-most common motivator was cost, which was cited by 12 interviewees . The motivator of ‘meeting specific field conditions/needs’ was cited by seven interviewees . Precision weeding technologies were also recognized for their ability to address specific field conditions or needs such as varying soil conditions, banding, and thinning. Four interviewees noted precision weeding’s potential to ‘transform agriculture’ and to provide positive ‘returns on investments.’ Within the category of ‘transform agriculture,’ five interviewees, primarily startups, addressed precision weeding’s potential to add value and increase farm profitability through sensors, additional data collection, advanced computation abilities, and automation. The motivators of ‘more weeding options’ and ‘environmental sustainability’ were acknowledged by a few interviewees . According to growers Gr3 and Gr4, the motivator of having additional weeding options through precision weeding technologies is partially a result of increased pesticide regulation in California, which may cause growers to lose access to certain types of pesticides. The government stakeholder added concerns about “glyphosate-resistant varieties out there…these weeds are mutating and they’re resistant…then these new formulations come out…these spray-resistant weeds are mutating and getting worse and worse. I would love to see anything that can do targeted spraying or manual weeding come out to the front.” Other interviewees added that weeds eventually adapt to weed management tools and thus effective regimes vary both temporally and in terms of products used. Three interviewees explicitly mentioned ‘environmental sustainability’ but the umbrella of environmental sustainability includes the benefits of fewer inputs and chemicals, cited by six interviewees , animal and human health outcomes, cited by three interviewees , soil conservation, cited by one interviewee , and following the United Nations’ Sustainable Development Goals, also cited by one interviewee . The motivation for aesthetics was evoked by three out of seven of the growers and refers to the negative impact of weeds on the aesthetic or visual appeal of the fields . Interviewee Gr2 shared that “growers like their fields to look nice and so [weeds] are also removed for aesthetic reasons” and interviewee Gr3 added that weeding is also a preventative measure so that harvesters do not accidentally harvest weeds in addition to the crops. Though there was a consensus across all stakeholder groups about the importance of labor concerns, other motivators were more polarized . For the most part, grow rack only growers expressed concerns about weeds harboring diseases, pests, and viruses, weeds competing with crops for resources, and the aesthetic value of weeding. In addition, only growers mentioned–under the motivator of ‘more weeding options’–that precision weeding adoption was partly driven by concerns that increased regulation in California could cause growers to refuse access to herbicides.
To answer the second sub-question regarding collaborative models between stakeholders, the comprehensive indexing of themes revealed that the average number of constructs, defined as key words and/or concepts that address stakeholder interactions and limitations, for all individual interviewees was 60. Excluding labels and descriptors, the startup stakeholder group produced an average of 59 constructs, growers had 58 constructs, and VCs had 63 constructs. All stakeholder groups produced a similar range of constructs, indicating the universality of the interview questions asked. Growers identified several blockers to adoption, such as competition between growers, old-school mentalities, and a lack of connection between startups and growers . Precision weeding requires the bandwagon effect for growers to want to try new technologies. However, competition between growers might hinder the bandwagon effect because growers may not wish to share their competitive advantages with their neighbors. This stakeholder group also asserted that many growers view working with startups as a high-risk endeavor, citing the high capital expenditures of most precision weeding machinery and the history of unsuccessful agtech startups. Because of these perceived risks, the multiple farm managers who work at one company may disagree with one another and prevent adoption. Startups also added that concerns about startup longevity are especially intensified because most traditional agricultural companies, such as John Deere, have been around for decades or centuries . Furthermore, growers perceived old-school mentalities and a potential lack of on-paper education as a blocker to the adoption of precision weeding technologies . Some may view new technologies as unnecessary and the mark of ‘true’ growers as putting in the hard work twelve hours a day, seven days a week. In addition, because many startups compare their products’ efficiencies and costs to hand crews, some growers fear automation replacing their jobs. Although many growers want to own their own equipment, they may be reluctant to hire specialized staff to run the equipment. Startups and VCs also mentioned and expanded upon the growers’ urge to own their own equipment, coming into conflict with the weeding-as-a-service business model that some startups have ventured into. Interviewee V2 also added that “eventually farmers need to own the equipment [because of] timing. As your operations become larger, timing becomes absolutely critical. As you grow different crops in variable environments, you need the machine. You may be in a field and discover; I need the machine right now and you phoned the service guy and he’s got three farms ahead of you.” Other startups have turned to the weeding-as-a-service business model to abate prohibitive capital costs and to ensure the machinery out on the fields are up-to-date with the startups’ latest developments . S3 illustrated this point by saying “the first-generation spray that we’ve built is like the iPhone 1, and technology is changing so fast that I know in three months I’m going to have iPhone 3 coming out.” S3 added that startup-centric reasons for the service model include allowing the startups to have constant access to new data and the ability to quickly relay failure points to the R&D teams. In addition, the weeding-as-a-service model provides a more intimate experience between the startups and the growers, enabling startups to conduct in-depth customer discovery for their current products and future ideas. The startup interviewees brought up the limitation that some startups lack connections to growers . Many proposed that startups need to hire employees who have worked in the agricultural industry and have local connections, while some also brought up that startups could develop strategic alliances with a committee of growers. Another startup limitation was the long timelines for hardware research and development, raising concerns about financial runways and funding. Some startups asserted collaboration between startups could alleviate runway fears as many startups have complementary products; consolidation will save time and effort. According to VCs, startup-university and grower-university relationships are often difficult to navigate and are not always advantageous . With startup-university partnerships, patent battles may sometimes emerge, particularly if the startup’s distinguishing technology directly spun out of university-sponsored research. In terms of how growers interact with universities, university research topics and trial designs are usually limited in scope and not perfectly aligned with the goals of the growers. For both parties, interviewees agreed that portfolio support from VCs to startups includes hiring and marketing support, business acumen and advice, connections to lawyers, accountants, and other startup founders, advancing governance to create stable and mature companies , and financial advising. VCs also mentioned hands-on, agriculture-related support such as matching startups with growers for field trials, building a grower advisory board, and helping with plot designs and trialing systems.