Most agricultural organisations can describe, in some detail, what they have done to support innovation this year. Far fewer can describe, with any confidence, what farmers are still doing with it six months later. That gap between activity and adoption is not a reporting failure. It reflects how difficult adoption actually is to measure.
Agricultural organisations measure activity more easily than adoption
Annual reports and steering committees tend to rely on a familiar set of figures: technologies reviewed, meetings organised, pilots launched, events attended, content distributed, platform registrations. These numbers are easy to collect because they are generated by the organisation itself, on its own schedule, using its own systems.
None of them establish that a farmer is still using the technology, still following the practice, or still finding it worthwhile. A high registration count says something about outreach. It says nothing about whether the tool was opened twice or a hundred times. A well-attended demonstration day says something about interest. It says nothing about what happened on the farm the following month. Activity metrics answer "what did we do," not "what changed as a result."
Adoption does not have one universal definition
Part of the difficulty is that "adoption" means different things depending on what is being adopted. A sensor is adopted when it stays installed, maintained and consulted over time. A biological input is adopted when a farmer chooses to apply it again, unprompted, in a subsequent season. A software platform is adopted when it becomes part of a routine rather than an occasional login. A cultivation practice is adopted when it survives beyond the season it was introduced in, without continued reinforcement. An autonomous machine is adopted when it is trusted with tasks it was not closely supervised for at first.
Each of these definitions implies a different observation window, a different signal of success, and a different way a farmer could plausibly stop halfway. Applying one generic notion of adoption across all of them, as many reporting templates implicitly do, produces numbers that look comparable but are not measuring the same thing.
Agricultural cycles delay observable outcomes
Agriculture does not produce fast feedback. A practice introduced in one season may only reveal its effect on yield, soil health or pest pressure after several cycles. Weather variation from one year to the next can mask or exaggerate a technology's actual contribution. Biological processes, whether in soil, plant or livestock, unfold on their own timeline regardless of a reporting calendar. Economic effects, when a technology's real payoff is a cost saved or a loss avoided, often only become visible after a difficult season that did not happen.
An organisation that evaluates adoption on an annual reporting cycle is frequently asking a biological or economic question before biology or economics have had time to answer it.
Initial use can be mistaken for sustained adoption
A farmer trying a new input during a subsidised pilot, attending a supported trial, or participating because an adviser asked them to is not the same as a farmer choosing, on their own initiative and at their own cost, to keep using it. The behaviour that matters most, what happens once the support, the subsidy or the attention is withdrawn, is exactly the behaviour that is hardest to observe, because that is usually also when active data collection stops. This is closely related to why promising pilots rarely translate into lasting use once external support ends. The distinction between initial trial and continued adoption is often invisible in the data an organisation already has, because the data collection itself tends to end at the same point the pilot does.
Technology performance is difficult to separate from context
Even when a technology is clearly still in use, attributing an outcome to it is rarely straightforward. Agronomic conditions differ from one plot to the next. A farmer's prior experience shapes how well any tool is implemented. Market prices determine whether a marginal yield gain translates into a meaningful economic result. Climate variation between seasons changes the baseline a technology is being compared against. The quality of implementation, how carefully a sensor was installed or a practice was followed, varies even within a single supported cohort. Complementary practices, adopted alongside the technology being evaluated, contribute to the same result without being credited or discounted.
Two farms using the identical technology can show different outcomes for reasons that have nothing to do with the technology itself. Without accounting for that variation, a result can be read as a technology success or failure when it was actually a context effect.
Data is distributed across several actors
Evidence of adoption is rarely held by any single party. Farmers know what they actually did on their land, which is not always what was reported. Advisers know what they observed during field visits, which is partial and intermittent. Technology providers know usage data for their own product, but usually nothing about the farmer's broader practices. Cooperatives know delivery and membership records. Researchers hold trial data collected for a different purpose, often on a different timeline. Internal departments, marketing, technical, financial, each hold a fragment relevant to their own function.
No one of these sources gives a complete picture, and few organisations have a reliable way to bring them together. This fragmentation of information across actors is a specific expression of a broader pattern in how agricultural innovation activity gets scattered across an organisation and its partners, making almost any adoption question harder to answer than it initially appears.
Negative and abandoned experiences often disappear
Organisations tend to document what continued. A successful pilot becomes a case study. An active user becomes a testimonial. A technology that was quietly abandoned, a practice that was tried once and dropped, or a farmer who disengaged without explanation rarely generates a report. Nobody is usually assigned to write one.
This produces a form of survivorship bias in the organisation's own evidence base. The visible record skews toward what worked, or at least toward what is still active enough to be visible, while the reasons things stopped, arguably some of the most useful information available, go largely unrecorded.
Measurement affects future decisions
Weak adoption measurement does not stay contained to a reporting exercise. It makes it harder to justify further investment in a technology whose real uptake is unknown. It makes it harder to compare two initiatives against each other when the underlying numbers were never measuring the same thing. It makes it harder to learn from a failed effort when the failure was never documented in the first place. And it makes it harder to communicate credible impact to funders, members or boards, who increasingly ask for adoption evidence rather than activity counts.
Over time, an organisation operating without reliable adoption data tends to default to funding what is visible and comfortable to report, rather than what is genuinely working on the ground.
Questions agricultural organisations should be asking
Before adding another metric or another dashboard, it is worth asking a smaller set of harder questions:
- What do we currently call "adoption," and does that definition hold across every type of technology we support?
- Are we measuring participation, or are we measuring continued, independent use?
- Over what period should results realistically be assessed, given the biological or economic cycle involved?
- Which contextual factors could plausibly explain this result, aside from the technology itself?
- Who currently holds the information we would need to answer this question, and do we have access to it?
- Are unsuccessful or abandoned experiences being documented anywhere, or only the ones that continued?
- Which of our current conclusions are actually supported by evidence, and which are assumptions we have simply repeated often enough to trust?
A problem of definitions, continuity and context
Measuring agricultural innovation adoption is not primarily a data collection problem. It is a problem of definitions that shift by technology, timelines that outlast reporting cycles, context that is rarely accounted for, and information that sits scattered across farmers, advisers, providers and internal teams who were never asked to compare notes. Until those underlying issues are addressed, more dashboards will mostly produce more precise versions of the same incomplete picture.
Is your organisation facing this challenge?
Vitto works with agricultural organisations confronting the complexity of innovation discovery, coordination and adoption. If measuring the real impact of your innovation efforts is a recurring challenge, contact our team to discuss it.