Kwabena’s Policy Gap. The Counter Is Closing and His Money Is at Risk.

A policy notice can state a rule, but it cannot anticipate every decision a cocoa farmer must make next. Institutional guidance therefore needs reviewed follow-up answers and a clear route to an accountable person when the approved material does not cover the question.

Imagine Kwabena, a composite young cocoa farmer, standing at an input counter outside Kumasi late on a Friday afternoon. He is holding the notice in one hand and a folded receipt in the other. Two customers wait behind him.

He has understood the announcement. His problem begins with the next sentence.

In Asante Twi, he asks whether the new guidance applies to supplies he arranged before the notice appeared. If it does, he may need to change what he buys. If it does not, changing the order could leave him without enough money to complete the work he planned for his farm.

The person behind the counter reads the notice again. It says nothing about earlier arrangements.

Kwabena cannot tell which choice will cost him more. The counter will soon close, and the supplier may release the goods to another buyer.

The hardest question often comes after the announcement

Public guidance usually answers the question its authors expected: What is changing?

A farmer’s next question is narrower and more urgent. Does the change apply to the decision I am making today?

That gap matters because a clear notice can still produce unsafe or costly improvisation. A clerk may offer a personal interpretation. A neighbour may remember a similar policy differently. A confident digital assistant may generate an answer that sounds complete while resting on no reviewed source.

For Kwabena, repetition would not help. Reading the same notice aloud in Asante Twi could improve access to the words, yet his real question would remain unanswered. The system must recognise the boundary between “I can explain this notice” and “I have authority to decide how it applies to your case.”

This is the same distinction explored in what happens when guidance cannot reach farmers in a language they understand. Language access gets the farmer to the policy. Follow-up support helps the farmer act on it.

A safe voice workflow needs permission to stop

AgriVoice is being prepared around a constrained approach. A farmer asks a question in Twi, and the system selects from content that people have already reviewed. It does not invent agronomy or fill gaps with plausible language.

That limit becomes especially important when the answer is missing or uncertain. The safe response is a plain acknowledgement, followed by escalation to the named extension officer responsible for handling the question.

For Kwabena, that means the voice workflow could explain the approved part of the notice, capture his follow-up, and route it to a person. It could not promise that his earlier arrangement qualifies. Until an authorised answer comes back, the uncertainty stays visible.

This refusal may feel less impressive than an instant answer. It is more useful than manufactured certainty.

The same principle applies when the stakes involve pesticide dosage, re-entry periods, purchasing rules, or any instruction that could affect a household’s income or safety. A practical comparison of constrained selection and human escalation shows why a system must know when its reviewed material has run out.

Routing must end with a named owner

“Ask an officer” is weak guidance when nobody owns the question.

A workable escalation path identifies the person or role receiving the request, preserves the farmer’s question in the language used, and records whether someone responded. The institution also needs a stated response commitment. Otherwise, escalation becomes a polite way to lose the farmer in another queue.

The recent petition from the Young Cocoa Farmers Association, asking President John Dramani Mahama to clarify policies and funding arrangements ahead of the 2026/2027 cocoa purchasing season, illustrates the wider need. Announcements create follow-up questions about how policy reaches individual decisions. Those questions require accountable answers, especially when money or timing is involved.

AgriVoice’s planned farmer pilot treats escalation as a measured operation. A cocoa-sector partner must name the extension officer who receives unresolved questions. The pilot will track whether that owner responds within the intended working window. If nobody owns the queue, the workflow has failed even when the speech technology performs perfectly.

Design the follow-up before publishing the notice

Before an institution releases guidance, its team can test it with a short exercise: place one farmer at a counter and ask what they would need to decide next.

Which cases fall outside the wording? Which terms need spoken Asante Twi explanations? What could cause financial loss or physical harm if someone guessed? Who has authority to answer, and how will that answer return to the farmer?

These questions expose the missing operational layer while there is still time to build it.

Back at the counter, Kwabena does not need another confident interpretation. He needs his exact question preserved, sent to an accountable officer, and answered from an approved source before he commits his money. The useful system is the one that keeps his uncertainty intact until the right person resolves it.