The One Follow-Up Plan Cocoa Households Need, and What Its Absence Can Cost

When a health referral and a crop-advice escalation reach one cocoa-farming household, the programme should coordinate the handoffs around that household. Shared ownership, consent-based information exchange and one visible follow-up plan can prevent missed visits, repeated questions and conflicting instructions.

In April 1970, Apollo 13’s crew faced rising carbon dioxide inside the lunar module. The command module’s square scrubber cartridges could not fit the lunar module’s round openings. The astronauts had the materials on board, but no proven way to assemble them into a working adapter.

NASA’s engineers in Houston had to solve the problem using only items available to Jim Lovell, Jack Swigert and Fred Haise in space. The ground team developed and tested a makeshift adapter, then passed precise assembly instructions to the crew. The astronauts built it. Carbon dioxide levels fell.

NASA’s account of Apollo 13 documents the wider rescue as a coordinated effort across flight controllers, engineers and astronauts. No single specialist could bring the crew home. The outcome depended on each person understanding the same emergency, owning a defined part of the response and handing useful information to the next person.

Two referrals can create one household problem

A cocoa farmer may ask for help with a diseased crop while someone in the same household receives a health referral. Those cases may enter through different programmes, phones or field workers, yet they compete for the same household’s time, transport, attention and trust.

A crop recommendation can also affect health. Advice about pesticide use may raise questions about protective clothing, exposure, re-entry or the safety of other people near the farm. AgriVoice is designed to select reviewed agronomy content rather than generate fresh instructions. Chemical guidance remains withheld until a Ghanaian agronomist has verified dosage, re-entry and pre-harvest intervals. Missing or uncertain answers go to a person.

That escalation becomes weaker if the extension officer sees only a crop ticket while a health worker separately handles a related referral. Each person may complete their assigned task while the household still receives an incomplete response.

Coordination does not mean combining every record. It means recognizing when two human handoffs affect the same people, gaining consent to connect them and assigning responsibility for the next step.

Build one accountable handoff plan

Apollo 13’s improvised scrubber worked because responsibility was explicit. Engineers on the ground designed with the crew’s actual materials. Flight control carried tested instructions across the gap. The astronauts executed them and reported the result.

Cocoa programmes need the same operational clarity. Before a referral system reaches farmers, define:

• who receives crop escalations;
• who receives health referrals;
• what information each role may see with the household’s consent;
• who contacts the household when both cases are active;
• when either specialist must refer back, close the case or escalate further;
• how unresolved cases remain visible.

A shared case reference can connect the work without exposing unnecessary health or farm details. The health worker may need to know that a farm visit is pending, while the extension officer may need only an approved warning that health follow-up affects timing or participation. Access should follow purpose, not curiosity.

One person must own coordination. “Referred” cannot count as resolved. The case closes only when the responsible worker records what happened, what remains open and who acts next. This is the same ownership problem explored in The Escalation Queue: What Happens When Answers Have No Owner?.

Measure the household outcome

Separate programme dashboards can make a broken handoff look successful. The crop system records an escalation. The health system records a referral. Both teams count activity, even if nobody reaches the household or the instructions arrive in the wrong order.

Measure the combined outcome instead. Did the correct people receive both cases? Did the household understand the next steps? Was consent recorded before information crossed programme boundaries? Did a named person respond? Did either queue remain unresolved?

For AgriVoice’s planned two-week pilot with 20 to 50 Asante-Twi-speaking cocoa farmers, escalation already has a concrete standard: a named owner and a median response time below one working day. A coordinated programme should also test whether related referrals can be recognized without weakening privacy, and whether the household receives one coherent plan.

Pesticide cases deserve particular care. A voice system should refuse to complete unsafe advice when verified details are missing, as described in The Half-Filled Spray Tank AgriVoice Must Refuse to Complete. A human handoff then needs enough context to act, but no more personal information than the task requires.

Rehearse the handoff before farmer exposure

Run a tabletop exercise with one linked health and crop case. Use fixture data, record consent, route both referrals and ask each worker what they can see. Then follow the case until every action has an owner.

Test the awkward moments. What happens when one worker cannot reach the household? Who sees that the other referral remains open? Can a household withdraw consent or request deletion? Does either programme mistakenly treat forwarding as completion?

NASA’s engineers did not send an untested idea to Apollo 13. They assembled the adapter on Earth with the materials available in space, then communicated a procedure the crew could follow. Cocoa programmes can apply the same discipline: rehearse the complete human chain before a real household has to depend on it.