Farm technology is usually introduced at its most impressive.
The demonstration works. The dashboard is tidy. A sensor produces a number that was not available before, or a machine completes a task without someone sitting in the seat. The benefits sound obvious when they are presented one at a time.
But a farm does not buy an isolated feature. It takes on a complete operating system: equipment, software, subscriptions, connectivity, maintenance, support and a new set of daily decisions. The useful question is not simply whether the technology works. It is whether the complete system earns a durable place in the business.
These are the five questions The Farming Edge uses when looking at farm technology.
1. What work does it genuinely remove?
“Labour-saving” is one of the easiest claims to make and one of the hardest to define properly.
A system may automate a task while creating several smaller ones: cleaning a camera, charging batteries, maintaining tags, checking exception lists, correcting records, investigating alerts or helping a remote technician diagnose a fault. A robot may remove hours from one job but add supervision at awkward times. A dashboard may replace handwritten records without removing the physical inspection that gives those records meaning.
The starting point should be the existing job:
Who does it now?
How often?
How long does it actually take across a normal week?
What skill or judgement does it require?
Does the work occur at a predictable time, or does it disrupt other jobs?
Then record the new workflow with equal honesty. Include setup, routine supervision, cleaning, calibration, data correction, alert investigation and breakdowns. Time transferred from an experienced person to another member of staff is not automatically time saved. Neither is a job moved from daylight hours to an alarm at night.
The strongest labour claim is not “fully automated”. It is a measured before-and-after account of which jobs disappeared, which remained and which new jobs arrived.
2. What measurable result improves?
Producing more data is not the same as improving a result.
A weighing system may create a daily growth curve, but the important question is whether that information changes feed, drafting or sale decisions accurately enough to improve performance. A health alert may identify an animal sooner, but value comes only if the farm can examine and treat that animal promptly. A camera may count, score or classify more consistently than a person, yet still fail to change the decision that follows.
Before installation, choose the outcome that matters and establish the baseline. Depending on the system, that might be:
minutes or hours of labour per week;
mortality, treatment or relapse rate;
feed conversion, daily gain or days to slaughter;
yield, quality loss or rejected output;
energy use or unplanned downtime;
variation between batches or groups;
frequency and cost of missed events;
staff safety or exposure to a hazardous task.
The outcome should be something the farm can check. “Better insight”, “greater efficiency” and “improved sustainability” are directions, not measurements.
It is also worth deciding in advance what would count as failure. If a trial produces interesting data but no change in labour, cost, welfare or performance, the farm should be able to say so without moving the goalposts.
3. What else must it work with?
Very little farm technology operates alone.
It may depend on an existing controller, RFID system, mobile signal, broadband connection, cloud service, herd-management package, machine interface or supplier data format. It may require every animal or item to pass one camera point in the correct order. It may need clean, consistent records before its recommendations are reliable.
Integration questions are often more important than the headline feature:
Which makes, models and software versions are supported?
Is the connection proven on this farm’s exact configuration?
Does information move automatically, or must someone enter it twice?
Can the farm export its own records in a useful format?
What happens to the history if the supplier, subscription or machine changes?
Does the system still produce a safe, visible result when connectivity is poor?
“Compatible” should mean more than the two systems can theoretically exchange data. It should mean the complete workflow has been installed, commissioned and supported on a comparable farm.
The physical fit matters too. Camera position, lighting, dust, water, metal-clad buildings, animal flow, multiple races and seasonal changes can turn a neat demonstration into a difficult installation. A written site survey is more valuable than a general assurance that it should work.
4. What happens when it fails—and who supports it?
Every system will eventually lose power, connectivity, a sensor, a camera, a tag or a software service. The useful distinction is between a failure that is obvious and contained, and one that quietly produces incomplete or misleading information.
A farmer needs to know:
How does the system show missing, delayed or low-confidence data?
Does it fail safe, or can it continue presenting an apparently normal result?
What can farm staff diagnose and replace themselves?
Which parts are held locally?
Is support remote, on farm, or routed through a dealer?
Who answers outside office hours when the system controls a time-critical job?
What response and restoration times are actually promised?
Uptime percentages are not enough on their own. A short failure during an unimportant period may have little effect; the same failure at feeding, milking, ventilation or drafting time can dominate the value of the system.
User evidence is particularly useful here. A manufacturer can explain how support is designed. A farmer who has used the product through a fault can explain how support behaves.
The most revealing buying question may be: tell me about the last time it went wrong.
5. Does the complete cost produce believable retained value?
Purchase price is only the first line of the calculation.
The full cost may include installation, building alterations, cabling, gateways, tags, commissioning, integration, staff training, subscriptions, mobile data, replacement parts, calibration, servicing, additional inspections and the time spent acting on alerts. Finance and contract exit costs may matter as well.
Some of those costs replace existing expenditure. Others are genuinely new. They should not be mixed together.
The other side of the calculation needs the same discipline. Supplier claims about theoretical losses are not automatically savings achieved by the buyer. An early alert is not proof that a costly failure was prevented. More accurate information is not a financial return until it changes a decision and the result can be measured.
A conservative calculation looks like this:
annual labour genuinely removed + verified performance or loss reduction
minus
annualised installation + subscription + maintenance + support + new labour and operating costs
The result should then be tested against less favourable assumptions. What happens if the system delivers half the expected benefit, requires more staff time or lasts fewer years? Does it still make sense? How dependent is the return on one optimistic claim?
The final question is not how much value the technology creates in theory. It is how much value the farm keeps after paying for and operating the complete system.
Start with one decision, not an entire platform
The safest trial begins with a clearly defined problem.
Choose one task or decision, record the existing baseline and agree what success would look like. Keep a log of missing data, false alerts, manual corrections, support contacts and staff time—not only the occasions when the system performs well.
Where possible, use a representative period. A short demonstration may prove that equipment can connect and produce data. It rarely proves reliability across seasons, a full production cycle or the failures that determine long-term value.
At the end of the trial, the decision should be straightforward:
adopt it because the measured result justifies the complete cost;
alter the workflow and test again;
or stop without treating the experiment as a failure.
Walking away from technology that does not fit is a useful result. It is cheaper than keeping a system because time and money have already been spent on it.
Evidence before excitement
Good farm technology does not need to remove people from every decision. It needs to put reliable information or machinery in the right place, at the right time, with less work or a better result than the available alternative.
That is a higher standard than an impressive feature list. It is also a fairer one.
Before technology earns its place on farm, ask:
What work does it genuinely remove?
What measurable result improves?
What else must it work with?
What happens when it fails—and who supports it?
Does the complete cost produce believable retained value?
The answers are where the real product begins.