To find out what your crew really thinks, hand out a short survey with no names on it twice a year, and count the answers one question at a time. That way you know how many people share each complaint, instead of guessing. At the end you have the count for each question and one page with the three complaints that came up most, and next to each one what you will do, who is in charge of it and by when.
Maybe good hands have left, the crew barely talks at lunch, and you cannot tell whether two people are unhappy or a dozen. Ask a worker how the job is going and you will usually hear that it is fine, so you end up acting on the last complaint you heard, or the loudest one. When a good worker quits, it comes as a surprise.
Why does asking face to face rarely get you the honest answer?
Because when people can be identified, they tend to shift what they say toward what sounds acceptable, and more so when the person asking pays their wages. Survey researchers call this social desirability bias, which Zaal and colleagues describe as the tendency of people to “present themselves in socially acceptable terms”. In a 2026 article in Quality & Quantity, a journal about research methods, these survey researchers reviewed 121 experiments on ways to reduce this bias.
As they put it, providing more anonymous research environments is generally expected to reduce that bias. Studies that compared letting people answer on their own, on paper or on a screen, with asking them face to face found a significant advantage for self-administered surveys: fewer answers were shaped to please. No method removed the bias completely, so a survey gets you a more honest answer, not a perfect one.
Which questions should you ask, and why so few?
Ask about five or six things that weigh on whether people stay, such as pay, training or how people are treated, with one short question for each and the same answer choices for everyone: yes, no or not sure. If you are unsure what matters on your farm, look at why people left and what they complained about this year.
Then add one open question, where people write the one thing they would change. Keep it to one page, because people are more likely to finish a short survey. If some of your crew read another language better than English, make a copy in theirs too; the farm certification GLOBALG.A.P. asks that a farm’s complaint process be “available in the predominant language(s) of the workforce”.
Count each question on its own: how many said yes, how many said no, how many were not sure. Do not add everything into one score for the whole crew, because a good result on pay can hide a bad one on training. On a small farm, do not split the answers by age or language either, because that could reveal who answered. Where nearly everyone complains on the same question, the whole crew shares that complaint. Where the crew splits about half and half, the problem may weigh on only part of the crew, and you may find why in the open answers.
Why ask, instead of just raising the pay?
Because the wage may not be what is wrong, and a raise may solve a problem your crew did not have. Roy and colleagues, researchers who surveyed 1,202 people about work in Australia’s banana industry, found the appeal of those jobs “driven more strongly by qualitative employment conditions than by purely economic incentives”, such as training and a supportive place to work; that is one crop in one country, a warning rather than proof for your farm.
Researchers know little about what farm workers think. Sullivan and colleagues, who surveyed 329 workers on 29 feedlots in the United States, note that the views of the people doing daily cattle care have “received limited attention” next to the research on animal health and performance. Campler and colleagues, who surveyed workers on a special work visa on pig farms in the United States, point out that turnover among them is high and its “underlying causes are mainly unknown”. Asking your own crew is how you find the causes on your farm.
If you already keep numbers on your crew, such as turnover and absences, write the survey count next to them. If you keep none yet, this guide helps you choose which crew numbers to track.
Is a survey a real management tool, or soft stuff?
It is a real tool, as long as you act on what you counted. The working conditions module of GLOBALG.A.P., a certification for good farming practice, requires that workers have a way to raise what matters to them, such as a regularly repeated meeting. It does not require a survey, but you can use one for the same purpose.
At a meeting, though, you hear only what people will say in front of others. The survey adds what they will say only with no name on it, and because nobody signs it, you cannot use it to grade or rank anyone.
The survey also does not replace the everyday way a worker brings you a single problem the day it comes up; that is a separate tool, covered in how your people can raise a problem before it blows up.
When does a survey do more harm than good?
In two cases. The first is counting the answers and then deciding nothing: the crew answered for nothing and has less reason to answer the next one. The second is reading a low count as a verdict on one person, for example blaming the foreman when few people say their boss listens. If the count turns into blame, people go back to being careful with what they answer.
When you share the results, give only the count for each question and the open answers already grouped, never one person’s sheet. With few answers, look at which way the numbers move from one survey to the next rather than at any single result.
Where to start
You need a page of questions, a sealed box and an afternoon, twice a year.
In this overview of farm management you can see where a crew survey fits among everything else it takes to run a farm, and there are more articles on managing people in this section of the library.