The room was set for a fight.
On one side: David Trimner, milking 3,000 cows across 30 robots and a conventional parlor in Athens, Wis., and Dan Venteicher, running 200 cows on three robots with his brother in Strawberry Point, Iowa.
On the other: Courtney Halbach, outreach specialist for the University of Wisconsin’s Dairyland Initiative, who alongside Nigel Cook has raised pointed questions about whether robotic milking really works at scale.
The conversation that followed wasn’t a debate. It was something rarer in agriculture: an honest reckoning with what’s actually working, what’s failing and what the data can’t yet explain.
The “Drunk Person Milking” Problem
Trimner doesn’t sugarcoat the challenges.
“I had a fellow dairy farmer who milks in robots say that with his Lely robot, he sees that sometimes it’s a little bit like a drunk person milking when it comes to the prep,” Trimner says. “The prep can be good, but it can also miss.”
The issue? Inconsistent teat preparation. The robot’s brushes are supposed to clean and stimulate teats before attachment, but if the cow moves or the robot’s positioning is slightly off, the prep fails — and mastitis risk climbs.
“So, he sets up his settings so that it can do the best job possible, have the best opportunity with the brushes to hit those teats, really do a good job cleaning,” Trimner explains.
Venteicher jumps in: “I don’t have the new lasers yet either, but I’ve heard they’re supposed to scan while they brush. So, they’re supposed to be like a less-drunk person while scrub-cleaning the teats.”
Trimner’s response cuts to the heart of what producers are betting on: “I would like to think that in the next five years, if the robots are not attaching with some human-like performance, I’d be disappointed. Just where AI is going and everything.”
It’s a moment that captures the entire robotic milking proposition: imperfect technology improving fast enough to justify the bet.
The Economics That Changed Everything
Both Trimner and Venteicher built in 2019 — and they know they got lucky.
“The whole thing was done for about $9,500 a cow — manure pits, feed storage, everything,” Trimner says of his 30-robot, 1,800-cow system. “That is not a thing today.”
He’s right. Today, that same project would cost between $15,000 to more than $20,000 per cow, depending on location and whether you’re building new or retrofitting.
But the 2019 timing gave both operations something critical: low interest rates and predictable payments.
“I knew if I built this barn and financed these robots for seven, 10 years, my payment every single year was going to be the same,” Venteicher says. “I knew if I were to hire somebody, that wage for 10, 15 years is going to keep going up every single year.”
The math was brutal and simple: Hiring people wouldn’t guarantee more milk. Hiring robots would.
“And we did get more milk. The amount of milk we gained for our small farm more than made the robot payment,” Venteicher says.
3 Million Pounds Per Person
The labor efficiency numbers are staggering — but they tell different stories at different scales.
Trimner’s operation:
- 1,800 cows on 30 robots
- Six people run the robot barn (not counting feeding or manure, which are identical to the parlor side)
- 300 cows per person
Venteicher’s operation:
- 200 cows on three robots
- Just him and his brother Don
- They ship 6 million pounds of milk annually
- 3 million pounds per person
“For us as a small farm, that’s kind of where we find our niche advantage,” Venteicher says.
But Trimner clarifies an important distinction: “The six people I speak of — I don’t count feeding because that’s identical in the system, and I also do not count sanding because that’s also identical. These six people take care of all the cattle in that barn — cleaning robots, fetching, taking care of robots.”
In other words, he’s comparing robot-specific labor to what it would take to milk 1,800 cows in a parlor. And six people managing that many cows through robots is a number most dairy producers would have called impossible a decade ago.
The Researcher’s Caution
Halbach isn’t here to rain on anyone’s parade. But she’s not convinced the success stories tell the whole story either.
“You don’t necessarily save in labor — your labor changes,” she says. “You get the flexibility back. But what I’m hearing from David is that you’ve got more skilled people now in the barn doing more sophisticated tasks.”
Then she drops a concern that gets at the scalability question: task repetition.
“When you start getting up to 1,000-plus cows in a barn, you now have this module system spread out throughout the facility,” Halbach explains. “Instead of just having a parlor where you’re doing all of your tasks in one area, now you are having those tasks repeated multiple times throughout the barn, which then adds to overall labor.”
Trimner pushes back: “I would disagree a little bit. We do feel that with all the robots in the center, not a lot of work out in the barn per se, the employees are pretty efficient on tasks.”
He mentions they have two golf carts to help employees move around the facility.
The Milk Production Mystery
Halbach raises another concern that’s harder to dismiss: milk per cow.
“We’ve seen farms that were getting over 100 pounds of milk per cow per day, and they go and build these AMS facilities — relatively new, mostly do things right in terms of facility design — and they are having a really hard time being over 90 pounds of milk per cow per day,” she says.
Trimner’s response? His operation is hitting 112 lb. to 113 lb. of energy-corrected milk per day — in both the robot barn and the parlor.
“We have on occasion 200-lb.-per-day cows, and they’re always in the robots,” Trimner says. “It’s hard to reach that capability on a 3x in the parlor versus maybe a 4x in the robot.”
Venteicher doesn’t share specific numbers but hints he’s also exceeding expectations.
The Marshmallow Principle
Venteicher drops one of the conversation’s most memorable insights: culling for “box time” — how long a cow spends in the robot per milking.
“We had a cow, Marshmallow. She averaged for three lactations in a row 155 lb., and her box time was 5.5 minutes,” he says. “My joke on social media was, can you imagine if we had a barn full of Marshmallows? That would be insane.”
Then there was Franny.
“She averaged 130 lb. a milking, but her box time was 15 minutes,” Venteicher says. “In the first six months of robot world, we’re like, ‘I’m sorry, Franny, but you have got to go. This is where we part.’ I can milk three Marshmallows for one Franny.”
It’s a management philosophy that challenges conventional dairy thinking: Don’t just cull for production. Cull for efficiency.
“I think the automation allows a good manager to become a better one,” he says. “But I got to be honest, I think it’ll make a poor manager a worse one if you expect it to do your job for you.”
The Hybrid Model: Problem Cows and Perfect Udders
Trimner’s decision to split his 3,000 cows between robots (1,800 head) and a conventional parlor (1,200 head) is strategic, not indecisive.
“I really like the fact that you can keep the problem cows outside of the robot barn and focus on them in the parlor,” he says. “You can just maximize the efficiency of the robots.”
How does he decide which cows go where?
“We just pick cows based on great udders,” Trimner says. “If she’s a fresh heifer and she’s got a pretty square udder, she goes into the robot.”
Fresh heifers start in the parlor. At 10 to 14 days fresh, those with good udder conformation move to the robot barn. They’re pushed into the robot four times a day for four days to learn the routine. After that, they’re on their own.
Cows with udder issues, health problems or behavioral quirks? They stay in the parlor.
The result? Both systems average 112 lb. to 113 lb. energy-corrected milk per day.
“If I were to build a brand new dairy today of similar size, honestly, if you had like a nice double-eight parallel for the fresh cows and all your problem cows, and the rest were milking robots, I think that’s a really nice setup,” Trimner says.
The Heifer Training Secret
Venteicher has an enviable problem: His heifers almost never need fetching.
“I never fetch that heifer pen, ever,” he says. “We keep them behind the robot until they get trained — sometimes it’s the next day they start going on their own. Sometimes you get a stubborn one; it takes two weeks. But after they start going on their own, I kick them out in gen pop. I never see them again.”
In six years, he can count on two hands the number of times he’s had to fetch a heifer.
Trimner is intrigued. “Do you think your success with the heifers is because you have a higher heifer population, so you can be a little more selective?”
Venteicher agrees genomic culling helps — they’re drying up heifers that test higher than the whole herd average was when he graduated college.
But he credits something else too: his brother Don.
“We are not alike in any way. He probably would have said five words on this whole podcast so far,” he says. “But he raises all the heifers, and he is as slow and quiet as they come.”
Calm handling matters. So might automatic calf feeders.
“I have heard calves that come from the calf feeder do better, and I feel like maybe this experience attests to that,” Venteicher says.
Halbach notes emerging research: “We’ve seen research that calves raised in groups or in pairs have an easier time adapting to novelty, which an AMS would be.”
The Lameness Paradox: No Foot Baths, Minimal Problems
Halbach raises a concern that’s been documented in research: lameness in robotic herds.
“Early on, AMS herds had more lameness rates than conventional herds,” she says. “A lot of that can be attributed to farms making housing decisions that weren’t in line with what our better conventional herds were doing — like putting in mattresses instead of sand.”
Trimner runs reclaimed sand and foot baths twice a week (down from three times recently).
But Venteicher drops a bombshell: “I don’t run foot baths ever.”
He doesn’t recommend it. But he explains why it works for him:
- Automatic alley scrapers running nonstop
- Shorter barn length (114 ft. to 128 ft. per section)
- Rubber matting under the Lely Vector feeding system
- Sand bedding
- Hoof trimming every 40 days
“We run the scrapers nonstop. They don’t have to go very far. They drop in the middle,” Venteicher explains. “I feel like our lameness issues are significantly better moving to the robots.”
He runs the economics every trim using hoof-tracking software.
“Seven trims out of 10, it’ll be like, ‘Nope, you’d spend more money in foot baths than you save,’” Venteicher says.
It’s a counterintuitive finding Halbach can’t fully explain — but the data from the hoof trimmer backs it up.
The ROI Question
Both producers are emphatic: The Lely robots have paid for themselves.
“We’re definitely very satisfied with where we’re at,” Trimner says. “To the point where, because of the way we built the barn, we could throw in 10 more robots and not touch the barn for the most part. It’d be almost financially silly not to make an investment in the future.”
Venteicher is even more direct. “We have our one section of that loan paid off now. But even if it wasn’t, we’re looking at what I would pay an employee per year versus what the robots give me in flexibility and the return on extra milk production.”
The 500-Cow Threshold
Halbach sees a decision point at 500 cows.
“This is where you start making that decision: Do we keep staying in the parallel parlor? Does that make sense as we start to get into much larger herd sizes? Should we go to rotaries?” she says.
Below 500 cows, robots make sense for flexibility — allowing someone to work off-farm, go to kids’ football games, manage a diversified operation.
Above 500 cows, the economics get murkier. Capital costs balloon. Labor doesn’t disappear; it changes. And the comparison shifts to rotaries, not parlors.
Trimner acknowledges the challenge: “If you told me I want to milk 3,600 cows, do I throw in a 72-stall rotary or do I put in 60 robots? Boy, 60 robots would be a lot of money.”
But he’s also intrigued by the possibility.
“A 100-robot dairy doesn’t exist today, but if that was built, boy, I’d love to go see it. That’d be awesome,” he says.
The Line That Says It All
In the end, the most telling moment of the entire conversation might be Venteicher’s emphatic declaration: “If you were to ask me, would I go back to the way we were, it would be a resounding no. I would go beg Courtney or David for a job before I would go back to how we used to have to do it.”
That’s not a ringing endorsement of robots for everyone, everywhere, at every scale.
It’s a deeply personal statement from a producer who found a system that works for his operation, his family, his life.
And maybe that’s the real lesson.
Robotic milking isn’t about whether the technology is “ready” or whether the economics “pencil.”
It’s about whether a farm is ready — ready to manage differently, think differently, cull differently and invest differently.
Trimner and Venteicher were ready. They built systems that work.
Halbach has seen plenty of producers who thought they were ready but weren’t.
The difference between those two outcomes isn’t just about robots.
It’s about everything else.
