Tom Kent is sure of a few things when it comes to dairy farming.
The 24-year-old wants to milk more cows, but he does not want to milk "bad cows".
The second is the importance of data in making informed management decisions.
For Mr Kent, the two are intrinsically linked thanks to herd recording.
"It gives me something to target because I can actually see what I did last year," he said of herd recording.
"I think as far as culling and joining goes, it takes the guess work out of it.
"We know who our best cows are.
"Some of those will get joined to sexed semen.
"Then we know which ones our bad cows are and understanding this ensures we don't keep heifers from the bottom portion of the herd."
Mr Kent manages his family's 340 cow split-calving herd at Lang Lang East in Gippsland.
He returned to the farm two years ago and instigated herd recording, something his parents Stephen and Anne stopped 20 years earlier.
At the time, the cost of herd recording every second month - during a tight year - was difficult to justify.
"It took me a while to come around to it, but I'd read a few articles which stated when times were tough and you keep herd testing it is going to save you money," he said.
"I think I get that now, because every time I sell a cow, I sell the right cow.
"Every time I sell five chopper cows, I know they are the worst five cows, so the herd gets better every time I sell some and that's a good feeling too."
In a poor season, this makes even more financial sense, according to Mr Kent.
"If you have to buy in feed, for example, you are going to get an economic response if you have good cows to utilize that feed," he said.
The Kent's herd - and its meticulous records - is part of DataGene's Ginfo project, the industry's national reference database of genomic information.
For Mr Kent, genomic (DNA) testing would provide a deeper herd dataset and a way to determine an animal's expected performance before it even enters the milking herd.
Herd recording not only plays a vital role in culling and joining selections, Tom also uses it at dry-off to selectively treat cows.
Any cow with a somatic cell count (SCC) of less than 150,000 cells a millilitre at the herd recording three-to-four weeks before drying-off doesn't receive dry-cow treatment.
Those with a SCC more than 150,000 cells/ml receive a dry-cow treatment.
This practice not only saves money, it has also limited antibiotic use.
The Kent's average SCC sits at about 180,000 cells/ml.
At joining, this data also helps Tom select the top 10 per cent of the herd to breed to sexed semen.
The bottom 10pc is joined to Angus semen.
Those joined to beef include carry-over cows or those with a low production index (PI).
Herd recording has also been used to track whole-farm progress as the performance per hectare and per day gets recorded and compared to previous years.
Mr Kent wants to grow the business to 500 cows and become a share farmer.
Increasing cow numbers without decreasing quality will be central to achieving this goal.
He's aiming to breed a "strong" 500-kilogram Friesian with good fertility and milk solid production.
Mr Kent used the DataGene Good Bulls App to select sires because it provided a level playing field to compare local and international bulls.
The Kent's current average herd production is 475kg of milk solids from a diet including: pasture plus 1 tonne per cow of pasture silage, 1.3t/cow of grain and 250-300kg/cow of summer crops - either turnips or rape.
Mr Kent wanted to lift milk solids production to a minimum of body weight and, ideally, up to 120pc with good fertility.
"I find it really exciting where it is all going," he said of herd improvement.
"Every time a heifer comes in from a bull I chose last year, I know it's an elite bull and it is probably a good cow, it is exciting times.
"Every time you get another heifer in the shed it feels like another step forward."