As CDCB implements data quality improvements this month, industry experts see preparations for a fundamental shift in how genetic evaluations could work within three years
EXECUTIVE SUMMARY: Tuesday’s genetic evaluation delivers four technical changes that signal dairy breeding’s shift from traditional statistics to artificial intelligence. CDCB is eliminating seven years of duplicate health records and 1.1 million outdated type evaluations—data quality issues that have subtly influenced every breeding decision since 2018. While most PTAs will barely budge (0.97-0.99 correlations), producers using bulls whose daughters transfer between elite herds should expect adjustments, especially for milk fever resistance. The February 2026 mandatory switch to HTTPS authentication isn’t just a security upgrade—it’s infrastructure for AI systems that need controlled data access. These December modifications, following April’s disruptive base changes, confirm that genetic evaluations are being systematically rebuilt for machine learning integration that industry experts project will revolutionize breeding accuracy by 2028.
You know, Tuesday’s December triannual evaluations from CDCB will bring more than just updated rankings and PTAs. Here’s what’s interesting—when you dig into these “operational efficiency improvements” they announced on November 5th, there’s something bigger happening beneath the surface.
The changes themselves sound routine enough. They’re eliminating duplicate health records, removing 1.1 million outdated type evaluations, optimizing data pipelines, and transitioning from anonymous FTP to secure HTTPS access. But after talking with folks across the industry and really examining the details, what I’ve found is we’re seeing essential groundwork being laid for machine learning systems that could reshape how we make breeding decisions in the next few years.
THE FOUR KEY CHANGES:
- Duplicate health records eliminated
- 1.1 million pre-1998 type records removed
- Pipeline processing optimized
- FTP to HTTPS transition (Feb 2026)
KEY DATES TO REMEMBER
- Dec. 10, 2025: December genetic evaluation release
- Feb. 2, 2026: Anonymous FTP access ends
- 2026-2028: Projected AI implementation timeline (based on current research trends and industry analysis)
Understanding December’s Technical Changes
So here’s the thing—the CDCB December 2025 genetic evaluation changes include four primary modifications that the CDCB team—Kristen Gaddis, Sam Comstock, Jason Graham, Ezequiel Nicolazzi, Jay Megonigal, and Frank Ross—describe as refinements to improve data quality and system efficiency. Let me walk you through what’s actually happening and why it matters for your breeding decisions.
Health Record Deduplication
What’s particularly interesting about this first change is that CDCB discovered cows transferring between herds during lactation were generating duplicate health records, artificially influencing disease resistance PTAs for their sires. Makes sense when you think about it—when a cow moved from one farm to another mid-lactation, both DHI systems could report the same health events. Double-counting in the national database.
Testing with August 2025 data revealed correlations between old and new health PTAs ranging from 0.97 for milk fever resistance to 0.99 for displaced abomasum and metritis. Now, while these correlations suggest minimal population-level impact, individual bulls whose daughters frequently transfer between elite herds may see more significant adjustments. Worth checking if you’re using popular genomic young sires.
You know what I find fascinating? Milk fever showed more variation than other traits, and there’s a biological reason. Research by Santos and colleagues published in the Journal of Dairy Science shows that subclinical hypocalcemia affects up to 60% of dairy cows in the first three days postpartum. That 0-4 day window—when cows are coming fresh and hypocalcemia typically occurs—coincides precisely with when many elite heifers transfer between herds. And with multiparous cows experiencing such high incidence rates, plus elite genetics programs routinely testing blood calcium levels… well, these duplicate records really messed with milk fever evaluations more than other health traits.
Historic Type Record Removal
This one’s interesting when you understand the history. Holstein Association USA and CDCB are removing 1.1 million animals born before 1998 from genomic type evaluations. These animals never qualified for traditional type evaluation but remained in the system through “supplementary evaluations”—basically, statistical adjustments that provided slightly better predictions than parent averages alone.
Now, in the pre-genomic era, these supplementary evaluations made perfect sense. They added real value. But today? With Holstein’s type reference population exceeding 750,000 animals and genomic predictions achieving 60-70% reliability without any phenotypic data, these outdated records just contribute noise. Testing showed a 99.99% correlation between evaluations with and without these records. So their removal won’t disrupt your breeding program.
Pipeline Modernization
I know this sounds technical, but bear with me—it actually matters. Recent updates to Interbull’s international evaluation schedule now deliver MACE (Multiple-trait Across Country Evaluation) results on Day 1 of CDCB’s genomic cycle, rather than partway through the cycle. So basically, CDCB can now eliminate redundant processing steps and incorporate international data earlier in the evaluation sequence.
Testing showed correlations exceeding 99.9% across all traits and breeds between old and new pipelines. Most of the variation came from slightly fresher international data influencing domestic evaluations—which is actually an improvement if you’re using international genetics, as many Upper Midwest operations are these days. Those Pennsylvania tie-stall operations importing Canadian genetics will particularly benefit from this fresher data integration.
Access Control Implementation
Starting February 2, 2026, CDCB will discontinue anonymous FTP access and require all users to authenticate via HTTPS. Sure, it provides enhanced encryption and improved access control. But here’s what matters for us as producers: CDCB will now know exactly who’s accessing genetic evaluation data, when, and how frequently.
If you’re using any third-party services that pull CDCB data—and let’s face it, most progressive operations are—you’ll want to verify they’ve secured authenticated access before February. I’ve been hearing from several consultants who haven’t even started this transition yet. A Wisconsin producer mentioned his consultant hadn’t even heard about the HTTPS change yet, so don’t assume anything.
The Timing Strategy: Why December, After April’s Major Changes
You might be wondering—I certainly was—why CDCB is implementing more changes in December, eight months after April’s significant genetic base change and Net Merit formula revision already shook things up.
Here’s what I’ve learned: April 2025 delivered the most comprehensive changes to U.S. evaluations we’ve seen in years. CDCB’s documentation shows PTAs reset to 2020-born cow averages, resulting in drops of 50 pounds of fat, 30 pounds of protein, and 2.5 months of productive life. And simultaneously, Net Merit weights shifted substantially—butterfat emphasis increased 11%, feed efficiency jumped 48%, protein emphasis dropped 33%.
So why add more changes now? The pattern suggests several strategic objectives:
- Separating data quality improvements from formula changes prevents confusion about what caused specific ranking shifts
- We had eight months to adapt to new Net Merit weights and base adjustments before facing additional modifications
- CDCB likely used the April-December period to identify and resolve issues that only became apparent after the base change
What We Learned from April’s Reset
You know, watching bulls that showed +2500 milk in December 2024 evaluations suddenly display +1800-1900 in April 2025 was jarring for everyone. Same genetics, completely different numbers after the base reset. It took months to retrain our eyes about what “good” looks like.
The Net Merit formula revision proved equally challenging. Bulls that ranked highly under the old system’s protein emphasis suddenly fell behind competitors with superior butterfat and feed efficiency profiles. What we all learned—sometimes the hard way—is that genetic merit isn’t absolute. It reflects current economic priorities that change with market conditions.
And different operations are adapted differently, as you’d expect:
- Large-scale operations in California and the Southwest, focused on component production, generally transitioned smoothly to the butterfat emphasis
- Grazing-based operations in Vermont and Wisconsin that traditionally prioritized protein for cheese market premiums had to reconsider their breeding strategies completely
- Those New York and Michigan operations with mixed market contracts found themselves recalibrating somewhere in between
These regional differences still matter as we navigate the changes in December.
What These Changes Reveal About Data Quality
Looking at December’s modifications, what strikes me is how long these data quality issues persisted before being addressed.
Duplicate health records from herd transfers have apparently influenced evaluations since the launch of health traits in 2018. That’s seven years of subtle contamination affecting our breeding decisions on disease resistance. Similarly, pre-1998 type records influenced genomic predictions throughout the genomic era from 2009 to 2025, affecting every breeding decision that incorporated type traits via Net Merit, TPI, or custom indices.
“Genetic evaluation systems are inherently conservative about implementing changes, even when problems are suspected. Given the stakes—every evaluation affects thousands of breeding decisions worth millions of dollars collectively—this caution makes sense. But it also means known issues can persist for years before resolution.”
The Hidden Story: Preparing for AI-Powered Evaluations
Here’s what I think many of us are missing: December’s changes serve a dual purpose beyond correcting historical problems. They’re establishing infrastructure for artificial intelligence and machine learning systems that could transform genetic evaluations sooner than we think.
Clean Data for AI Training
Having worked with data scientists on various projects, here’s what I’ve learned—machine learning algorithms need pristine training datasets. The duplicate health records being eliminated? They’d propagate errors exponentially in AI models. Those 1.1 million outdated type records would introduce inconsistencies that deep learning systems just can’t tolerate.
Dr. Victor Cabrera from UW-Madison’s Dairy Brain Initiative has some fascinating perspectives on this. Modern neural networks for genomic prediction show promise for improved accuracy compared to traditional methods—there’s a 2023 review in Frontiers in Genetics by Chafai and colleagues exploring various machine learning models, though specific performance improvements vary by trait and population.
“What CDCB calls ‘operational efficiency improvements’ are actually essential preprocessing for AI implementation. Data quality is everything.”
Real-Time Evaluation Infrastructure
The pipeline optimization isn’t just about speed. It’s about enabling continuous-learning AI systems that can update predictions in real time as new data flows in. Companies like EmGenisys are already demonstrating this with AI-powered embryo viability evaluation. CDCB’s infrastructure changes create similar potential for genetic evaluations.
Controlled Access for AI Development
The shift from anonymous FTP to authenticated HTTPS gives CDCB visibility into who’s developing AI models with their data. With over 11 million genotypes in the National Cooperator Database as of June 2025 and roughly 100 million lactation records… that’s extraordinary value for machine learning applications. Controlled access becomes essential for both security and potential commercialization.
Industry Perspectives and Cost Considerations
The reactions I’m hearing from industry stakeholders have been really interesting:
- Data service providers are emphasizing the technical challenges—hundreds of automated scripts need rewriting for the authentication transition
- Advisory services, especially smaller consultants who’ve relied on simple FTP downloads, worry about increased administrative requirements
- Academic researchers note that graduate students who previously accessed data instantly now need formal data-use agreements
Looking at what this might mean for your operation, industry sources suggest these changes could involve various costs—though actual expenses will vary significantly:
- HTTPS authentication setup: $500-1500 in one-time programming costs if you’re working with a consultant transitioning from FTP
- AI-literate genetic advisors: Generally commanding 15-25% premium over traditional advisors—roughly $150-200 additional per consultation
- Future AI evaluation subscriptions: Based on similar ag-tech services, we’re probably looking at $50-150 monthly for basic access to $500+ for premium predictive analytics
WHAT THIS MEANS FOR YOU The real question isn’t whether these changes matter—it’s how quickly you can adapt to maintain your competitive edge as genetic evaluations evolve from traditional calculations to AI-powered predictions.
Practical Implications by Operation Type
Let me break down what December’s changes mean for different types of operations:
Elite Genetics Programs
If you’re marketing high-genomic females, using contract heifer growers, or exporting genetics internationally, you likely had higher exposure to duplicate health record issues. Bulls heavily used in these programs may show more variation in health PTAs this December—on top of adjustments you’re still processing from April.
What you should do:
- Compare December health PTAs against both pre-April and post-April baselines to understand cumulative impacts
- Pay particular attention to milk fever resistance (which showed the most variation during testing)
- Success here means maintaining your genetic progress rates despite evaluation adjustments
Data-Dependent Operations
Farms relying on third-party software, consultants, or services that access CDCB data via FTP need immediate action.
Your action items:
- Verify providers have secured HTTPS access before February 2026
- Don’t assume compliance—I know several consultants who haven’t started the transition
- Document current data access methods as backup
Technology-Forward Operations
Progressive dairies should recognize December’s changes as early indicators of the AI transformation coming to genetic evaluations.
What to focus on:
- Build relationships with AI-literate genetics advisors now
- Invest in farm data quality—every accurate record improves future AI predictions
- Start budgeting for potential AI evaluation subscription costs
Understanding the Broader Context
December’s refinements are happening within a rapidly evolving dairy genetics landscape that’s still adjusting to April’s disruptions. Genomic testing volume continues to expand—CDCB processed its six millionth genotype in February 2022, and by January 2023, the database contained over 7 million genotypes. What’s really interesting? 92% of those are from females.
Novel traits like feed efficiency and methane emissions gained real prominence following April’s 48% increase in Feed Saved emphasis. International competition keeps intensifying as global genetics companies leverage advanced analytics. And technology adoption—sensors, robotics, precision management systems—is becoming standard on progressive operations from California’s Central Valley to Pennsylvania’s tie-stall barns.
These trends are creating demand for sophisticated evaluation systems that can integrate diverse data streams and deliver real-time insights. Traditional linear models can’t provide these capabilities, but AI systems potentially can. The genetic evaluations we rely on today may look primitive compared to what’s coming.
Looking Ahead: The Next Three Years
Based on current research trends and what I’m seeing in the industry, here’s what we might expect:
2026: Foundation Building
- We’ll likely see continued data quality improvements framed as technical maintenance
- First commercial AI tools for specific traits—mastitis prediction, feed efficiency optimization—should hit the market
- Universities will start publishing research using CDCB’s cleaned datasets for deep learning models
2027: Parallel Systems
- I expect AI evaluations will run alongside traditional models for validation
- Early adopters will begin incorporating AI predictions into breeding decisions
- CDCB might announce pilot programs for enhanced evaluations, following patterns from their Producer Advisory Committee, founded in 2019
2028: The Transition
- AI predictions could become primary, with traditional models serving validation roles
- Genomic prediction accuracy is potentially improving significantly, with preliminary machine learning studies showing trait-specific gains
- At that point, evaluation interpretation will require specialized expertise that most of us don’t currently have
As Dr. Cabrera suggests, producers who understand this trajectory and begin preparing now will likely maintain competitive advantages.
Key Takeaways for Dairy Producers
As we approach Tuesday’s evaluation release and continue adapting to April’s major changes, here’s what I think matters most:
Immediate Actions
- Check high-value animals: Compare December PTAs against both pre- and post-April baselines to understand cumulative impacts
- Verify data access: Confirm all third-party software and consultants have secured HTTPS access before the February deadline
- Document current PTAs: Track how successive changes affect your genetics
Strategic Considerations
- Invest in data quality: Research from Weber and colleagues demonstrates that data quality is the primary factor determining AI model accuracy
- Build technology literacy: Understanding AI basics will likely become as essential as understanding EPDs became in the 1990s
- Maintain flexibility: April showed us that long-standing assumptions can change rapidly
Long-Term Planning
- Accept temporary stability: Technology and economics drive continuous change in evaluations
- Focus on principles: Genetic principles matter more than specific numerical values
- Prepare for subscriptions: AI-powered evaluations probably won’t remain free public services forever
The Bottom Line
As these CDCB December 2025 genetic evaluation changes take effect, dairy breeding decisions will increasingly rely on clean data and sophisticated analysis. These changes represent more than routine maintenance—they’re essential preparations for what could be a fundamental transformation in dairy breeding. While CDCB frames these as “operational efficiency improvements,” coming eight months after April’s disruptive base change and Net Merit revision, the pattern seems pretty clear: the industry is systematically upgrading infrastructure for next-generation evaluation systems.
For those of us still adjusting to April’s new reality—where historical benchmarks shifted dramatically, and component emphasis changed substantially—December’s modifications might feel like one more thing to deal with. But you know what? These changes are actually stabilizing forces, addressing long-standing data quality issues while preparing systems for future improvements.
What strikes me most is that success in this evolving environment won’t require becoming a computer scientist. But it will demand openness to continuous change, investment in data quality, and strategic partnerships with advisors who understand both traditional genetics and emerging technologies.
The December 2025 evaluation changes, following April’s significant adjustments, confirm that transformation is accelerating—not slowing. Those who recognize this trajectory and adapt accordingly will discover opportunities within the evolution. The future holds exciting possibilities for operations ready to embrace precision breeding powered by AI-enhanced evaluations.
The opportunity—and the responsibility—rests with each of us as dairy producers. We need to embrace change while maintaining focus on what matters most: breeding better cows for profitable, sustainable dairy farming in an era of continuous innovation. That’s what I’m taking away from all this, and I hope it helps you navigate these changes in your own operation successfully.
KEY TAKEAWAYS:
- URGENT: Verify all genetics services have HTTPS authentication before February 2, 2026—failure means lost data access
- PTAs TUESDAY: Most bulls unchanged, but check elite sires with transferred daughters for milk fever adjustments
- HIDDEN FIXES: CDCB eliminated 7 years of duplicate health records + 1.1M obsolete evaluations contaminating your decisions
- FUTURE READY: December’s cleanup enables AI breeding systems projected to boost prediction accuracy 10-25% by 2028
Complete references and supporting documentation are available upon request by contacting the editorial team at editor@thebullvine.com.
Learn More:
- Fertility Bulls Failing? Your PTAs Are 30% Inflated – Here’s the Fix – Dives deeper into the “data quality” theme, revealing why current fertility PTAs are systemically inflated and offering immediate strategies to adjust your breeding program for real-world results while awaiting systemic fixes.
- Is AI The Next Frontier in Dairy Breeding? – This article directly supports the main piece’s thesis by exploring the specific machine learning models being tested and how AI can overcome the limitations of traditional statistical methods in genomic selection.
- Genomic Testing: A Producer’s Guide to Smarter Breeding Decisions – This guide provides the tactical “how-to” for implementing on-farm genomic testing, helping you optimize the quality of the exact data that will soon be used to train the next generation of AI evaluations.
Join the Revolution!
Join the Revolution!Join over 30,000 successful dairy professionals who rely on Bullvine Weekly for their competitive edge. Delivered directly to your inbox each week, our exclusive industry insights help you make smarter decisions while saving precious hours every week. Never miss critical updates on milk production trends, breakthrough technologies, and profit-boosting strategies that top producers are already implementing. Subscribe now to transform your dairy operation’s efficiency and profitability—your future success is just one click away.
