How Herd Manager Forecasts Lactation Production
If you've opened a doe's profile in Herd Manager and seen a "Projected 305-day production" number, you might be wondering: where does that come from, and how much should you trust it? This guide walks through exactly how the forecast is built, what makes it more or less accurate, and what you can do to improve it.
The short version
Herd Manager fits a mathematical model called Wood's curve to your test-day milk weights, then projects forward to estimate the total production over a full 305-day lactation. When she has enough of her own records, that's the basis. When she doesn't, the app falls back through three more tiers — using her previous lactations, her dam and grand-dam's lactations, or breed averages — depending on what's available.
Every projection comes with a confidence indicator (a colored dot) so you know how much weight to put on it.
What is Wood's curve?
A goat's lactation has a recognizable shape. She kids, ramps up to peak production over the first 4-8 weeks, then declines slowly across the rest of the lactation. Wood's curve is the dairy industry's standard model for this shape. The math:
Where Y(t) is daily lbs at day t of lactation, and a, b, c are three parameters that control the curve's height, ramp-up speed, and decline rate.
When Herd Manager has at least three test-day records spanning more than 30 days, it fits these three parameters to your doe's actual data. The result is a personalized curve — one that reflects how your doe is actually performing this lactation.
From there, calculating the 305-day total is just the area under the curve from day 1 to day 305.
The four-tier fallback system
Most does don't have a complete dataset. A first-freshener at day 30 has only one or two records. A mid-lactation doe might have five. The model picks the best available data source automatically:
-
Tier 1 — Her own current lactation (high confidence)
When she has 3+ test-day records spanning 30+ days, the curve is fit directly to her data. This is the most accurate tier and the one Herd Manager prefers whenever possible. -
Tier 2 — Her last freshening (medium-high confidence)
If her current lactation doesn't have enough data yet, but she had a previous lactation with good records, the model uses her previous curve as a starting point — adjusted by any current-lactation data we do have. -
Tier 2.5 — Genetic prior (medium confidence)
For first-fresheners and others without prior lactation data, Herd Manager looks at her dam and her sire's dam (paternal grand-dam). If they're in your herd with milk records, their lactation history informs her projection. Standard dairy genetics weighting: 50% from dam, 25% from sire's dam, 25% from breed baseline. -
Tier 3 — Breed averages (low/low-medium confidence)
If no other data is available, the model uses breed-typical curves from a built-in reference table. Accuracy here is rough, and the projection improves dramatically the moment you start logging her test-day weights.
Reading the confidence indicators
Every projection in the app is paired with a colored dot:
| Color | Confidence | What it means |
|---|---|---|
| Green | High | Tier 1 fit. Her own data is doing the work. |
| Light green | Medium-high | Tier 2. Previous lactation data is reliable but historical. |
| Yellow | Medium | Tier 2.5. Genetic prior — directionally useful, but heritability is only ~25-35%. |
| Amber | Low-medium | Tier 3 with some data. Breed average scaled by your few records. |
| Red | Low | Tier 3 with no data. Pure breed average. Treat as rough. |
About the genetic prior
The "Tier 2.5" genetic prior is most useful for first-fresheners — does that haven't lactated yet but whose dam (and ideally sire's dam) you already have data for. The math is the standard breeding-value blend:
0.5 × dam's average production
+ 0.25 × sire's dam's average production
+ 0.25 × breed baseline
This reflects genetic share — a daughter inherits 50% from her dam and 50% from her sire. Since the buck himself doesn't lactate, his dam's production is the standard proxy for his contribution (which works out to 25% of the daughter's expected production).
The remaining 25% is the breed baseline, which captures both random variation and the heritability cap on milk production. Heritability of milk yield in dairy goats is approximately 25-35%, meaning even a perfect genetic prediction would still leave around two-thirds of the variance up to environment, nutrition, management, and luck.
Improving accuracy
Herd Manager's projections get better as your data accumulates. Specific things that move the needle:
- Log test-day weights consistently. Even informal weights (no DHIA test required) help. The more records, the better the curve fit.
- Record at varied points across the lactation. A few weights from days 30-60 are good. A spread from day 30 to day 200 is much better — it lets the curve fit the whole shape.
- Add your dam's lactation history. If your doe's dam isn't in your herd but you have her records (paper milk tests, registry data), enter them as historical lactation data. The genetic prior will pick them up automatically.
- Track sire genetics. Adding the sire to your herd, plus his dam's lactation records, dramatically strengthens projections for first-fresheners — especially when planning future breedings.
- Mark in-heat test days. If a doe is in heat at test, her milk drops noticeably. The app already filters these from curve fits when you flag them.
Frequently asked questions
Wood's curve accounts for the rise-then-decline shape of lactation. A simple average × 305 either over-estimates (if you're calculating it after she's already past peak) or under-estimates (if she hasn't peaked yet). The curve model gives you a number that doesn't depend on where in the lactation you happen to be.
It reveals a chart showing your doe's actual test-day records (orange dots) overlaid on the projected curve (green line). If genetic data is available, you'll also see her dam's projected curve (purple dashed) and sire's dam's curve (blue dashed) for visual comparison. A red vertical line marks today's day-in-milk.
Yes — that's exactly what the next-lactation forecast on the breeding planning view is for. When you have a doe bred and view her pregnancy details, you'll see "Forecast next lactation: ~X lbs" based on either her last lactation, her lineage, or breed averages. This helps with decisions like whether to retain her kids, what to expect from feed budgeting, or whether a particular cross is worth pursuing.
The app falls back to using Nubian curves as the closest middle-of-the-road reference. The projection gets a "your breed not in reference table; using closest" note. If you log her own test-day weights, the system stops needing the reference table at all and uses her data instead.
That's the long-term vision — being able to tell you "does like yours typically produce X-Y lbs" based on data from across the Herd Manager community. It's not in the current version, but the architecture supports it. As the platform grows and more farms log production data, the genetic prior tier will become increasingly powerful.
The bottom line
Lactation projections in Herd Manager are best understood as informed estimates, not guarantees. They get better with data — yours and your relatives' — and worse without. Use them for planning, comparison, and trend analysis. Don't treat them as fixed numbers to budget against.
If you ever see a projection that looks dramatically off for a doe you know well, your data is the answer: log a few more test-day weights, and the projection will pull itself in line.
Track everything you learn
Herd Manager helps you put this knowledge into practice — track FAMACHA scores, schedule hoof trims, record milk tests, and manage your whole herd from any device.
Try Herd Manager Free →