It is vital that fertility key performance indicators (KPIs) for every herd are monitored to track performance and identify weaknesses to optimise fertility.
But with so much data now available at the touch of a button, exactly what information should you dismiss and what metrics should you really focus your attention on?
Vet Stuart Russell, from Nantwich Farm Vets in Cheshire, answers some of these key questions and also takes a look at what targets should be depending on the software you use.
What do you need from a KPI?
- Tells the truth (rises reliably when fertility improves and falls reliably when it deteriorates).
- Reflects the present. Last year’s success is irrelevant today unless it helps predict tomorrow.
- Changes quickly, but not in response to luck. If a KPI tracks luck (or “cow-level effects”) too closely, it can’t teach you anything about your herd management.
What not to monitor
At best, it reflects a herd’s fertility 12-21 months ago and, at worst, it lies.
If you save a few stale cows from the barren list, your calving interval will go up (eventually), yet fertility has improved.
Calving interval should be locked securely in the history books. It has no valid purpose.
100-day in-calf rate and related “snapshot in lactation” metrics (for example, 200-day not in-calf rate or percentage served by 80 days in-milk).
Based on monitoring strategies for seasonal herds, these KPIs work well with tight calving patterns, but are heavily affected by “luck” when monthly calving numbers aren’t large.
To make them useful, these KPIs are normally averaged over multiple months or even a full year, making them slow to change. They should only be used when we have nothing better.
Conception risk (often ‘rate’)
Confusingly, some UK advisers call this “pregnancy rate”, which is different from the similarly-named USA KPI (below).
All but the largest herds should not monitor conception risk as it is heavily influenced by which cows were served that week when insemination numbers aren’t huge.
Conception risk becomes useful when averaged over a longer period, especially when troubleshooting.
Higher doesn’t always mean better: you could achieve a 100% annual conception risk if you only ever served one cow per year.
Which KPIs are crucial?
21-d pregnancy risk (or rate) by calendar date
Similar terms: fertility efficiency or % eligible for service which conceived
Relevance: did I generate enough pregnancies in this 21-day calendar period?
Target: using the default DairyComp 305 version of 21-d pregnancy risk, the classic target is more than 21%, but those using Uniform-Agri should add 2-3% to their target.
Other versions of this metric need lower targets. Changing the voluntary waiting period (VWP) used in any of the calculations means higher targets are required, but this is a common trick to make a herd’s 21-day pregnancy rate look awesome.
This is because cows inseminated before the VWP should not be eligible, but this is a mistake some of the software packages make – cows that become pregnant before the VWP are counted as eligible and this distorts the numbers.
The best in the world truly achieve more than 32% with the default DairyComp 305 version, but a net financial return from chasing extremely high figures is not guaranteed.
21-day insemination risk (or rate)
- Similar: 21-day submission rate, if used to refer to an entire lactation, not just VWP plus 21 days.
- Relevance: did I inseminate enough cows in this 21-d calendar period?
- Target: more than 60-65% depending on reproductive strategy, the software and the voluntary waiting period used in calculations.
Which KPIs for very quick feedback?
Weekly number of pregnancies produced
- Relevance: quick and crude feedback on overall fertility.
- Target: more than 1.5 pregnancies per 100 cows per week, but this assumes a flat calving. pattern and no fluctuations in fertility. The 21-day pregnancy rate is more reliable but takes two weeks longer to become available.
Percentage pregnant of those presented at pregnancy diagnosis (PD)
- Relevance: quick and crude feedback on heat detection (normally).
- Target: more than 70% at 32-39 days after insemination. The longer you allow yourself to see repeats, the fewer negatives should reach PD. Usually, lower figures reflect poorer heat detection/expression, although occasionally they reflect extremely low conception rates.
Why do software packages differ when it comes to pregnancy rate?
It all comes down to how you define an eligible cow. This should be simple:
- Not yet pregnant
- After VWP
- Not marked DNB (barren)
However, the matrix has to take account for a number of scenarios. For example, let’s take the 1-20 January window.
All of these animals have to be accounted for:
- Cow passes VWP on 5 January (she is eligible for 15 days)
- Cow passes VWP on 19 January (eligible for one day)
- Cow passes VWP on 19 January, but is served on 15 January (before VWP is up)
- Cow marked DNB on 8 January (eligible for first seven days)
- Cow aborts on 25 December (does she become eligible?)
- Cow that was diagnosed pregnant on 15 December is then found to be empty on 2 February. (Should she have been eligible for the whole period?)
You can see by the example above, designers of the software packages have lots of complexities to build into a model.
- Although crude and sometimes misleading, pregnancy hard count and the percentage pregnant at pregnancy diagnosis have important roles as early warning indicators.
- Otherwise, 21-d pregnancy rate (or similar) is the fundamental fertility KPI for non-seasonal dairy herds, and every herd manager needs to know this figure for both the last year and the most recent calendar 21-day window.
- There are at least six completely different definitions of pregnancy rate, and at least three for submission rate. Be careful – they may not be speaking your language.
- Compare like with like. Different software packages define “eligible for service” differently. The 21-d pregnancy rate and 21-d insemination rate are also very sensitive to the voluntary waiting period used in the calculation. Don’t try to compare different versions of these KPIs.
- The versions produced by different software packages are often not directly comparable, but as always, the most important comparison is against yourself.
- Find a software package that most accurately affects herd economics.