Shelter Insights

Animal shelter data glossary

Plain-language definitions of the shelter data terms used on this site.

Length of stay (LOS)
Days from an animal's intake to its outcome. We count calendar days; same-day outcomes are 0. Shelter Animals Count defines it the same way, for animals that already have an outcome.
Days in care
How long an animal currently in the shelter has been there so far. Unlike LOS, it applies to animals still waiting for an outcome.
Average daily population (ADP)
The average number of animals in care on a given day. Average daily admissions × average length of stay.
Capacity for care
The number of animals a shelter can care for humanely at one time, given housing, staff, resources and available outcomes. See capacity for care, explained.
Capacity for Care (C4C) model
A population management approach from UC Davis Koret Shelter Medicine Program that sets a target daily population and manages stays to stay within it.
Pathway planning
Planning, at intake, the services an animal will need and the outcome it's headed toward, so steps happen without delay.
Population rounds
A regular (often daily) walk-through or meeting where staff review every animal's status, pathway and next steps.
Median
The middle value: half of animals left sooner, half later. Less affected by a few very long stays than the average.
90th percentile
9 in 10 animals left sooner than this. A good measure of the long tail of stays.
Mean (average) stay
Total days in care ÷ number of stays. Use it for cost and capacity math, because every day counts.
Intake types
How an animal arrived: stray, owner surrender, seized or held for custody, transfer in, returned adoption, return from foster, community cat / TNR, owner-requested euthanasia, public assist, born in care, or other.
Outcome types
How a stay ended: adoption, return to owner, transfer or rescue, return to field, foster placement, euthanasia, died in care, or other. We include every outcome type in benchmarks and models.
Return to field (RTF)
Healthy community cats who are sterilized, vaccinated and returned to where they were found.
Foster placement
An animal moving from the shelter to a foster home. Some shelters record this as an outcome, which ends that shelter stay in their data.
Live release rate
The share of outcomes that are live outcomes. Definitions vary; Shelter Animals Count publishes standard ones.
Stray hold
The legal waiting period for owners to reclaim a stray before other outcomes are possible. It sets a minimum stay for strays.
Little's law
A queueing rule: average number in a system = arrival rate × average time in the system. It's the math behind ADP.
Censoring
When recent animals haven't had time to reach an outcome yet. We leave out the last 180 days of intakes from benchmarks so long stays aren't undercounted.
AUC
A score from 0.5 (no better than a coin flip) to 1.0 (perfect) for how well a model ranks animals who will and won't have long stays.
Calibration
Whether predicted chances match reality: of animals given a 30% chance, about 30% should stay over 30 days.
Time-based holdout
Testing a model only on animals that arrived after all its training data, the way it will be used in practice.
De-identified data
Records with names, ID numbers, addresses and owner, adopter or finder details removed.
Interrupted time series
A before-and-after method that compares the months after a change to the trend from before, adjusting for season.