Write the ad
Most caregiver ads say the same three things: "compassionate," "reliable," "team player." None of that gets you a shortlist, it gets you a hundred nearly-identical applications with nothing to sort them by. The ad that actually works states the things a caregiver needs to know before they apply, not the things you wish were true about them.
Put the shifts in the ad, not "flexible hours." Say Tuesday and Thursday mornings, or overnight weekends, whatever it actually is. Put a pay range, even a wide one, because "competitive pay" filters out exactly nobody and wastes everyone's time on the phone finding out it's below what they need. Say whether the role needs a car and a valid license, because transport is one of the first things that eliminates a candidate and it should eliminate them before they apply, not after you've called them.
State the checks you require up front too: registry check, background screening, TB test, CPR. Not because anyone reads a job ad and decides not to apply over a background check, they don't, but because it sets the applicant's expectation for what happens between "hired" and "first shift," and it gives you something to check the application against later.
If writing that from scratch every time gets old, the job description template has the state credential block already built in, editable per state, so the ad states the right checks for Texas or California or Florida without you having to remember all of them from the last time you posted.
Post it where the caregivers are
Indeed's free listing is still where most of the caregiver labor pool actually looks, and it costs nothing to post. It gets you into search and the app the same as a sponsored ad, just without the ranking boost. That's usually enough for an independent agency that isn't trying to fill twenty roles a month. See what the free post gives you, and what it doesn't.
Your own careers page matters more than it sounds like it should. It's the one place an applicant can see every open role at your agency at once, and it's the form that catches whoever clicks through from Indeed, from a referral, from a Facebook group, from anywhere. Applications from every source land in the same list instead of scattered across an inbox, a phone number written on a sticky note, and whatever Indeed shows you that week. See the home care page for what that looks like for an independent agency specifically, rather than the multi-location version most ATS tools are actually built for.
Referrals are still the best source you have and the easiest to neglect. A current caregiver who refers a friend is vouching for someone who already knows the work is hard. Ask for referrals the same week you post the ad, not as a last resort three weeks in when the ad's gone cold.
Screen from the inbox
However the applications arrive, the sorting method is the same: decide the five things the shift actually needs before you open a single application, then score each one 0, 1 or 2. Registry check stated, availability that matches the shift, transport, the right experience type, and distance that doesn't eat the shift in a commute. The full method is here, but the short version is: sort by total, ring the top ten, call those first.
This is also the point where software genuinely helps, not by deciding who you hire, but by doing that first pass before you've opened the inbox. First Shift reads every application against the job you posted and scores it 0 to 100, with the strengths and gaps written out in plain English. If an applicant never mentions a registry check, it says "not stated," not a guess in either direction, because that's the honest answer and it's still your call to make. It doesn't run the registry check itself, it flags when the applicant hasn't told you about one.
The scoring works the same whether ten applications come in or a hundred, which matters more than it sounds like it should. The agency that gets thirty applications on a quiet week and the one that gets a hundred and twelve are running the same method, just at different volumes, and neither one benefits from reading in arrival order.
Interview
A caregiver interview that's just a conversation tells you whether you liked them. It doesn't tell you whether they'll show up to a dementia care shift and handle a resident who's agitated at 2am, or what they do when a client's family member is difficult, or whether they'll actually call you before a no-show instead of just not turning up.
Structure the interview around five question areas instead: ADL scenarios (can they walk you through helping someone bathe, dress, transfer), dementia behaviors (what do they actually do when a resident is confused or agitated), reliability and no-shows (what's happened before, and how did they handle it), boundaries (what do they do when a client or family member crosses a line), and incident reporting (do they know what gets reported, and to whom). The interview scorecard turns those five into ten structured questions with 1-5 anchors, weighted, with room to score three candidates side by side so the interview isn't just a feeling you write down afterward.
Booking the interview matters as much as running it. A lot of caregiver no-shows aren't really no-shows, they're applicants who took another agency's shift while yours was still deciding when to call. Put a real time in the first message, not "we'll be in touch," and the interview you booked is more likely to still be needed by the time it happens.
Check before the first shift
The checks before a caregiver's first shift vary by state, and there's no federal rule for private-pay home care, so this is the part every agency already does by hand, and every agency does slightly differently. A few examples of what agencies in these states already check:
- Texas checks the Nurse Aide Registry and the Employee Misconduct Registry before first contact with a client, runs a criminal history check as well, and re-checks yearly.
- California checks through Live Scan, a CDSS clearance, and a Home Care Aide Registry listing.
- Florida runs an AHCA Level 2 background screening.
Most states also expect a current TB test, CPR or first aid certification, a valid driver's license and insured vehicle if the role needs one, work authorization (I-9), and whatever training hours your state mandates. None of this is legal advice, check your state's current rule before you rely on any of it, because registries and thresholds change and a summary on a blog post won't catch that.
This is also the list First Shift's fit score checks an application against: it flags what's missing or unstated so you're not the one re-reading every application by hand to catch a gap. It doesn't run the registry check, the background check, the TB test or the CPR verification itself, and it never will pretend to. Those stay yours.
Keeping the dates themselves straight is its own job once you've got more than a handful of caregivers on the books. A TB test from eighteen months ago, a CPR card that lapsed in March, a registry re-check that was due last week, none of that shows up until a state surveyor or a client's family asks. The credential tracker keeps a date per caregiver per check and goes amber at 90 days out, orange at 60, and red at 30 days or past, which is the same warning a spreadsheet gives you if someone remembers to update it every week. Most don't.
Onboard
First shift goes wrong more from nobody saying where to park or who to call than from the caregiver not knowing the job. A short first-shift checklist fixes most of it: confirm the address and any access instructions the day before, send the client's care plan and any specific preferences, give a direct number for the office and an after-hours number if something goes wrong, and check in partway through the first shift, not just at the end of the week.
None of that is complicated. It's just easy to skip when you're already moving on to the next open shift, and the caregiver who has a rough, unsupported first day is the one most likely to be gone by the second.
Keep them
Median caregiver turnover across the industry runs about 75%, according to Activated Insights' benchmarking, which means the hiring described above isn't a once-a-year project, it's most of the job, most of the year. The same benchmarking puts the median agency funnel at roughly 800 applicants to 195 interviews to 63 hires, a 7.9% conversion from application to hire, and that's on the larger end of the market. An independent 10 to 40 caregiver agency works the same funnel at a smaller scale, which is exactly why the sorting and reply-time habits above matter more, not less, when every application is a bigger share of your total pool.
Reply time keeps mattering after the hire, not just during it. Caregivers who get steady, predictable hours and a fast response when they raise an issue stay longer than caregivers who have to chase the office for a schedule. None of that is a software problem. It's the same discipline that got the application answered in an hour, applied to everything that comes after.
None of this costs much to start doing right. See how the whole workflow fits together or what it costs if you'd rather see the software version of this playbook than build the spreadsheet version yourself.