SHRM NORTHERN CALIFORNIA From Screening to Strategy

Key takeaways from Andrew Gardner

Last week I got to do a breakout session at the SHRM Northern California conference in San Francisco, and I'm still thinking about some of the conversations that came out of it.

My session was called "From Screening to Strategy," and I opened with a guess. Almost everyone in the room was already using AI somewhere in their hiring process, and for most people that somewhere was resume screening.

Screening Strategies

Here's the thing though: Resume screening is where most teams start, and a lot of them never get much further.

I spend most of my time with teams hiring frontline and hourly workers at real volume. Healthcare, hospitality, retail, logistics. And what I see over and over is that the slow part of hiring usually isn't the decision. It's waiting. Waiting on a callback, waiting on a reply, waiting for someone's calendar to open up.

Meanwhile, the candidate applied to five other jobs that same night. Whoever actually talks to them first usually wins.

That's why I think resume screening is honestly the least useful place to point AI. Sorting the pile faster doesn't shorten the wait. And for an hourly role, the resume doesn't tell you much anyway. You want to know if they can work weekends, if they can get to the site, if they have the right certification. That's rarely on the page.

So my pitch to the room was pretty simple. “Stop asking AI to judge documents. Ask it to have the first conversation.”

The example I used was Valet Living. They do amenity services for apartment communities in 40+ states, and about 90% of their people are part-time frontline associates. They see 20,000+ applicants a month and make around 25,000 hires a year. That's a lot of phone tag.

They didn't rip anything out. They kept their existing workflow and added automated outreach and reminders, screening that's open 24/7 in English or Spanish, an honest preview of the job upfront, and auto-advancing qualified candidates straight to a recruiter.

The results they reported: screening completes in about 19 hours on average, their recruiters are making 140+ hires a month each, and 75% of phone screens turn into an offer. That last number is my favorite, because it means speed didn't come at the cost of quality. The people getting to recruiters are the right people.

Antonio Fiorentino, their VP of Talent Acquisition, said something I love. “Humanly doesn't sleep. You return, and now you have candidates already scheduled to speak with and have a better conversation.” That's really the whole point. Nobody got replaced. The admin just went away.

A few topics kept coming back up, in Q&A and in the hallway after.

Fairness and Compliance in the age of AI

The first was fairness and compliance, which makes sense in a California room. FEHA's rules on automated-decision systems kicked in last October for employers with five or more people, and the definition is broad. If a tool scores, filters, or ranks candidates, it probably counts, and using a vendor doesn't take the responsibility off your plate. I'm not a lawyer, so please talk to yours. But the point I tried to make is that AI isn't automatically more fair. An unstructured process is biased in ways you can't see. A structured one is biased in ways you can actually measure and fix. Asking every candidate the same questions in the same order is good for fairness and good for compliance.

The second was where to keep a human in the loop. The way I think about it, some stuff should just be automated (reminders, scheduling, status updates). Some stuff AI should help a person do better, like summaries, scorecards, and spotting where the funnel is leaking. And some stuff should never be automated: the final hire decision, turning down a qualified candidate, accommodations, pay conversations. Those stay with a human.

Getting Started

The last one was really about getting started. A lot of the folks in that room were on small teams, sometimes an HR department of one, and the question wasn't whether to do this. It was "okay, but what do I actually do on Monday?" My advice: resist the urge to start with a tool. Spend the first 30 days just getting a clear picture. Understand how long it takes to first contact a candidate, and where people are dropping off. Then pick one role at one location, change one thing, and compare it against where you started. Small teams can move fast here. You don't need a big rollout plan, you just need one honest baseline and one pilot.

If you only remember three things from my session, I'd want it to be these.

1. Automate the waiting, not the judging.
2. Structure is your fairness strategy and your compliance strategy at the same time.
3. Pick one type of role, one pilot, one baseline, and start in the next 30 days.

Big thanks to the SHRM NorCal team for having me, and to everyone who came by and stuck around to chat after.

If you're hiring hourly or frontline folks at volume and want to talk through where things are getting stuck, shoot me a message. Always happy to compare notes.

LinkedIn post to accompany this:

Last week at the @SHRM Northern California conference, I asked a room full of HR leaders where they were using AI in hiring.

For most, the answer was resume screening.

That’s where many teams start. It’s also where too many stop.

For frontline and hourly hiring, the biggest problem usually isn’t how long it takes to review a resume. It’s all the waiting that comes next:

- Waiting for a callback- Waiting for a response- Waiting for a recruiter’s calendar to open

Meanwhile, that candidate applied to five other jobs. The employer who starts a real conversation first often wins.

My message during the session was simple:

Stop asking AI to judge documents. Ask it to have the first conversation.

That means using AI to reach candidates quickly, ask consistent job-relevant questions, answer questions, schedule next steps and keep the process moving, even when recruiters are offline.

We talked about how @Valet Living approaches this across a workforce that is roughly 90% part-time frontline associates. With more than 20,000 applicants a month, they added automated outreach, 24/7 screening in English and Spanish, realistic job previews and automatic advancement for qualified candidates.

The result? Screening completes in about 19 hours on average, recruiters make more than 140 hires per month each, and 75% of phone screens lead to an offer.

The goal isn’t to remove people from hiring. It’s to remove the administrative work that keeps people from having the conversations that matter.

If I had to boil the session down to three ideas:

1. Automate the waiting, not the judging
2. Use structure as both a fairness and compliance strategy
3. Start with one role, one location and one measurable pilot before buying any software

Thank you to the @SHRM NorCal team and to everyone who stayed afterward to compare notes.

I shared more from the session here: [BLOG LINK]

If you’re hiring frontline or hourly workers at volume, where does your process slow down most?