Operations and Technology
AI should eliminate work, not ownership.
Over the last five years multifamily automated much of the leasing office, and site payroll per unit still climbed roughly 25%. Two costs, one job, and nothing came out. A view from inside a 13,000+ unit third-party asset management portfolio.
Current portfolio: 13,000+ units under asset management across 60+ properties and 20+ active markets. Repositioning track record reflects leadership experience across prior institutional platforms.
The Trade
The list of what we handed off is longer than most people realize.
Market surveys. Pricing. Concessions. Renewal offers. Prospect follow up. Review requests and review responses. Resident issues. Maintenance intake and triage. Renewal conversations. Delinquency outreach.
Every one of those is either a relationship or a judgment call, which are the two things we hire on-site teams to do in the first place.
Portions of the trade were fair, and it is worth saying so plainly. These took real work off a human being and did it better than we ever did by hand:
- Package lockers
- Screening and fraud review
- Utility billing
Nobody misses logging packages off a clipboard.
None of this is a skeptic talking. We use AI heavily at CRES and built our own operating system from scratch. I have spent enough of our own capital on this to be annoyed about where the industry took it, and we see the result daily across the 13,000+ units we third-party asset manage.
The Real Cost
The bot never owns the outcome.
The one that gets me most is EliseAI and related products. Not because the product is bad, but because of what it quietly did to accountability. Elise chats. Elise texts. Elise emails. Elise never owns the outcome.
A prospect submits a guest card at 9 p.m. and gets an automated response four seconds later. Nobody on site has actually spoken to someone who wanted to rent an apartment.
Voice AI is the answer I keep hearing to that. It is coming, it is getting better, and I still think it misses the point. When a bot picks up, what is the first thing you do? I am hitting “0” before it finishes the sentence. Nobody has ever hung up with a bot and felt taken care of, and we are about to point that at people deciding where they are going to live.
A leasing agent used to own that prospect. If they did not tour, that was on the agent, and everyone knew it. Now the response time is perfect, the follow up count is stellar, the dashboard is lit up, and the prospect is sitting on a dozen messages from something that has never walked the property. Polite, timely, and identical to what they are getting from the four communities down the street.
That is not follow up. That is spam in a prospect’s inbox and on a property’s leasing calendar. Tour cancellations are up on top of it, which makes sense once you think about who they made the appointment with. Nobody stands up a bot.
So pick up the phone. That is the entire thing.
A 90 second call from someone who knows the property and can hear hesitation in a voice will beat an infinite number of perfectly timed text messages. We all know it, and we stopped doing it anyway.
Higher Stakes
Renewals, delinquency, and the comp shop we gave away.
Renewals and delinquency are the same problem with more on the line. A resident deciding whether to stay another year is having a conversation with a bot. A resident who is behind and probably embarrassed about it is getting an automated nudge instead of a manager who might have worked something out. Those are two of the most important conversations we have all year, and we handed both to software because it was easier than making the call.
Market surveys are another version of this. Site teams used to shop their own comps, and that was never really just about the data. It was training. It was how a manager learned what a unit was worth and why the property down the street was beating us on Tuesdays.
Now nobody has time for it, so we pay a subscription to scrape the same ILS listings the team could have opened themselves, and we call the result data infrastructure. We took a core competency, gave it away, and now rent it back by the unit.
The Economics
Two costs, one job, and nothing came out.
Assumptions Site payroll per unit ranges reflect CRES observed and underwritten figures across our own managed portfolio. They are not a published third-party market index. The illustration below uses a 300 unit asset, the midpoint of each range, and a 5.5% capitalization rate applied to the annual payroll delta. Figures are illustrative and are not a projection for any specific asset.
| Metric | Five years ago | Today | Variance |
|---|---|---|---|
| Site payroll per unit, low end | $1,450 | $1,800 | +24.1% |
| Site payroll per unit, high end | $1,750 | $2,200 | +25.7% |
| Range midpoint | $1,600 | $2,000 | +25.0% |
| Annual payroll, 300 unit asset | $480,000 | $600,000 | +$120,000 |
| Value impact at 5.5% cap | ~$2.2M |
Some of that is inflation, and I want to be fair about it. Wages grew, they needed to, and paying people more to do a hard job is not the problem I am describing.
The problem is what we did on top of it. We layered on a full stack of new expense and never levered the payroll function down to offset any of it. Site staffing models look almost identical to what they did in 2019, except now the software is doing a third of the work and the payroll line is a quarter higher.
The Point
We did not automate leasing. We automated accountability.
What we forgot is that this is a low margin business. Adding expense only works if the productivity gain is large enough to let you take out a bigger expense. If payroll never falls, then the software was never an efficiency investment to begin with. It is margin compression with a nicer dashboard.
If software takes 30% of someone’s job, either that person’s responsibilities expand or the staffing model changes. Otherwise you have simply added another expense line to the P&L.
Over the last couple of years, as this all went parabolic, we told ourselves we were buying efficiency.
What we actually bought was convenience. Those are not the same thing.
Your accountability layer above property management.
CRES holds site teams to a business plan and holds the technology stack to a return. Whether the plan is triage, stabilization, or a full reposition, we run your asset the way we run our own portfolio. Let’s talk about staffing models, expense discipline, and the path to value.
Schedule a Portfolio Review
