AI Maintenance Math: Quantifying the Load Lifted Off Your Team

🎧 Listen to the Audio Version Below

A conversation about door counts vs headcounts vs quality of life broke out on LinkedIn last week:

Laura Bruyere asked if the multifamily rule of thumb of one employee per 100 units is still relevant- challenging the notion that it should be lower because of growing complexity in that asset class.

Peter Lohmann weighed in, saying the multifamily asset class is so bloated with overhead that they wouldn’t be able to hang with SFR managers.

Yet he put forth a tougher baseline: ~1 employee per 60 single-family homes.

Not gonna lie- this debate fascinates me.  Not because I have any horse in the race between who can do it better. 

But because the historical assumptions around headcount needed to properly manage the living environment, and more importantly, the quality of life for those who choose that as a profession, are at the center of what I have been questioning since joining Vendoroo.

I want to make it very clear that I have a limited scope of view into the matter.  Laura and Peter are discussing full-time work per employee per door for the entire scope of managing the property. 

I am discussing only maintenance.  But still-AI is definitely changing the old math.

The Moment the Math Flips

When our Chief PM, David Normand,r eviewed the date, he expected to see faster response times and tighter documentation. But he found a reduction of work burden.

Across teams joining our network, a familiar baseline shows up in their historical data: about 17% of doors have an open work order at any given time. After AI starts coordinating- intake, triage, vendor scheduling, resident/owner updates, and the dreaded “did anyone follow up?”- that Point In Time Open WO % trends toward ~10% (often 9–10%). 

That’s a ~41% relative reduction in residents living with an unresolved issue at any given moment. 

The effect of this became the most obvious: when we added 10,000 doors, the system and the Total Point In Time Open WOs still went down from roughly 2,800 to ~2,200. In other words, AI wasn’t just helping humans go faster; it was quietly eliminating the wait states that keep tickets stuck.

That’s the value to your bottom line.  But the value to your team is eliminating that mental burden of the neverending busywork. What we call the Forever Firefight.

Life on the other side of “busy”

Picture a 300-door portfolio on a normal Tuesday.

In the old rhythm, about 17% of those doors have an active work order at any moment. That’s 0.17 × 300 = 51 open WOs. If each WO typically takes ~18 touchpoints (intake, triage, WO creation, vendor selection/coordination, follow ups, etc.) to close, your team is carrying 51 × 18 = 918 touches across a ~6-day completion window, about 918 ÷ 6 = 153 human touches per day. Death by to-do list.

Now fast-forward to an operation that’s cleaned up the wait states so the point-in-time % open trends toward ~10%. Same 300 doors, now 0.10 × 300 = 30 open WOs. At 18 touches each, that’s 30 × 18 = 540 touches per 6-day window, 540 ÷ 6 = 90 touches per day. Better… still a lot of plates to spin.

Here’s where AI-Assisted Coordination (AIC) changes the job, not just the speed. If the coordination layer is ~85% automated end-to-end, only 15% of those touches need a human at all. 15% × 540 = 81 human touches per 6-day window; 81 ÷ 6 ≈ 13.5, call it ~14 touches per day.

That’s the flip: from ~153 touches/day (17% world) → 90 (10% world, manual) → ~14 (10% world with 85% automation).

Fourteen real actions (approvals, clarifications, true exceptions) instead of a hundred-plus little fires.

It’s not “do more with less.” It’s do more of what matters: the owner callbacks that prevent churn, the vendor conversations that improve coverage and SLAs, and the transparent updates that turn anxiety into trust.

Meet AIC: AI-Assisted Coordination

Traditional software made people faster but still kept them buried in tasks. When we first started at Vendoroo, we realized we needed to completely own the outcome if we were going to ever reach a truly agentic level of AI. 

So we’ve always had a mix of AI + expert maintenance coordinator taking the work off our clients’ plates. 

That’s why we needed a way to measure the improvement of the efficacy of the AI, and shifted from counting doors per employee to counting active WOs per AI-assisted coordinator. 

This gave rise to a new metric: AIC (AI-Assisted Coordination). 

 As time has gone on, the impact of AI has become very clear:

• How It Started: Our maintenance experts could juggle 65 to 85 open work orders at a time when we first started (eerily similar to that 65 doors per employee benchmark).

• How It’s Going: Today, our maintenance experts are managing 200 open WOs (AIC200), with a horizon of AIC 400 in clear sight.

As Vendoroo’s Chief PM, David Normand, put it, even at 200 open work orders, that person is "sitting with their hands under the desk” because the phone tag and context-hunting are gone.

What People Feel When the Math Changes

Residents notice first. Fewer of them live with unresolved issues (~41% fewer, remember?) at any moment. Frustration doesn’t have time to harden into churn. 

Vendors feel it next: approvals arrive faster, NTEs are clear, and accountability pings happen on time, which means speed-to-revenue improves. 

And your team?

They stop carrying 200 tiny worries in their heads. They make ten real decisions, then use the rest of their attention for the conversations that keep portfolios healthy.

Do The Math, Yourself

Give yourself five minutes and a notepad. This is maintenance only.

Step 1. Capture your current load

• Doors = D

• Your current percent open (point in time) = %open_now

• Touchpoints per work order (use 18 unless you have your own) = TP

• Average completion time in days (use 6 unless you have your own) = Days

Your current human touches per day
 Touches_day_now = (D × %open_now × TP) ÷ Days

Step 2. Compare to the Vendoroo benchmark
Benchmark percent open we are seeing when AI coordinates is about 10% = 0.10.
Assume automation of coordination is 85%, so the human share is 15% = 0.15.

Your projected human touches per day at the benchmark
 Touches_day_AIC = (D × 0.10 × TP × 0.15) ÷ Days

Step 3. Calculate capacity gained

• Touches saved per day
   Touches_saved_day = Touches_day_now − Touches_day_AIC
  

• Percent reduction
   %reduction = Touches_saved_day ÷ Touches_day_now

If you want to translate touches into hours, multiply by your average minutes per touch and divide by 60.
Example helper: Hours_saved_week = Touches_saved_day × Min_per_touch ÷ 60 × 5

Optional sanity check using AIC capacity

• Active work orders now
   Active_WOs_now = D × %open_now
  

• Coordinators needed at AIC200 (today)
   MC_AIC200 = (D × 0.10) ÷ 200

• Coordinators needed at AIC400 (horizon)
   MC_AIC400 = (D × 0.10) ÷ 400

Send me these calculations, please.  I’m dying to see them (and can give you my opinion)!  Remember- this does not change your total PM headcount model. It isolates maintenance coordination so you can see what AI gives back to your team each week.

Back to what matters

The thread that kicked this off asked if old headcount ratios still make sense. Helpful, but the practical question is simpler. What would it take for your people to finish the day with fewer touches and more time in the work that moves your business. That answer does not depend on asset class. It depends on removing the coordination load.

AI is a means to that end. When AI owns maintenance coordination, percent open trends toward a healthier benchmark, human touch count collapses, and the workday gets calmer. Owners get clear updates. Vendors get faster approvals. Most importantly, your people become great at their job without burning out.

Start with maintenance. Measure your percent open and your AIC for a few weeks. Convert touches saved into hours saved. Decide what to do with the time you created. Better owner communication. Tighter SLAs. Stronger renewals.

You do not need to pick a side in an old debate. Set your people up to win, and let the math show you the path.

—

Pablo Gonzalez, 

Chief Evangelist at Vendoroo