A 30-day pilot of property operations automation needs five decisions before day one: one operation, one building, a two-week baseline, a written standard, and a decision rule with a date. Week one writes the policy; weeks two to four run the operation while you watch three numbers; day 30 is expand, extend, or stop.
Most pilots of property operations automation end without a decision. They start with enthusiasm, run for a quarter, produce some anecdotes, and drift into a permanent trial that nobody wants to cancel or expand. JLL's research puts the pattern in numbers: 92% of real estate companies are piloting AI and only 5% report achieving most of their goals. The cause is rarely the technology. It is that the pilot was never designed to end.
A 30-day pilot is a design choice. Thirty days is long enough to run every weekly and monthly cycle in a property operation at least once, and short enough that the people involved stay attentive. Here is the structure, whether you are trialing a software platform, an AI agent, or a managed operations partner. Some vendors offer this shape themselves; Premise's engagement model starts with exactly this, one operation on one property with an SLA from day one, which is why the structure below matches it closely.
What must be decided before day one?
- One operation. Not "AI for the portfolio." One of: tenant and vendor communication, COI compliance, lease events, sales data collection, building records. Pick the one where the pain is most visible and the outcome most measurable. Compliance is often the easiest to measure; communication is often the most painful.
- One building. Mid-size, typical of the portfolio, with a property manager who wants the pilot to work and will tell you honestly if it does not.
- A baseline. Two weeks of logging before the pilot: hours by task, response times from the tenant's side, current compliance rate verified (not filed), open lease events. Without this the pilot cannot produce a number; the ROI calculation depends on it.
- A standard. Written, one page: what "acknowledged," "verified," and "executed" mean, the windows for each, and what counts as an exception. If a partner is involved, this is the SLA.
- A decision rule and a date. "On day 30 we will expand, extend for one more building, or stop, based on these three numbers." Written down, with names.
Rule of thumb: if you cannot state the pilot's success criteria in one sentence with a number in it, you are not ready to start. "Every certificate verified within one business day of receipt and every deficiency chased the same day, at a verified compliance rate above the baseline" is a sentence. "See if it helps" is not.
Week one: policy and setup
The first week is spent on the document that makes the rest possible: the policy the automation or the partner will run.
- For communication: response windows by request type, what can be answered without approval, tone, escalation names.
- For compliance: the requirement per lease and per vendor tier, the exception authority, the day-zero action.
- For lease events: lead times by event type, who verifies money terms, notice addresses.
- For sales data: deadlines and formats per lease, validation rules, the chase cadence.
Then the setup: access to the system of record, the inbox or channel, the contact lists, the current open items. If the vendor or partner cannot be running by day seven with your policy, the pilot has already told you something.
The standard is easier to write than it looks, because the industry has already published the shape. Provider SLA templates set tiers such as emergencies acknowledged within hours and routine work within two business days, and platforms offer SLA tracking that warns before a breach. Staff the pilot with someone other than the property manager, whose week is already full: in US residential property management the National Apartment Association reported 78% of property management companies with critical staffing shortages in 2025, and commercial offices are not better staffed.
Weeks two to four: run it, and watch three numbers
The operation runs under the standard. Your involvement is a weekly thirty-minute review and a daily glance at the exception list. Watch three numbers every week, against the baseline:
| Operation | Number 1 | Number 2 | Number 3 |
|---|---|---|---|
| Communication | Median and P90 acknowledge time by type | Resolution time | Items escalated to the property manager, and whether each should have been |
| Compliance | Verified compliance rate | Days from deficiency to chase, and to compliant | Vendors dispatched while non-compliant (should be zero) |
| Lease events | Events due, executed on time | Notices with money terms verified before sending | Tenant acknowledgments logged |
| Sales data | Reports received by deadline | Reports validated within two days | Exception list length at month end |
In the weekly review, ask three questions: what did the operation do that we would not have done? What did it do wrong? What did it hand to the property manager that it should have handled? The third answer is the one that decides whether the pilot removed the queue or moved it.
What usually goes wrong?
- The exception list is long. Either the policy is too tight (fix the policy) or the operation cannot handle judgment (fix the vendor).
- The property manager is still doing the chase. The vendor is a tool, not an operation. Decide whether that is what you wanted.
- Tenants or vendors are confused by the new sender. Normal in week two, a problem in week four. Check the messages.
- The numbers look great and the property manager is unhappy. Read the escalations; something is being counted as done that is not.
- Nothing measurable changed. The operation was not the bottleneck. Useful to know; pick another.
Day 30: the decision
Sit down with the baseline, the three numbers, the exception review, and the property manager. Apply the decision rule written before day one. Three outcomes:
- Expand to more buildings or a second operation, with the same structure.
- Extend for one more building or one more month, with a specific thing to fix and a new date.
- Stop, with the reasons written down, because a clear no is a good result.
Then do the ROI arithmetic with real numbers, and put the result in front of whoever will fund the expansion. The point of the pilot was never the thirty days. It was the decision, and the number it produced.
Frequently asked questions
How long should a property operations automation pilot run?
Thirty days: long enough to run every weekly and monthly cycle in a property operation at least once, short enough that the people involved stay attentive. Pilots that run a quarter without a decision date drift into permanent trials that nobody cancels or expands.
What should I measure during a 30-day pilot?
Three numbers per operation, weekly, against the baseline. Communication: acknowledge and resolution times by type, and items escalated to the property manager. Compliance: verified compliance rate, days from deficiency to compliant, vendors dispatched while non-compliant. Lease events: due versus executed on time, money terms verified, acknowledgments logged.
What has to be ready before a pilot starts?
A two-week baseline log (hours by task, response times from the tenant's side, verified compliance rate, open lease events), a one-page written standard defining acknowledged, verified, and executed with windows, and a decision rule with names and a date. If the success criteria cannot be stated in one sentence with a number, you are not ready.
What are the signs a pilot is going wrong?
A long exception list (the policy is too tight or the vendor cannot handle judgment), the property manager still doing the chase (you bought a tool, not an operation), tenants confused by the new sender in week four, numbers that look great while the property manager is unhappy, or nothing measurable changing (the operation was not the bottleneck).
What happens on day 30?
Apply the decision rule written before day one: expand to more buildings or a second operation, extend for one more building or month with a specific fix and a new date, or stop with the reasons written down. Then do the ROI arithmetic with real numbers and put it in front of whoever funds the expansion.