Are Multifamily Properties Understaffed?
The recent report by NAA published that for every group of 45 apartments at least one full-time employee has to be added to the payroll. Labor shortage and talent retention remain the top concerns for owners and operators. At the same time, the need for business intelligence for improved remote working is becoming mission-critical. According to the report, the need for on-site staff has increased due to bad resident reviews and increasing expectations. But the data from the ground gives us a more detailed and nuanced picture of the problem.
Domain-specific natural language models enable us to examine resident and prospect conversations at scale while drawing both automation and insight which would normally be invisible to multifamily owners. In 2020, ADAM, our proprietary AI platform for multifamily operations, had over 21,000 unique conversations with both prospects & residents across the US. Here is our aggregated insight from on-site operations.
Timing matters
Residents demand support as and when they need it. The volume of interactions with residents surges by 2.5 times on the first two days and last two days of the month. This trend is in fact predictable and repeats consistently every month. Also, over the course of the year, 36% of the interactions and 46% of leads came from outside of the typical office hours of 10AM to 6PM.
This variance creates a major staffing challenge. In order to do their jobs, the staff interacting with the residents need to be not just knowledgeable but also scalable by a factor of 2.5 . This pattern also results in task requests coming in like a deluge and creating backlogs for the rest of the month.
Variance is less acute for the leasing staff, who see a steady flow of prospects throughout the month. However, the months of June, July, and August had 5 times the number of prospect conversations that December and created 3.5 times more leads than any other period of the year.
A very long tail
The volume and importance of maintenance requests are being overestimated and all other topics are in fact underestimated. The largest cluster of interactions making up 42% of all resident interactions involved people asking questions about the property or policies of use. This cluster itself breaks down into over 700 different types of questions as specific as ‘how to use a gate’ to ‘how to use the app’. Only 14% of the resident interactions were related to property maintenance. Payment-related inquiries contributed to 10% of the volume eventually leading to a very long tail of various unique topics ranging from deliveries, complaints, waste disposal, promotions, move-ins, move-outs & beyond.
The challenge that long tails create for on-sites teams is that they constantly feel that they are wearing multiple hats and are being pulled in too many directions. At the same time, residents feel like they are not getting a consistent experience or a response to their queries.
Process gaps
The use of natural language models allows visibility to the on-site data with precision but it also exposes gaps in processes with which the site staff often improvise. In 25% of all conversations, the interaction had to be passed to someone on-site in order to figure out what needed to be done, despite the models predicting the expectation of the resident correctly. These topics included items, often repeated, which do not have a defined process or system to be inputted into. This is not a problem of interoperability of software but of missing processes that come from the historic opacity of the sites.
The onsite teams will often mention burnout and stress as the main factors for talent retention issues and turnover. In the absence of organizational learning, much knowledge is lost with staff departure. This makes it challenging to set performance metrics for teams where the SLA’s and KPIs are often arbitrary or based on a very small part of their role.
Finding micro-trends
The long tail challenge of property management is addressed well by the opportunity of finding early micro trends by leveraging machine learning techniques. The on-site teams perform better with intelligence support. However, this requires the adoption of very high quality and specialized models.
Some examples of insight gathered with micro-clusters include a relative increase in prospects wanting in-unit washers & dryers. The sharer lifestyle was also highlighted in 2020 with more prospects asking about shared kitchenettes in the community and quiet places to work or take exams.
Microtrends were also observed with more residents asking to use the management office printers & faxes. Conversations about rental promotions were had in equal measures by both residents & prospects, indicating a new pipeline for apartment transfers.
In conclusion, multifamily properties are not understaffed but they are dealing with a resource allocation problem which includes a variance in both volume & variety of tasks. It is the reason they always seem to be overstretched and under-resourced. Using cutting edge machine learning assistance can not just reduce the challenge of recruiting and training. It can also create a base for a consistent and improved customer experience.