In anticipation of our next Keystone Awards, we have examined 1000+ completions of our Property Management Software Selector Tool and enlisted the data science services of Wes Melton of Nokori to help find patterns and visualize trends. With Part 1 published last week, each Part of this series will feature one graph (to avoid overload) with audio discussion points.
Each Property Manager was asked a series of questions, each specifically identifying the importance they placed on a number of PMS specific features, such as powerful revenue management, trust accounting & reporting, work order management, etc. and so on.
While no data exercise in this range can be thought of as 100% statistically significant, no objective poll performed by a truly objective party has been performed at this scale within the Vacation Rental Industry before now.
Therefore, it is believed that the current sample size is still directionally accurate, and a helpful tool to understand overall trends in software needs that Vacation Rental Managers have demonstratively expressed as real needs over the course of their own growth journeys, both through voice and vendor selection/changes.
All participant feedback has been anonamized, consolidated, and verified. Most data analysis projects begin with a large amount of 'data cleansing' - this effort has been no different. Significant time has been placed in cleaning the data, removing obvious outliers, removing participants who did not complete the entire survey, double entries, and any other feedback that appeared to be intentionally trying to sway the results by vested parties.
In this analysis, each survery response value has been normalized to a range from 0.0-1.0. This is a value which converts a 3-part scale to a relative percentage (0%-100%) for easier evaluation. For clarity, the original survey gave respondents the 3 following choices:
- 1 - Not Important
- 2 - Somewhat Important
- 3 - Extremely Important
Additionally, Property Managers were segmented along the following unit quantities:
- 1 Property Managed
- 2-5 Properties Managed
- 6-20 Properties Managed
- 20-50 Properties Managed
- 50-100 Properties Managed
- 100+ Properties Managed
Part 2: The Co-Importance of Features
"I don't know what I don't know." This is a comment we've heard a lot when it comes to selecting a Property Management Platform. Use the graph below to connect features you know you need with closely ranked features you're not aware of. Click PLAY below and listen to Wes and Terry explain how to interpret your second graph. Then read Wes' analysis and Takeaways below.
INVITATION: If you currently use a PMS, please contribute to our research by completing the Current User PMS Feedback Survey and you will receive VIP findings before they go public. Sorry, no PMS representatives permitted.
Feature Co-Importance Matrices
How to read this matrix
This matrix primarily shows areas of 'hotness' that are visible across multiple features. For instance, we might conclude from this matrix that the same segment of users who heavily weight the importance of 'API Management' might also heavily weight 'Channel Management' or 'Property Detail Configurability'.
Property Managers should use this matrix to derive possible needs they may not realize they already have, or whether the importance demonstrated by the collaborative nature of the survey results should cause a reassessment of feature importance previously thought unimportant.
How this matrix was created
For the curious, this matrix is created by taking the average of all ratings for a given PMP feature, thus creating a "feature vector" (not in the typical meaning for those with a heavier stats background) that can then be used to create a heat map matrix such as this one using basic linear algebra. We fill in all cells by cross-averaging the two vectors to create this co-occurrence matrix of average importance.
The above feature heat map is insightful for sure, but it is an aggregate of all participants in the study. For additional visibility in to the parts that make the whole, below is the same matrix segmented by each major segment that we specified above in the introduction.
UPDATE: At the beginning of the audio, Matt suggests that along the diagonal axis we should see the strongest (hottest) correlations emerge. PLEASE NOTE that unlike traditional co-occurrence matrices this matrix is intended to give a high-level understanding of the features that coalesce into core groupings of PMS functionalities most often ranked as being the most important. And so, the diagonal values of two identical features will not always result in the highest correlation. This matrix highlights strong co-averages NOT co-occurences.