Hometime.io Research Repository

Introducing research operations to the business to store feedback from leaving customers, to give greater visibility on churn and inform retention strategies into the future

Research ops

A system for storing research insights


Losing customers without knowing why

One of my first projects at hometime, was to work on something that would improve retention. Project lead Rosa, explained that customers were churning at a high rate, and we needed to do something urgently to remedy that.

Key Issues

  • There was no standard way for employees to react to a leaving customer
  • Some account managers would ask for a reason over the phone and not write it down. Allowing the vital information to disappear into the ether. 
  • We had no simple way of tracking who had left, even if we did store the information. We would only see the total number of active properties fluctuate.

Introducing research operations 

The inital question of "how do we decrease churn?" Triggered a question in my mind that I saw as needing to be answered first, which would in turn make it a lot easier to know how to approach retention.

Why are customers churning?


At that point in time, we had no idea. Since there was no process or system in place to gather that information in a standard manner, and then store it to explore later.

Luckily, at my previous role as a designer at Channel 7, coworker and user researcher Monica Regalado had introduced me to a former Google senior researcher, called Tomer Sharon, who had implemented a system at WeWork to do that just. They called it Polaris. 

A tool Tomer Sharon implemented at WeWork to store all the evidence and insights gathered from interviews with customers.

Polaris was a tool to store gathered information, where insights were connected to primary evidence like sound and video recordings from interviews. Many key elements would be connected.

  • What elements were involved in the customer's experience?
  • Which areas?
  • Which people?
  • What emotions did they feel? And so on.

Tomer's idea was that searching through Polaris for something like:

"How many customers had a negative experience with our beer facilities in New York?" 

Would provide a certain amount of results, and that collection, which he called "a playlist of research nuggets", would replace the old-fashioned research report. Which was time consuming to create, existing with no functional connection to any system, typically in a PDF, and would typically get forgotten about.

I'd spent weeks examining the system with Monica at Channel 7, and I decided to implement it at hometime to help us deal with churn.

The plan for an MVP research system

Operations analyst, Mariam had already created a form for employees to go through each time a customer left, to deal with internal actions that needed to be taken. That form was important and needed to stay, but I wanted to build around it with an outward facing form, for customers to fill out when they left.

Until we had the capacity to attempt to interview leaving customers in our offices across the country. I was hopeful that the MVP of this new system, would show its value and result in company buy in, to allow me to take it to the next step and aim for face to face interviews to gather information.

  • Leverage existing "Offboarding checklist" for employees
  • Use Hubspot to allow account managers to trigger an email to a leaving customer
  • Create a new "Offboarding form" for leaving customers to be sent in the Hubspot email
  • Create a new database inspired by WeWork's Polaris to store the information gathered
  • Use Zapier to automate syncing between the various apps
offboarding process2
We didn't expect our MVP process to result in 100% of leaving clients filling out our form. But since we were currently collecting zero data, we saw value in having access to even a small percentage of the reasons our customers were churning. With that, we could already start planning retention strategies.

The Offboarding form's logic

Offboarding form structure v3
I worked on several drafts of fields and logic for the form, with the aim of covering as many possible realistic scenarios as possible. My team mates Rosa and Mariam were crucial in helping me to fill in the gaps.

Some key aspects of the form were:

  • Making a distinction between people that were leaving due to a negative experience, and those leaving due to being in a new life situation
  • Distinguishing between customers leaving us to use a different Airbnb property management company and those leaving the Airbnb platform entirely
  • Allowing them to tell us in detail about the negative experience that convinced them to leave. Making it easier to prevent that occuring again.
Eventually we decided to build the form with all its conditional logic, in Typeform, and although it was very involved, it wasn't too much of a headache. It left me very impressed with Typeform's capabilities. 
This image is how Typeform depicts the form and its logic.

The final version of the MVP offboarding form


Building our own Polaris

A significant part of the power of Tomer Sharon's Polaris, is that it allows you associate so many details of a customer's experience that you discovered through research, with the stored evidence.
All combined, those parts represent the research "nugget." I went through his Polaris tool to see what his tag structure was like, to see what I could use and adapt for our research repository.
There was a lot...



The aftermath

  • Eventually we had set up the various parts and synced everything through Zapier with the help of Zapier and Airtable expert Zoe
  • But at that point we were told that the business had shifted priorities from retention to growth, right when we were ready to have the system be used for the 1st time with real customers

Regardless, it was great to plan and build an end to end research process and system. I look forward to the next time I get a chance to be involved with this type of project, eventually using it to gain insights and utilising them to improve a customer's experience.

Selected Work