Pre-launch cancellations were at 18%, which was a significant problem for the overall health of the product. We researched and identified opportunities to improve the new client onboarding process, which when tested, saw a nearly 90% decrease in pre-launch cancellations.
This project lasted about 3 months in total, from research through test implementation. I worked closely with our Director of Product, CS stakeholders, Program Management, and our UX intern throughout this process.
We didn't really know. But here's what we did know.
In the fall of 2017, pre-launch cancellations were at 18%. This meant that almost 1 out of 5 clients were cancelling during the trial period, before they even got to use the product. Solving this was important to the business, and would help us address our two objectives for the year: improving client retention and the end-to-end customer experience. We initially hypothesized that this was due to the average 4 weeks it took to launch a new client's account (over 60% of clients weren't launching within 30 days). We had to dig deeper and see if we could back up our assumptions, so our next step was to interview customers and the internal teams that were involved in the launch process.
I put together a research plan detailing our objectives, research questions, hypotheses, logistics, and research participants. We started by interviewing three groups of current and former clients:
Current clients who were launched successfully.
Current clients who called in to cancel during the launch process but were saved by our Retention team.
Former clients who cancelled before their website was launched.
After our interviews, we organized our findings and were able to group those client experiences under three themes:
“I thought the site would go up in 2-3 days, but it took 3 months.”
“They promised they would take care of everything, but later I found out I would have to buy my own domain and email, go to my realtor association and get my own IDX.”
“My concern was wasting too much time to begin, and at the same time they already have my money.”
“The [process] that your backend team had to navigate was a surprise to them...I’m the one left holding the bag. I paid all this money upfront...it was deficient.”
“After the site was live, I waited close to 10 days for the first training call...”
“We only had 30 minutes to train a team of 10 agents, I wish we had an hour.”
“I don’t have a lot of time to go in the back-end and learn that [after my site is already live].”
The pain points were becoming a little more clear, and these interviews helped us define the problems we wanted to tackle:
How might we set better expectations so clients have a clear understanding of what's going to happen during the launch process?
How might we start providing value during the launch process?
How might we start training clients before their website is launched?
Furthermore, it became painfully obvious that we didn't really have an effective onboarding process. We interviewed team members who were involved in the launch process, and mapped out what that looked like for both clients and TORCHx employees:
This map helped illustrate several problems with our launch process:
Clients were getting kicked around from phone call to phone call with different TORCHx employees, from sales to fulfillment to customer support, so they didn't have a go-to representative they could call or email with questions or concerns while their website was getting built.
We were doing little to communicate with clients during the build process, other than asking them for information (e.g. "What domain do you want to use for your website?"), and it was during this period that many clients started to doubt their investment in our product.
The training schedule lasted for almost a month after their website was launched, which meant clients weren't able to really get any value from the platform for almost 2 months after buying our product.
There was little we could do to shorten the time-to-launch, since most of these delays were due to MLSs (multiple listing services) being unresponsive, or taking weeks to returned signed paperwork.
MLSs are databases of regional "listings" (e.g. homes for sale). Agents use these databases to populate their websites with real estate listings. In most cases, accessing this information is free for the public (people like us!) because an agent or brokerage is paying the MLS fees to use that data. MLSs are a great tool for both the agent and homebuyer. The issue lies in their number - there are over 600 MLSs in the United States, each with their own processes and regulations. So for a product like TORCHx to scale, it needs to find efficient ways to work with all the different MLSs. But this is easier said than done.
Armed with this information, as well as research on successful onboarding processes, we ran a brainstorming session with team members from Customer Support, Sales, Configuration and Fulfillment. Here's the solutions we aligned on, along with the new timeline we wanted to test:
Update the sales script to make sure sales representatives weren't over-promising just to get their commission.
Start training clients as soon as they buy TORCHx so they can get value early in the process, and communicate with a single CS Representative during the whole process.
Provide a new training seminar during the onboarding process from a subject-matter expert on nurturing real estate leads.
Create an "Onboarding Status" page so clients have a go-to reference for where they are in the process and what information we need from them
The new onboarding experience was tested in October 2017 with a cohort of 50 new clients that were sold during that month, along with a control group that was given the "old" onboarding experience. We didn't set a success metric, since we didn't have data to inform us of what that could be. When the test concluded, we were pleasantly surprised by the results:
We expected a drop in the churn rate but not such a massive one! Additionally, a few other metrics we were tracking were positively affected, including training call completion rates and even a tiny improvement in launch time. We also interviewed CS Representatives and they felt clients were more engaged during the new onboarding process.
Thanks to the success of this test, the new onboarding process was launched to all new clients in December of 2017. We've been monitoring churn rate over time, and while it was never as low as the 2% rate we saw during the test, the rate has now hovered around 8%-10% which is a drastic improvement to our bottom line and the clients' overall experience with our product.
After the test concluded, we called several of the clients that had launched through the new onboarding process. Overall, they didn't have complaints (beyond the time-to-launch) and they felt like they were ready to use the product as soon as their site was launched. One aspect of the new process that didn't seem as successful was the "Onboarding Status" page. Clients we spoke to couldn't recall seeing or using this page, even though it was used as a reference during their training and follow-up emails included a link to the page.
This page was cobbled together by yours truly since we didn't have engineers to support us during this project. Using rudimentary HTML and CSS, I built a page that we could easily scale to the 50 test clients, and CS Representatives could also build the page using a tool within the CRM ("Custom Pages"). It's easy to see why this page (left) was likely ineffective, with problems ranging from poor hierarchy to cognitive overload.
Fortunately, we've recently redesigned it so clients can better understand where they are in the onboarding process, and what information we need to collect from them (right).
Personally, I had two key takeaways from this project:
We should have aimed to test a "thinner" slice of the experience. It became difficult to tell what had the greatest impact on the churn rate since we ended up changing so many things about our onboarding process. If we had started smaller with our test, we could have more easily pinpointed what levers would have the greatest effect, so we could build on those in the future as we continue to improve the onboarding process.
Service Design! This was my first real foray into Service Design and I really enjoyed it. If anything, it requires even greater collaboration and trust-building than our usual product work, and it was great to see how we could have a positive impact on our customers despite being limited by our team's engineering resources.