Choosing the right opportunity
Shelter’s digital product team for housing advice services has been working closely with colleagues, researching and ideating on how we might solve our biggest problem: we do not have the capacity needed to meet demand for our one-to-one housing advice services.
As a charity we had some early discussions about using AI and chatbot technology to develop personalised housing advice journeys on our website, automating a process of asking questions to gather information about users who were seeking advice to curate content specific to their needs.
However, I had serious doubts about whether this was the right opportunity for us to commit to. It seemed like a significant risk to invest in developing something that was so well-defined when there were so many unknowns. I wasn’t confident this approach of focusing on a quick solution would offer the most value to Shelter or our users.
From solution to discovery
We started to look more closely at the issues we were encountering to seek better ways to help. That was when our senior user researcher identified a common pattern: the users with the most urgent and most complex needs were not always able to get help from advisers, because people who could have been helped by finding a piece of advice themselves, or identifying a simple next step to take, quite naturally preferred to talk to someone directly.
Our national services (helpline and webchat) take a first come, first served approach – just like many face-to-face services (pre-pandemic) where people would queue at drop-in centres. This meant that people with the most urgent and complex needs, who would benefit the most from one-to-one advice, on average only made up 29%* of the queries we helped with.
It was becoming clear to us that a triage system might help prioritise those with the most urgent and complex needs, so they would have a better chance of getting one-to-one support. If we could direct other users whose needs were less urgent and easier to resolve to the advice articles they needed, then they would not be disadvantaged by this either.
The discussion about making these improvements all took place in the context of working with a funding partner who wanted to contribute to Shelter’s work, with the goal of helping the charity to support people in getting essential housing advice.
After early discussions about the possibilities of using a chatbot, our deeper analysis now led us to formulate a new proposal, which meant taking a quite different approach:
- we asked our funding partner for less money than we would have needed to start developing a chatbot, and we only applied for enough support to fund a discovery investigation related to triage.
- the outputs of this discovery phase would not result in any immediate digital development, but would ensure a deeper understanding of our user needs and technical capabilities needed to make it happen.
- the learnings gained from our discovery activities would be the key outcomes we needed to ensure this was the right direction for our product/service development.
Being bold and direct about this change to selecting opportunities created the basis for a new agreement with a new direction that we felt would achieve a great deal more for housing advice seekers. We successfully received the funding for a discovery investigation and worked in collaboration with two specialist agencies to conduct the research, delivering a roadmap of recommendations to implement.
This supported a second application for funding, which we successfully received, to support the full implementation and resource to ensure the solution is accessible and inclusive. Our next step is exploring how we might define who our priority users are in an ethical way, and prototype a concept on one of our advice channels. Our service designer will be discussing this process in a follow up blog.
It was essential that we were transparent with our funding partner about the opportunity for improvement that we selected, and our processes. The nature of digital innovation means that we must embrace a level of uncertainty, and that things will change as we start the design phase. We also had to make it clear that our key goal was about offering the greatest value to the greatest number of people.
So rather than reducing the volume of demand, we refocused our efforts on increasing the number of high-priority cases that we needed to reach. The value of this approach would also be cumulative, helping more people in ways that didn’t prevent others being helped. And we were confident based on the discovery investigation that this was the right direction, and would help achieve the best outcomes for our most vulnerable users longer term.
The impact of the coronavirus (COVID-19) pandemic made it difficult to work in the way we wanted to, so we took a fully remote approach to the discovery, meaning there are still gaps in our understanding despite the progress we made.
We did deploy a tailored housing advice user journey within our existing website, which is specific to a low complexity/low urgency need, and we didn’t need to adopt AI or a chatbot to create this. We are reviewing user feedback and plan to iterate and continuously improve before looking at how we might scale this.
Digital Project Coordinator, Jemima Stevens, will be discussing the discovery in more detail in our next blog.
*Sample of data taken over 6 months between 2019-2020