How AI saved £400k/year for Citizens Advice Scotland
Written by Andy Bell, CEO, helpFirst.ai
Citizens Advice Scotland (CAS) has an unusual operating model. It’s an umbrella organisation of 59 local bureaux. Each bureau is its own independent charity, organised to best suit the needs of their local community. That local knowledge is key.
So far so good.
But this structure makes it hard to offer a single point of access. During the pandemic, CAS established the Scottish Citizens Advice Helpline: a central number for advice. To get local advice, clients would have to wait for a callback from their local bureau.
There were a few problems: it could take 24 hours to get a callback; volunteering to staff the service was not popular; advisors wanted to be giving advice, not routing calls; and the system cost nearly £500k per year to run.
Working with Civtech, my team successfully deployed an AI-driven system (thanks to the support from CivTech), which tackled these challenges that: reduced wait times for connecting to a local bureau to under a minute; eliminated the need for staffing, allowing advisors to be redeployed from the contact centre to the frontline; and reduced annual costs below £30k.
CAS said that the project was a significant game changer in bridging the gap between national coverage and local delivery. By smart use of AI, we effectively decentralised a call centre model. The project won a 2023 ScotlandIS Tech for Good Award and was runner-up for DigiLeaders AI Innovation of the Year.
Here’s 3 things we learnt from this project:
Service Design is key to AI
Advances in AI are bringing human-like writing and understanding within reach. Similar to previous tech advances (whether the internet, the computer or even electricity before that), the key issue becomes how to use the technology well. This is a service design question: the challenge is deeply understanding user needs, exploring innovative solutions, and continuously refining the user experience.
Cost-Effectiveness Doesn’t Compromise Quality
One significant takeaway is that AI can not only make operations more cost-effective but also improve the quality of service. The transition from a £500,000 per year system to one costing only tens of thousands without compromising — in fact, improving — response time is a wonderful example of this.
Buy Over Build
Instead of creating a solution from the ground up, leveraging existing, well-funded technologies (like PolyAI, in this project) can speed up implementation and reduce risks. This ‘buy over build’ strategy is a crucial lesson if you are seeking to innovate swiftly.