Can chatbots and AI help solve service design problems?

Full Application: Funded

There are problems for local authorities looking to adopt chatbot and AI solutions;

  • There is no shared understanding of the technology and its advantages/disadvantages as a solution to key issues. 
  • There are a variety of products and platforms that already exist on the market, but no objective evaluation of each to identify the most appropriate to use.  
  • Without a sound evidence base, there is a risk of inappropriate chatbot/AI projects being pursued that have poor prospects of meeting their objectives, wasting further time and resource

This discovery project aims to resolve these problems by delivering;

  • a methodology for evaluating the suitability or otherwise of a particular application for developing into a chatbot or AI product
  • a research base to assist local authorities developing their individual business cases to save time and resource rather than duplicating research
  • a summary of the potential technology solutions that are available with their individual advantages and disadvantages of each
  • a set of case studies drawn from participating councils

In pursuing this it will also;

  • explore the feasibility of adopting or developing a common, open source platform for local government to use to develop chatbot/AI solutions
  • deliver a set of recommendations on how to proceed with their use cases to alpha stage
  • Enable (an) alpha project(s) to take over from the discovery phase

We will use the Digital Marketplace Outcomes and Specialists framework to find a digital agency partner with suitable knowledge and experience to;

  • Work with partner councils in setting expectations and principles for the project as a whole
  • Instruct the partner councils in how to approach their user research (data to collect, methods to use, ways to capture and share findings)
  • explore the feasibility of adopting or developing a common, open source platform for local government to use to develop chatbot/AI solutions
  • Compile the user research and publishing the findings
  • Develop and publish a business case for analysing the findings
  • Producing a summary report and recommendations

There are 81 working days (from announcement to delivery date) for the project to run, so an effective work plan is essential. This will be managed by the lead authority (Oxford City Council)

We will begin preparing our Digital Marketplace requirements shortly after submission of this application to ensure the procurement can start immediately after.

The scope of the project has been scaled to address the time available (8 weeks), for instance by limiting the service user research to four areas. This will allow for up to three councils to collaborate on each area to get a wider spread of user research and identify benefits wider than a single authority.

Using a ‘publish as we progress’ approach will mean that deliverables will be created during the course of the project as well as at its conclusion. Time will be set aside in the project plan to ensure this takes place.

The project milestones are set out on our full project plan

The project will research four key service areas across all tiers of local government. A common (customer) benefit of extending will be delivered through extending service hours for each of the services

Planning advice

Currently the planning service funds a planning advice drop-in service for the public for 3 hours each day (equivalent to 105 person days per year) using a qualified planning officer. Estimated costs for this are 0.41FTE (or £15,400 per annum) which could be released. Service could be extended beyond the current delivery hours using an alternative delivery method.

Highways Defect reporting and enquiries

72,000 visitors on average per year visit Surrey County Council Highways web pages to report defects but only 21% of visitors complete reporting a defect.

Assuming that these visitors are calling the contact centre to report those defects, this equates to 4,350 avoidable calls per year. This is reflected as 0.5 FTE additional costs to the contact centre which could potentially be released.

Waste and recycling enquiries

Doncaster Council currently receives about 95,000 waste enquiry phone calls per annum which makes up a third of all phone contact despite a good online offer regularly promoted. This is equivalent to about 8 Customer Service Advisors at a cost of approximately 191k per annum during normal working hours.

Revenues and Benefits

Bromsgrove and Redditch issue approximately 50,000 recovery notices per year across the authorities, requiring a minimum of 5 FTE to be available for fielding customer enquiries (approx. £125,000 pa).

Calls received in relation to chasing claims, and provision of information in relation to Housing Benefits change in circumstances is equivalent of 1 FTE (approx. £25,000 pa).

A fifth research area for dealing with Licencing enquiries and handling their validation is anticipated to save 2000 hours or £30,000, but will only be undertaken if time and resource allows

The large number of partner councils and their distribution across the country will make collaboration on project delivery more challenging.

We will use collaborative tools as we have to develop this application; using Google Docs for creating and sharing information, and Google Hangouts for our meetings.

We will create and update a blog throughout the project with each participating council posting their progress during discovery. A lightweight project website using GitHub will be set up early on to facilitate this, with access for publishing via a CMS interface

We will adopt an ‘open first’ approach to how we work, including any mistakes, to help others learn from our experience.

The outputs from the project will be published on our project website in a format that can ensure their re-use by others for their own discovery of alpha phases.

By 29 March 2019 the project will have delivered;

Output Action
Research methodology We will use our digital agency to design an approach to information discovery. This will be captured in a series of templates that can be published and re-used by others. This will be an early deliverable of the project to guide the initial research work.
Business case We will commission our digital agency to develop a framework for evaluating the suitability or otherwise of a particular application for developing into a chatbot or AI product. 
Technology summary Will will commission our digital agency to develop an overview of the potential technology solutions that are currently available with their individual advantages and disadvantages, and their applicability to different circumstances.
Feasibility report Following on from the technology summary, our digital agency will also examine the feasibility of developing a common platform for localgov to adopt, and which technology to pursue. This will be published as a report that will be used for the final recommendations in moving to an alpha phase.
User research reports We will publish all outputs on our website on our user research from the four service areas, and will produce a final summary report in a format ready to be published. The user research teams will make use of existing templates to present user research findings in a consistent way after every round of user research.
Summary report and recommendations Our digital agency will produce a final report that either proposes how to take the individual projects forward to alpha stage or explains why each should not continue.
Case studies (use cases) The outputs from our user research and business case analysis will be developed into four case studies to assist local authorities developing their individual business cases to save time and resource rather than duplicating research.

 

User research will be used to;

  • Understand the current user experience in each of the service areas chosen
  • Understand the costs of current service delivery
  • Identify issues and opportunities for service design improvement
  • Identify why users choose the current channels that they do
  • Identify shortfalls in the user experience; frustrations that they face
  • Understand the attitudes towards using webchat/chatbots as a contact channel

For each of the four services chosen, the research will;

  • Pool existing data on the area (volumes of contact, methods of contact, existing feedback, current costs/staffing)
  • Identify the user groups to involve and recruit participants
  • Map the current user journey with users
  • Use interviews with users, including any current use of webchat services
  • Record findings in a common format

The users for each area will be;

  • Customers/residents that interact with services to obtain information or resolution
  • Front line staff that interact with customers/residents to assist them
  • Business customers or agents that act on behalf of customers/residents as a proxy

Users will be recruited to the research process through;

  • invitation of existing users that use the service, recruited at the point of contact
  • seeking out the most relevant staff in council front line services
  • for known (regular) business users or agents, inviting them directly via email

The appointed digital agency will plan, direct and (where feasible) undertake the user research. They will also complete the summary of findings.

We anticipate using the appointed digital agency to provide most of the support, direction and guidance required for the project.

No funding has been provided for this project previously

2 thoughts on “Oxford City Council

  1. As Project Manager - Cardiff Council:

    This research work looks interesting. To ensure that the outcome can have a UK-wide benefit it would be worth adding the availability of bilingual service to the assessment process. As a Welsh authority it is essential to ensure that English and Welsh are treated in such a way that neither is offered a lesser service. We have been looking at chatbot platforms and the provision of adequate automated translation, voice recognition and speech synthesis tools for the Welsh language is a hurdle to the provision of equivalent service. Please drop me a line if you’d like to discuss further

Comments are closed.