Local Digital Project: Chatbots and AI

Sprint #1: Planning services

This week we completed Sprint #1 of the project, looking at planning services across three councils. As well as user research interviews with 18 members of the public there were stakeholder interviews with 6 staff in councils, and a stack of data gathering to get completed.

Show and Tell

Friday saw us bring together initial findings from the week and a chance for all participating councils to get early headlines.

We had 10 people on the Hangout, including Andrew from the Planning team from Oxford, who was able to jump in and answer a question arising from the research.

Although we’d captured the event on video, our sound recording didn’t work out, so instead the slides are presented below, and we go into a bit more detail about the work after that.

Download the slide show

User research interviews


Luiza from Torchbox fed back on the interviews that she and Kyle had conducted in Oxford.

From the interviews they had identified two key user journeys; users managing their own application and users who are checking or challenging someone else’s application. Making an application was far less emotional than objecting to one, as users already felt invested in it.

The complexity and diversity of planning queries lends itself to calling to speak to someone rather than using online services;

“It’s a very complex, nuanced and subjective topic of conversation, and not having an answer that fits all is an issue when it comes to giving specialised customer service”

Users were using multiple sources to research their applications before submitting, and it appeared that more experienced users (i.e. agents and architects) were able to distinguish the cheaper route to progress these. Andrew from Oxford was able to clarify that the Planning Portal had introduced a handling charge for online applications

Users were sometimes given contradicting advice; a suggestion to email in a complex query was then met with the suggestion to use the drop-in service as the query was too complex!

The two user journey maps were walked through to show the key issues identified and user emotions at different points.


Claire Cook fed back on her work at Cheltenham.

She found that there was a period of time where users do a lot of their own research; using the website quite a bit, looking at other applications to get a sense of what would be acceptable, and even using it to find an architect that had been successful in having applications approved.

Claire had spent time analysing the types of customers that use planning services, which she’d used in her customer journey maps. Luiza is looking to echo this approach with some work Torchbox are doing on the results.


John Robinson fed back on Hertsmere’s three days of interviews.

Most of the people that responded to the user recruitment did so because they had an issue personal to them. It had to be explained to them that there was a fixed script and that their personal questions couldn’t be addressed.

They found that people really wanted to speak to a planning officer rather than contact centre staff; it was hard for them to get through to the back office. With the website not providing answers to the questions they had it was more likely to generate calls;

“I think here its frustrations of ‘I want to speak to a planning officer - I don’t want to speak to anyone else’ “

However, a key issue they shared with the other councils is the number of available planning officers to take those direct calls;

“Our underlying problem is recruiting and retaining planning officers - we just haven’t got enough”

John’s interviews echoed the findings from Oxford; for people responding to someone else’s application there was more on an emotional attachment to the process;

“With objections it’s very emotive and they’re not necessarily going to get the answer they want anyway”

Stakeholder interviews

Tom Williams from Torchbox had been speaking to key stakeholders in each council, as well as using data gathered to date on contact centre activity and website analytics, to get an early view on opportunities to introduce automation in Planning services.

Early estimates of the call volumes alongside the proportion of calls that can be resolved by front-line staff suggest that across three councils there would only be a potential £86k saving, which throws the business case into doubt based on return on investment.

A key issue identified is the reliance on paper-based information to reach decisions, which makes the application of artificial intelligence much more difficult.

Areas that would appear to be beneficial are;

Data gathering

As Torchbox were doing the heavy lifting with user interviews in Oxford, Neil Lawrence had spent time digging further into available data to learn more. This included;

This has been summarised on the Oxford Data Dive page.

Reactions by colleagues to the data had been interesting; it had been anticipated that a far higher proportion of calls would direct people to the website. Also, the volume of downloads were a surprise;

“A colleague said ‘there’s such a large volume of downloads going on but they’re still calling us’ so it’s not hard to move from that to ’is the stuff we’re giving people on the web in any way meeting user needs”

Learning points

Luiza, Kyle and Neil worked on a retrospective after the user research interviews in Oxford to identify things for others to take away from the process

Things that made it good:

Things that made it bad:


Next Steps

Next week is Sprint #2 looking at Waste and Recycling Services in Doncaster, Rotherham, Bolsover and North East Derbyshire.

The intrepid Claire Cook will also be undertaking research for Cheltenham for sprint #2 as well!