[day note] DVLA, Swansea


Thanks to Dudley, Lee and Ryan for taking time out of their days to chat to me and to Rohan for making it happen. It was a genuinely interesting couple of hours (and not only for finding a Grimsby fan in Swansea!).

The DVLA digital team have been a major player in the GDS ‘revolution’ over the last four years and there is a noticeable maturity in their approach. The nature of the organisation really reinforces a user/customer centric approach which I suspect has helped with the extent they have embraced things.

In no particular order I have just noted down a few things I learned below but my 3 main take-aways are:

(1) The power of having Service Managers have responsibility for the whole ‘service’ not just the digital aspects

(2) The lessons from the move to lots of in-sourcing to a blended model and finally in-house sustainable teams

(3) The amount of foundational ‘business’ work the teams do before (and during the build)

Oh and the place is blooming huge but the views are pretty great on the 14th Floor (at least on a lovely day like today!)


So some quick notes on the train:

  • DVLA Service Managers genuinely owning the services end-to-end (including non-digital channels where they exist)
  • Have a sizeable co-located Service ‘group’ split in to genuinely multi-disciplinary teams that include not only technical staff & product owners but also subject matter experts and project support staff.
  • Four years on from initial involvement with GDS on the exemplars they now have a wide portfolio of ‘transformed’ digital’ services.
  • They have used the ‘in-source’ model over the years in particular building a relationship with Kainos
  • Currently use Skyscape as their primary cloud provider but have ambitions/plans to move to AWS
  • They operate with a pretty Scrum flavoured agile but it is clear they have taken an iterative approach to working agile with an impressive mix of styles of ‘big boards’ visible all over the office.
  • They essentially have ‘User Research as a Service’ — a user insight team with lab facilities elsewhere in the city assign staff to the project and they undertake research nationwide.
  • Real commitment from the Product Owners to be at the research sessions and they actively encourage both technical and business members of the team to observe research sessions whenever possible.
  • Have involved interaction designers in the user research sessions to enable on the fly updates to prototypes (often using Heroku) based on the feedback to play that back to users asap
  • They do a lot more integrated ‘business analysis’ type work up front (all displayed on a Kanban board) than I am used to seeing but it makes sense as it provides the foundation for the development work.
  • Two week sprints with daily stand-ups (chicken/pig rules) and at times they have had ‘scrum of scrums’ as projects scaled up. Strict about stand-ups being 15 minutes.
  • ’Showcases’ at end of every sprint — have an area in the main room for this and it is open to everyone. The challenge is sometimes that the shows tell a story the evolves over the sessions and some attendees can derail by needing additional context setting as they have missed earlier sessions.
  • Retrospectives at the end of each sprint — careful to ensure that they come away with a handful of actionable changes after each retro and report back on how things went at the next one.
  • Continuous service improvement (CI) team is also embedded alongside the new product teams
  • deploying changes to a portfolio of web and API services as often as twice a week for more than a year
  • The team maintains commitment to user focus and agile principles — dealing with huge amounts of user feedback!
  • Larger features sometimes considered out of scope for CI and instead becomes a project itself requiring a team spun up around it
  • The ‘Driving with a medical condition’ service is the first team to be almost 100% DVLA staff without ‘in-sourced’ support from someone like Kainos
  • Have to increase in-house staff capability with gaps around Ops/SRE and front-end dev in particularly and they are building automated testing capacity.

Thanks again to everyone I met today.