The Beta Assessment is probably the one I get the most questions about; Primarily, “when do we actually go for our Beta Assessment and what does it involve?”
Firstly what is an Assessment? Why do we assess products and services?
If you’ve never been to a Digital Service Standard Assessment it can be daunting; so I thought it might be useful to pull together some notes from a group of assessors, to show what we are looking for when we assess a service.
Claire Harrison (Chief Architect at Homes England and leading Tech Assessor) and Gavin Elliot (Head of Design at DWP and a leading Design Assessor, you can find his blog here) helped me pull together some thoughts about what a good assessment looks like, and what we are specifically looking for when it comes to a Beta Assessment.
I always describe a good assessment as the team telling the assessment panel a story. So, what we want to hear is:
- What was the problem you were trying to solve?
- Who are you solving this problem for? (who are your users?)
- Why do you think this is a problem that needs solving? (What research have you done? Tell us about the users journey)
- How did you decide to solve it and what options did you consider? (What analysis have you done?)
- How did you prove the option you chose was the right one? (How did you test this?)
One of the great things about the Service Manual is that it explains what each delivery phase should look like, and what the assessment team are considering at each assessment.
So what are we looking for at a Beta Assessment?
By the time it comes to your Beta Assessment, you should have been running your service for a little while now with a restricted number of users in a Private Beta. You should have real data you’ve gathered from real users who were invited to use your service, and your service should have iterated several times by now given all the things you have learnt.
Before you are ready to move into Public Beta and roll your service out Nationally there are several things we want to check during an assessment.
You need to prove you have considered the whole service for your users and have provided a joined up experience across all channels.
- We don’t want to just hear about the ‘digital’ experience; we want to understand how you have/will provide a consistent and joined up experience across all channels.
- Are there any paper or telephony elements to your service? How have you ensured that those users have received a consistent experience?
- What changes have you made to the back end processes and how has this changed the user experience for any staff using the service?
- Were there any policy or legislative constraints you had to deal with to ensure a joined up experience?
- Has the scope of your MVP changed at all so far in Beta given the feedback you have received from users?
- Are there any changes you plan to implement in Public Beta?
As a Lead Assessor this is where I always find that teams who have suffered with empowerment or organisational silos may struggle.
If the team are only empowered to look at the Digital service, and have struggled to make any changes to the paper/ telephony or face to face channels due to siloed working in their Department between Digital and Ops (as an example) the Digital product will offer a very different experience to the rest of the service.
As part of that discussion we will also want to understand how you have supported users who need help getting online; and what assisted digital support you are providing.
At previous assessments you should have had a plan for the support you intended to provide, you should now be able to talk though how you are putting that into action. This could be telephony support or a web chat function; but we want to ensure the service being offered is/will be consistent to the wider service experience, and meeting your users needs. We also want to understand how it’s being funded and how you plan to publish your accessibility info on your service.
We also expect by this point that you have run an accessibility audit and have carried out regular accessibility testing. It’s worth noting, if you don’t have anyone in house who is trained in running Accessibility audits (We’re lucky in Difrent as we have a DAC assessor in house), then you will need to have factored in the time it takes to get an audit booked in and run well before you think about your Beta Assessment).
Similarly, by the time you go for your Beta Assessment we would generally expect a Welsh language version of your service available; again, this needs to be planned well in advance as this can take time to get, and is not (or shouldn’t be) a last minute job! Something in my experience a lot of teams forget to prioritise and plan for.
And finally assuming you are planning to put your service on GOV.UK, you’ll need to have agreed the following things with the GOV.UK team at GDS before going into public beta:
- how the transaction will join up with GOV.UK guidance or any start page
- a subdomain name and any content changes needed
Again, while it shouldn’t take long to get these things sorted with the GOV.UK team, they can sometimes have backlogs and as such it’s worth making sure you’ve planned in enough time to get this sorted.
The other things we will want to hear about are how you’ve ensured your service is scalable and secure. How have you dealt with any technical constraints?
The architecture and technology – Claire
From an architecture perspective, at the Beta phases I’m still interested in the design of the service but I also have a focus on it’s implementation, and the provisions in place to support sustainability of the service. My mantra is ‘end-to-end, top-to-bottom service architecture’!
An obvious consideration in both the design and deployment of a service is that of security – how the solution conforms to industry, Government and legal standards, and how security is baked into a good technical design. With data, I want to understand the characteristics and lifecycle of data, are data identifiable, how is it collected, where is it stored, hosted, who has access to it, is it encrypted, if so when, where and how? I find it encouraging that in recent years there has been a shift in thinking not only about how to prevent security breaches but also how to recover from them.
Security is sometimes cited as a reason not to code in the open but in actual fact this is hardly ever the case. As services are assessed on this there needs to be a very good reason why code can’t be open. After all a key principle of GDS is reuse – in both directions – for example making use of common government platforms, and also publishing code for it to be used by others.
Government services such as Pay and Notify can help with some of a Technologist’s decisions and should be used as the default, as should open standards and open source technologies. When this isn’t the case I’m really interested in the selection and evaluation of the tools, systems, products and technologies that form part of the service design. This might include integration and interoperability, constraints in the technology space, vendor lock-in, route to procurement, total cost of ownership, alignment with internal and external skills etc etc.
Some useful advice would be to think about the technology choices as a collective – rather than piecemeal, as and when a particular tool or technology is needed. Yesterday I gave a peer review of a solution under development where one tool had been deployed but in isolation, and not as part of an evaluation of the full technology stack. This meant that there were integration problems as new technologies were added to the stack.
The way that a service evolves is really important too along with the measures in place to support its growth. Cloud based solutions help take care of some of the more traditional scalability and capacity issues and I’m interested in understanding the designs around these, as well as any other mitigations in place to help assure availability of a service. As part of the Beta assessment, the team will need to show the plan to deal with the event of the service being taken temporarily offline – detail such as strategies for dealing with incidents that impact availability, and the strategy to recover from downtime and how these have been tested.
Although a GDS Beta assessment focuses on a specific service, it goes without saying that a good Technologist will be mindful of how the service they’ve architected impacts the enterprise architecture and vice-versa. For example if a new service built with microservices and also introduces an increased volume and velocity of data, does the network need to be strengthened to meet the increase in communications traversing the network?
Legacy technology (as well as legacy ‘Commercials’ and ways of working) is always on my mind. Obviously during an assessment a team can show how they address legacy in the scope of that particular service, be it some form of integration with legacy or applying the strangler pattern, but organisations really need to put the effort into dealing with legacy as much as they focus on new digital services. Furthermore they need to think about how to avoid creating ‘legacy systems of the future’ by ensuring sustainability of their service – be it from a technical, financial and resource perspective. I appreciate this isn’t always easy! However I do believe that GDS should and will put much more scrutiny on organisations’ plans to address legacy issues.
One final point from me is that teams should embrace an assessment. Clearly the focus is on passing an assessment but regardless of the outcome there’s lots of value in gaining that feedback. It’s far better to get constructive feedback during the assessment stages rather than having to deal with disappointed stakeholders further down the line, and probably having to spend more time and money to strengthen or redesign the technical architecture.
How do you decide when to go for your Beta Assessment?
Many services (for both good and bad reasons) have struggled with the MVP concept; and as such the journey to get their MVP rolled out nationally has taken a long time, and contained more features and functionality then teams might have initially imagined.
This can make it very hard to decide when you should go for an Assessment to move from Private to Public Beta. If your service is going to be rolled out to millions of people; or has a large number of user groups with very different needs; it can be hard to decide what functionality is needed in Private Beta vs. Public Beta or what can be saved until Live and rolled out as additional functionality.
The other things to consider is, what does your rollout plan actually look like? Are you able to go national with the service once you’ve tested with a few hundred people from each user group? Or, as is more common with large services like NHS Jobs, where you are replacing an older service, does the service need to be rolled out in a very set way? If so, you might need to keep inviting users in until full rollout is almost complete; making it hard to judge when the right time for your Beta assessment is.
There is no right or wrong answer here, the main thing to consider is that you will need to understand all of the above before you can roll your service out nationally, and be able to tell that story to the panel successfully.
This is because theoretically most of the heavy lifting is done in Private Beta, and once you have rolled out your service into Public Beta, the main things left to test are whether your service scaled and worked as you anticipated. Admittedly this (combined with a confusion about the scope of an MVP) is why most Services never actually bother with their Live Assessment. For most Services, once you’re in Public Beta the hard work has been done; there’s nothing more to do, so why bother with a Live Assessment? But that’s an entirely different blog!