August 16, 2017

How we became a world class engineering organisation (and saved a million pounds a year in the process)

Peet is Chief Technology Officer at Wealth Wizards. He heads up the AI Guild and is founder of the Birmingham AI Meetup.

Smart Platform blog
As a nation, we’re bad at looking after our money. UK personal debt is around £1.5tn. This means that on average, everyone has more debt than they earn in a year. Our savings, as a proportion of salary, are around half of the global average.

The average Briton has set aside just £87,000 in their pension pot which needs to last the rest of their life. Helping people to prepare for retirement is a particularly tricky problem to solve because the landscape is so complex. Regulated financial advice is typically out of the reach of most people.

Wealth Wizards’ mission is to make financial advice affordable and accessible to everyone. We do this by bringing together a combination of financial intelligence and smart software technology.

Wealth Wizards has grown dramatically over the past few years from a young start-up of ten people in shared office space in 2015 to a medium-sized scale-up of over sixty people in early 2017. The ways of working that are effective for a start-up will not usually scale well, and like most successful start-ups, we found ourselves needing to transform.

For example, our applications were becoming harder to change over time as they’d been written without sufficient test coverage. Our infrastructure had been hand-built, and only one person in the organisation understood how to operate it. We were releasing new software every two months or so. In addition, such aggressive growth had resulted in a 75/25% contract to permanent ratio. In short, the magic that had led us to our first major successes of winning large clients and securing investment would not necessarily help us to hit our new, loftier goal of changing a million lives by 2020.  We decided to run a Lean Enterprise transformation focusing on the four areas of People, Process, Technology, and Culture.

People

Drawing on the pool of local contractors had helped us to scale very quickly and get our products live and into the hands of customers and investors. However, the transformation ahead would require a lot of investment in people; there would be a lot of new technologies and processes to learn.  We knew that to make this work, we needed a strong foundation of permanent, internal talent.

We started a recruitment drive looking for the very best local engineers. Joel Spolsky, founder of Stack Exchange, and creator of Trello states that the best developers are five to ten times more productive than average, and we’ve certainly found that to be true.

So, what does ‘good’ mean in this context? We focus on three elements: “Smart, gets things done, plays well with others”. I personally also look for people with a growth mindset, passion for changing the world and a deep and enthusiastic love of tech.

The team we ended up with was smaller, but much more performant - able to deliver several features per two-week sprint instead of per two-month release. There was also a much closer attention to long-term quality with us moving from low doubledigit test coverage to being consistently above 80%.

Since moving to mostly perm, the team feels very invested and involved in all aspects of product discovery and delivery. And there’s a huge amount of buzz and excitement around hitting our organisational objectives. 

Process

The company has been agile from the beginning. However, an external assessment we commissioned showed that we were suffering from infrequent releases, low predictability and morale, things that would be addressed by doing an ‘Agile reset’; going back to basics and sticking to the core Scrum process before migrating to a more mature model.

We elevated the main Scrum ceremonies, making them more frequent and made attendance mandatory for everyone in the team. We started being a lot more public about our commitments, tracking delivery-to-commitment and reporting on it every fortnight. As a result, confidence grew in our ability to deliver on promises, and it became obvious how important prioritisation was as there was always too much to build. 

We recruited an Agile coach who helped build our Product Management discipline to conduct discovery, shaping and prioritisation of new features. This SVPG (Silicon Valley Product Group) inspired product team spend time on the front line with customers, forming hypotheses, designing and running experiments and gathering data about what kind of features our customers most want to see, and how they want to interact with them. 

To support these fast turn around interactions, we needed to streamline our delivery organisation to be able to take ideas and push them into live quickly, and to be able to measure everything – how many people used the new feature, what demographic were they in, how many completed the journey, where people dropped out of the process, who clicked where, which messages did people find confusing, which parts of the journey were unintuitive. Some of this data we can collect automatically from our analytics software. For others, we run sessions with customers, or have access to a UX lab where we can observe users’ interactions more closely

Technology

Our infrastructure estate had been hand-built by a single engineer, who was the only person in the organisation who really understood how it all fit together. In addition, she was the only person who could make changes, or push releases to live. As well as being a key-man dependency the workload was also growing to be unmanageable for a single person.

So, we invested a lot in our operations and have tried to follow an ‘everything as code’ philosophy. This has allowed us to build a platform with a huge amount of automation. All of our infrastructure is built using scripts stored in version control. This not only means that we’re able to deploy highly tested code to our production environment with minimal human input, but also means that our estate itself is easy to change. When the Dirty Cow vulnerability was discovered, for example, we were able to ensure that our whole estate was patched in under an hour.

The use of Microservices has meant that our products are built out of composable modules available via an API. This approach itself has helped us to be more reactive to new ideas or opportunities. For example, we recently reimagined one of our end user facing products as a productivity tool for IFAs. We were able to quickly plug together our existing microservices into a new configuration to create a new tool for a new user base. This ‘Robo-paraplanner’ is now saving human advisers around five hours per case handled.

Culture

Our culture is centred around focusing the whole organisation on a small set of business goals and empowering small teams to find creative ways of driving the company toward those goals. Our engineering teams are arranged in a Spotify-style squads who are given one or two business metrics to own. Each squad is made up of three developers, a QA, an ops engineer and a business product owner. Everything the product owner needs to take an idea from inception to running in live is within her team. There’s no more waiting for other teams to deliver dependencies, and we no longer spend any time on timesheets or resource management. Developers say that they feel a sense of ownership and get satisfaction from seeing their work have a direct and tangible impact on the business.

A highlight of our sprints is the all-hands show and tell, where the whole company comes together for an hour to see what we’ve achieved as a company over the past fortnight. Each delivery squad showcases what they’ve changed – where the emphasis is on what features have made it to live – and we often have presentations from sales, marketing, financial intelligence and other parts of the business too. 

How did it turn out? The first principle of the Agile Manifesto is: “Our highest priority is to satisfy the customer through early and continuous delivery of valuable software”. 

Helping people to prepare for retirement is a particularly tricky problem to solve because the landscape is so complex. Regulated financial advice is typically  out of the reach of  most people. 

Our culture is centred around focusing the whole organisation on a small set of business goals, and empowering small teams to find creative ways of driving the company toward those goals.

Our transformation work has allowed Wealth Wizards to begin shipping robust, well tested product features to their customers several times per week. The engineering organisation is now much more performant, while costing around £1 million less to run per year.

The velocity of our development function means that our product organisation is now able to perform several cycles of test and learn per month. This has proven to be invaluable in finding a user experience that really works.

We can iterate with speed, and have been able to concurrently try out many different approaches in live, gather data on how conversions are affected by change, and hone in on a mix of messaging and UX that has driven conversions way up.

Our Pension Wizard product was demonstrated at Finovate this year, to great acclaim. In the past quarter, we’ve launched both to our biggest client yet, and to our first FTSE 100 company.

This means that we’re now helping thousands of people get their finances back on track.

This article initially appeared in the Q2 edition of Financial Technologist

Blog

More blog articles