B&CE tasked RocketMill with building a digital measurement solution they could use to inform change across their business – and to develop it quickly.
As a rapidly-growing financial company, the business was generating increasing levels of data but were struggling to manage, and trust, it.
B&CE provides a workplace pension called The People’s Pension. As the demand for workplace pensions grew, so did B&CE, but with this growth came growing pains, one of which was data.
The brand has multiple domains and sub-domains, where employers, employees and financial advisors log-in. As a new product was developed, a new team would manage it and set up measurement methods. This led to an abundance of information, scattered in various places, which made it tough to tie up and make sense of collectively.
To tackle this, we began by carrying out stakeholder interviews. One of the biggest things we found was that, despite a real hunger to use data, many just didn’t trust it.
There were too many places to look for information, which caused confusion. B&CE’s experienced marketers and developers were struggling to find Google Analytics useful because the data was in such a mess.
When you don’t trust the data, it’s next to useless, so we had to come up with a solution, and fast.
Before developing anything, we set some guiding principles for the tool:
The vision we settled on was one, global Google Analytics account. It was paramount this was clearly labelled, to segment the different parts of their business, as well as provide a holistic view covering the entire company.
Previously, B&CE was using one Google Tag Manager container to measure activity and had set up tracking across all websites. There were multiple domains and sub-domains so, as a result, every time a rule was set up in a container – for example: “the URL contains ‘login/success’” – they weren’t sure it was firing in the right places.
To solve this, we created six different Google Tag Manager containers, one for each key section of the site. These would send information to the same Google Analytics property and consolidate it. This separation was important, as different areas have different tracking needs. Measurement strategies for an FAQ section and an account log-in section, for example, will be different.
We set up tracking in one of the containers and, as Google Tag Manager doesn’t easily allow you to copy from one container to another, exported and re-imported it as the global configuration, which gave us a sound foundation.
This saved almost a week’s worth of time – significant when completing such a time-sensitive project.
Now we had the set-up we needed, it was time to add business context. We didn’t just want to know the number of page views and events, we wanted to understand real business information. This included the number of people logging in, volumes of employees activating accounts and engaging with them, and amount of people opting out of auto-enrolment.
We didn’t, at this point, understand the revenue generated for the business through the enrolment fee for employers. We introduced user IDs, so we could merge information of what users were doing online with back-office systems.
The site also housed a lot of complicated forms, but no measurement was in place on who was using them and how. This tool changed that.
Finally, we provided extensive training and support for B&CE’s developers. We didn’t want to deliver this fantastic, but rather complex, setup without knowing it could be maintained.
We taught them to use the API, so when making a global change, instead of going through manually (and introducing the risk of human error), they could use the API to pull tags into other containers smoothly.
This training allowed them to actively engage with the tool. When developing a new tool – and they make a lot – this is now at the back of their minds, so they understand intuitively where measurement needs to be built into the process.
We built easily digestible, real-time reports, as well as Power BI dashboards, to provide B&CE with real business intelligence. More importantly, we improved the brand’s trust in data.
They have since introduced accurate forecasting which, when tested with back-dated information, has forecast to within 3-5% revenue figures.
The tool has also provided fresh behavioural insights, including the number of users logging in and cross-device usage. Other, more granular insights have also been unearthed; one example being that 7% of users are trying to reset their password but just 50% get to the point of being able to log in again. This highlighted a problem previously unknown to B&CE, which they can now rectify.
We also found 60% of employers signing up to the tool, were selecting “other” when asked to describe what type of business they are. This can be off-putting, as it suggests the brand isn’t tailored to support their needs. Again, this would never have been noted if it wasn’t for the tool.
Overall, we significantly improved B&CE’s data collection and analysis capabilities, improved the brand’s trust in the information they were getting and provided them with the capability to discover insights previously unobtainable, enabling them to make informed business decisions.