Rethinking the roadmap – using data to release the right features
We’re always looking to improve how we operate and serve our customers. Recently, we’ve been extracting more value from data that had been subjective, scattered or unreliable and using the resulting insights to inform our decision-making. As a result, we’re now able to deliver sought-after features more quickly than before.
At Access Worldpay, we love metrics. We can’t get enough of raw data that tells us in real time what’s happening, how we’re doing and, crucially, where we can be better. We’re always striving towards achieving and then tightening that classic loop of continuous improvement, in which different stages of our process inform and reinforce each other to help us deliver a faster, better and more responsive service.
In the past year, we’ve focused particularly on making better use of data that’s previously been under-used, usually because it’s time-consuming to find or because it’s imprecise, subjective or unreliable in some form. By gathering this data faster and more systematically, it becomes more relevant and reliable; this then gives us more confidence in using its insights to drive our decision-making.
Working both ends
A key element of this project has used the principles of Lean SDLC to automate the collection and streamline the use of important back end metrics, such as Change Fail Percentage. As
At the same time, we’ve been concentrating on the front end of our process, working to extract more value from the data we gain from existing and potential customers and then using the resulting insights to respond to their needs faster than before.
From sketchy & hazy…
Conversations with customers have always proved the best source of information and insight but its value had previously been sketchy. This is because the information was ad-hoc, collated manually from individual conversations with various solution consulting teams and implementation consultants.
It was hazy, too. “We weren’t getting the data from people who were talking to customers directly,” says Tom May, Product Manager at Access Worldpay. “Instead, everything was one stage removed – somebody talking to someone else who’d spoken to a customer.” This led to the loss of quality information, often because of accidental omissions, mixed messaging, or both.
…to trusty & timely
In 2020 we began to make this process more systematic by collecting the data automatically on Salesforce. This is the tool used across our business by implementation managers and implementation consultants to contact and email customers and prospects. We also use it internally to store information about backlogs and manage our pipeline, which also proved helpful.
Inside Salesforce, we inserted a series of tags and fields within specific interfaces. “These became mandatory for implementation managers, implementation consultants and so on to fill out at the right stages,” says Tom. “It meant that the right data was entered in its original, raw form at the appropriate point."
The simplest way to identify customer priorities was to log the features that they were asking for, especially where these highlighted current gaps in our functionality. This let us see how many customers wanted a particular function; slicing and dicing this data – for example, by region – meant we could drill down further so that we could identify not only the number of customers wanting a particular function, but also their location.
Rethinking the roadmap
After starting this process, it took a while before the information became truly usable. “We were still getting false positives and false negatives and other bugs, and we weren’t able to fully trust the stats that were coming out the other end,” says Tom. “So we went back and iterated on that month-on-month, making it incrementally better each time.”
While these refinements continue, we were able to start using the dataset from early 2021. Specifically, we can now chart the most requested features against our current roadmap. “It’s opened up a whole new conversation about whether some items should be re-prioritized because of customer demand,” says Tom. “That’s a huge benefit to our customers, and also being able to make these decisions based off data has value for us, too.”
Although some requested features require little more than providing a new option or an extra field, others are more complicated. In these situations, the harvested data helps us identify the relevant customers: “Our solution consultants can then talk directly with them to help flesh out their requirements, so we know that our feature will actually address their specific needs,” says Tom. “It means that we’re not just faster, but more responsive too.”
For example, recently we re-prioritized developing new functionality to cover Mail Order Telephone Order (MOTO) transactions that happen away from the customer’s website. “The data showed how much demand we’d got for this and we didn’t have a MOTO solution,” says Tom. “Thanks to this insight, we quickly produced an MVP solution that looks after our customers now – we also made sure that our full MOTO feature was bumped up the roadmap so it would be released sooner than originally planned.”
A broader, better future
Early results like MOTO show the potential of this work and further improvements are on their way. Besides making the information more user-friendly, Tom says: “We’re going to establish more robust processes around how we get projects onto the roadmap, as well as clearer thresholds to determine when an item becomes important enough for us to give it higher priority.”
There’s also the bigger picture. This project is a first step in our much larger goal to automate data collection for the full end-to-end life cycle. This will provide a comprehensive data-driven “insight platform” that will improve how we make decisions and will, therefore, tighten the loop of continuous improvement in everything we do. Besides Lean SDLC, this platform will include automated metrics from other key areas, such as integration, so that we can remain confident that Access Worldpay is always the fastest, easiest platform for our customers to join.
For now, our more systematic approach to using front end data has helped us prioritize the functionality that our customers most need so that we can go live faster than ever. In 2020, our average time from idea to production release was just 25 days, but we’re always looking to improve – and continuing to make better use of customer information definitely helps. “At the very least,” says Tom, “we’re acting on hard data rather than simply building features based on our own, subjective opinions of what we think we most need.”