Making data work for people not used to working with it.
Redesign of Alfie - a software as a service program used by hotels to connect guests with their staff in a way that increases hotel revenue and guest satisfaction levels.
1. The Problem
Alfie has been great at generating valuable data for hotels, just not at sharing it.
Since its 2018 launch Alfie has enabled guests to free themselves from the room phone and order food and drinks or book services such as a spa appointment from anywhere in the hotel. This system generates a lot of valuable data. However, hotels can’t extract this data themselves, and have to wait for a monthly report to be manually created and emailed through to them by our client, meaning that it was up to a month out of date before it could be used.
2. Task
Working with data that is a month out of date created challenges for both hotels (who need access to data that is as close to real time as possible, in order to make informed decisions across their departments) as well as for the company behind Alfie (which works on a commission model - the more revenue that goes through their program, the more they make).
Our task was therefore to create a self-service system for hotels to access and analyse the data that Alfie generated on demand and in a way that could provide insight and guide decision making in relation to staffing levels and product offerings.
3. The Research
We knew that hotels couldn’t extract any data from Alfie, however we didn’t yet know what data they actually did need. What we had was a laundry list of ideas provided by current and potential clients, but no way to prioritise these, or know what else could be useful to them.
Time was of the essence, we needed to design, carry out and analyse our research within three days to meet our delivery deadline. We knew that we would need three types of research:
Qualitative - to explore what types of data hotels need
Quantitative - to help prioritise needs
Competitive - to ensure that Alfie’s offering remained unique within the marketplace
I wrote our discussion guide and led the design of our survey. In addition I undertook a review of analytics software, with a particular eye on functionality.
In order to supplement our research, I also conducted an audit of online search behaviour in the travel category, in order to increase our understanding of the services that hotel guests want to use while they are on vacation (particularly at resorts, Alfie’s core user base).
With our entire team of four carrying out qual, quant and desktop research over several days, Miro became our shared workplace. A series of one on one interviews with guests and hotel management (both in person and remotely via Skype), completed surveys, search analysis and our competitive product audit generated a lot of rich data for us to analyse.
Individual user journey maps were created for each qualitative research respondent (top left). Video interviews took place with hotel management respondents located internationally (top right). I conducted a review of online search behaviour in the travel category (bottom right). I conducted a competitive review of analytics packages (bottom left).
A mixture of interviews (in person and via video), quantitative work (including a survey and search analysis) and competitive review informed our discovery.
4. Research Outputs
After all of our raw research data was charted we commenced affinity mapping, which produced a wide range of insights, initial ideas around functionality and a user persona for Alfie’s customers in hotel management:
A range of insights were formed from our research data(top).
We charted various options on a mock-up of the analytics page – our original “prototype” (above)..
Meet Thomas Peters – the user person I developed for Alfie’s hotel management clients (left).
5. The Co-Creation
With a lot of data to work with a mid-project client review wouldn’t be enough, we needed them to jump in and work with us in a co-creation workshop:
The co-creation workshop saw us present our personas, insights and the ideas for functionality of the analytics package. A series of exercises with our clients involved them giving a business priority to our ideas on functionality and helping us define what this looked like via crazy 8s.
Now we knew user and business priorities, it was time for us to design the product itself.
6. The Ideation
I co-led the design of the analytics functionality. A quick decision was made from our persona – our audience primarily featured users for whom English is a second language. Fortunately, numbers are a universal language, so my designs focused on them:
My sketches used a dashboard with a range of customisable widgets that could track the performance of data created by current and prior guests (by importing TripAdvisor data), compare it to past performance and export it to other systems.
Widget design emphasised numbers and graphical elements such as arrow up and arrow down due to varying levels of English amongst Alfie’s user base.
7. The Prototype
It was time to check in with the team.
Decisions were made quickly – we would feature key metrics as a series of widgets, leaving space for charts that would give depth to our headline numbers. This achieved a range of objectives – it shared data quickly, in a language that is common to everyone, and would facilitate informed decision making. Sharing this data was key – we had found out that this was typically done via screen capture, so we decided to offer it through the ability to export a range of different file types.
The analytics page is customisable, with hotels able to choose what is displayed via widgets. Widgets show change over time (the time period is also customisable) and one widget has its data charted over time.
Users can dig into the data by selecting an individual widget, this will automatically chart the data for them.
As well as being able to view and dig into the data, users can export it via a number of formats - CSV, JPEG or PDF - depending upon their needs (guidance is provided to users as to which format to use).
Our prototype was ready, now to test it with users.
8. User Testing
As with our initial research phase we did both domestic and international user-testing (via a desktop share using Skype).
Testing indicated that a number of updates would be required – renaming items in the navigation and being clearer with the comparative timespans on our charts. However, one user did call our work “utterly simple!”. It was clear that our design philosophy of leading with the international language – numbers – was working.
Conducting international user testing via Skype enabled us to hear what the user was saying, see their reaction and track their cursor’s movement across our prototype.