EXPEDIA GROUP TECHNOLOGY — DATA
How to Build an Actionable Scorecard
Impactful techniques in Tableau to simplify your data
The disclosed data is anonymized. No interpretations or decisions about Expedia’s business performance should be made based on this anonymized data.
No business in the world has escaped the impact of Covid-19, but with stringent travel restrictions continuing globally, travel has been especially hard hit. Expedia Group™ has demonstrated that a culture of data-driven decision making has ensured strategic and quick reactions to understand this changing market context, allowing us to adjust our marketing strategy accordingly. Expedia’s Data Visualisation Team developed a Market Recovery Scorecard to synthesise all the data points across the broad portfolio of global brands needed to help support the business during this critical time.
The business problem
We needed to rapidly understand how the business was being impacted so that we could make the necessary adjustments until the industry began to show signs of recovery, at which point we wanted to be able to react in the right way.
The Market Recovery Scorecard needed to summarise metrics in each of our key markets that could give a daily readout indicating at a glance:
· Is it currently possible to travel either domestically or internationally in this country (are there restrictions in place)?
· Are customers beginning to book in these countries again?
The Data
Our data was split into two sections: firstly, internal data covering a few indicative metrics: Search, Qualified Shoppers, Purchases, Cancellations, Flights & Car Hire. For this, we simply surfaced these metrics over a time series for each country. We collaborated with our data science team to run algorithms that removed anomalies and detected genuine increases and decreases in these metrics. However, to make a more basic version of this, it is also possible to show the % difference across a given time period and flag when it’s above or below a certain threshold. Our use case had to be more robust than this.
The other half of the data was gathered from various external sources which showed: travel restrictions in place in the country, a stringency index (an aggregated score showing the restrictions in place) and mobility data (showing whether people were travelling in that country). We covered more on this in a previous blog post here.
Finally, we also allowed a certain amount of user input: for instance, certain users were allowed to change the ‘Phase’ of each country (a 1–4 value, 1 being fully locked down, 4 being returned to normal).
Visualisation
The most simple thing we could have done would be to enter this data directly into Excel and refresh daily, which would look something like this:
However, this could be improved upon. A lot!
1. Highlight the high-value countries
We did this by adding a small dollar icon next to each country. This can be incorporated into the country name row itself:
This calculation uses the previous year’s booking value which was available in the dataset. It calculates the WINDOW_MIN(MIN(value)) and the WINDOW_MAX(MAX(value)) and finds the percentile that the value sits within these. Then rounds it to the nearest .20 to get a 1–5 value which then is put into a case statement to convert it into a value between $ and $$$$$. Then a remainder is added in which is the inverse of this to give us the gold and grey dollar signs.
2. Show which countries are at which ‘Phase’
For this we replaced the 1 to 4 value of the country’s ‘Phase’ with something more visual.
This is where we start introducing shapes. Tableau actually has some pretty good shapes for this in the Bars folder which would work fine. However, we created our own for a neater look. To add these just go to your My Tableau Repository > Shapes and add a new folder. In Tableau just assign the discrete value to each shape.
These shapes work great for quickly scanning down the page. This column alone could be enough for some users so it had to be eye-catching.
3. Visualise the travel restrictions
For this we used an online resource that contains stock icons and found a globe and a house icon to indicate whether there were international or domestic restrictions in place. We followed the same process as with the above shapes for this. Using these shapes combined with the colours allowed us to add in a really clear ON/OFF Switch for each country and as shown above, give another column that you can get your answer from in a matter of a few seconds. In this screenshot you can see that certain countries have eased restrictions domestically:
A great resource for shapes and icons is www.flaticon.com
4. Visualise the Country mobility Data
Stringency and mobility data could quite easily be shown as a bar chart. However, as Stringency is the opposite of mobility it works well to show Stringency as a negative and Mobility as a positive so you see the whole thing shifting to the left where there are a lot of restrictions in place and to the right when the country is recovering adding another bit of top-to-bottom readability.
5. Create a grid of indicators
Next we get into surfacing the indicators provided by data science that identify whether a metric is showing signs of increasing, decreasing or stable. We also had a few other fields to play around with that showed whether a metric was increasing at an accelerated rate. If you don’t have your own crack team of data scientists to hand, then a similar set of indicators can be produced by some table calculations looking backwards in your data that show the same results.
As the goal of this scorecard is for the user to be able to scan it in a few seconds and get a good idea of what is happening, we decided to use indicator arrows rather than show the numbers. For this you could always use the standard arrows in Tableau. However, as with some of the other icons, we sourced some slightly nicer icons from the web instead.
6. Provide a whole extra layer of information in the tooltips!
We wanted to keep this view as simple as possible as the main priority was reducing the time to insights in this scorecard. However, by adding tooltips it encourages users who wish to do so to explore the Market Recovery Scorecard in greater depth.
7. Proactively answer questions by adding an info icon
This is an addition that we add to all of our visualisation products. It just utilises Tableau’s collapsible containers functionality, replacing the default icon with an information icon. Use this as a place to add any FAQs as well as any other additional information about the data:
The result is a simple and clear summary which shows at a glance how all markets are performing across all metrics, as well as pretty much everything else you need to know. What’s more, is that there is a whole extra layer of insight by hovering over almost every point, inviting your user to explore.
This approach can be applied to anything and is a great way of showing your high-level metrics in one place, one level above a detailed deep-dive / dashboard.
To explore the Market Recovery Scorecard in more detail, check out the full version on our Tableau Public page here
