Mindshare Customers Blown Away By Interacting with Data

One moment we had was when we combined several different data sources on a dashboard: the client’s website traffic, media activity, and call center calls. Over the course of a month, we were able to see that when the client was being advertised on TV, there was an increase in website traffic and the increase in calls to the call center came a few days later. So you could start to get an idea of what the customer journey looked like.

That dashboard took just five or ten minutes to put together, but traditionally would have taken time to do because it was three different data source and you would have had to join them together in Excel, draw three separate charts, and make sure the axes were lined up through all of them. With Tableau, it was relatively easy and gave instant insight into that client’s business.

It varies quite a bit. Sometimes it might be just 50 to 100 rows that have already been summarized, such as a search ranking of different key words from the client. It might be medium-sized, like thousands or tens of thousands of rows. That might be something like media activity by channel over the course of a year or three years. But it can also start to get bigger. With something like sales by product for 800 stores over the last five years, you’re starting to look at a million or two million rows of data.

And then when you start thinking about digital data, the sky is the limit. We’ve got some websites where the traffic log from the server might be 18 million rows over the course of a month. So, the great thing I love about Tableau is that it’s quite agnostic about what data it’s dealing with. It doesn’t matter whether it’s 20 rows, 200 rows, or 200 million rows—you’ve still got a consistent way of looking at the data, and it doesn’t fall over when you get past a certain amount. We’ve tried to push it to deal with really big data, and it does fantastically well.

It started off as an internal idea. A colleague on another team asked if we could look at the bus routes in London to help a client understand buses as an advertising vehicle. The insight that the team came up with was that buses were the chariots for 15-19 year olds in London as they traverse the city. We wanted to use them as an advertising platform for posters on the outside and also potentially for advertising inside the buses as well.

So we started by getting information from the Transport for London website, including where the bus stops were and a 5% sample of all Oyster card journeys for one week in November 2009. There was some hard work getting a hold of the data, but then it was quick and easy to use Tableau to turn it into something interactive. It was no more than three or four hours of work to produce something that looked fantastic and answered the questions.

You could bring it to life during the pitch by asking people in the room, “Which bus route do you take?” and then choose the bus route and show lots of data. Very quickly it shows that we understand data and we can visualize it well in a way that no slide saying we’ve got all this experience or anything else can really do. That’s how it started, and the final visualization that I published on Tableau Public just pushed that forward a bit by looking at different times of the day and different days of week.

There’s no other piece of software that I’d be able to use for something web-based and so interactive with in that length of time. I’d end up needing a lot of code, such as Javascript, and I don’t have the expertise or the time to write a lot of code. Tableau is a fantastic tool for very quickly visualizing complex data.

Yes, and we’ve got Tableau Server as well, but we’re waiting for a web server to deploy it. As we move to Server, we’ll be able to share a lot more data. Internally, we’re a company of about 400 people in the UK, primarily doing buying and planning of media. For them to be able to access all their media plans online and have a consistent source of data is going to be fantastic.

With Server, we’ll also be able to deliver dashboards to clients. Historically, we’ve used tools like Xcelsius or Adobe Flex to deliver to clients and a lot of the time the data has only been updated weekly and tends to sit in their inboxes. By having a really interactive way of engaging with clients and being able to share an analysis very quickly, we can give the client the opportunity to really delve into the data. One thing we’ve developed using Tableau is the Global Sports Index, a tool to help sports rights holders better understand their fans around the world. Again, being able to deliver that via the web gives us a fantastic marketing tool that we can share very quickly and easily with potential clients.

Speaking personally, I think it’s been a great tool to get everyone on the same page for visualization. Historically, we’ve only used Excel and one other piece of open source software for our visualizations. There was a lot of variation in what was produced and limitations with both. So, trying to look at data on a map or look at small multiples of data to compare different media channels was not easy and often very time consuming.

Now I can work on a project with even junior members of the team and say, can we look at the data like this? It allows me to delve in after a lot of the hard work has already been done. It’s absolutely fantastic for dashboarding in pitch situations. Within two weeks of our trial version, we were using it in a live pitch. At the time, we made it work on an iPad. It was slightly clunky compared to know, but it was still a step beyond whatever else was out there. It’s a fantastic way to engage existing clients, as well as new clients by showing our insights and point of view on the data in their market.

We haven’t looked at that explicitly, but I’m definitely saving a lot of time. Things that used to take two hours to do, I can now do in 10 minutes. And once I’ve done something, I can repeat it time and time again. Personally, it’s the best piece of software I think I’ve ever used.

For our team, I think it’s improved our output substantially. It’s taken us from having to present 15 different slides to having one dashboard and looking at the most important things. It’s also a lifesaver when we get a dataset to be able to check that the data is correct, or at least there’s not anything obviously wrong like double counting or missing data periods. For example, a couple weeks ago we got a data source sent to us and within a couple of minutes I was able to see that nine weeks were missing in the middle of it. This is something you’d typically not notice until a lot later down the line. Tableau saves us from having to perform a lot more laborious checks on the data.

The first thing is to make sure that you take the time to make dashboard design a process. The blessing and the curse of Tableau is that it’s very quick to be able to produce something that looks good. But it still takes a little bit of time and investment up front to understand what the client’s needs are, what data is most important to them, and what they want to see. Using wireframes to do that is a fantastic solution, before you start building the dashboard itself.

I’d also say that training is a good thing to have. In my experience, you can get about 80% of the way there just by using the tool on your own, but it’s that final 20% of the features that you find out from going to things like the Tableau Conference or having some type of more formal training session. Or you can watch the videos on the Tableau website. It’s just that little bit of investment up front to know what you’re doing.

And finally, I’d say data structure and infrastructure are important. Again, the blessing and the curse of Tableau is that it can connect to pretty much any data source, which allows you to be quite lazy about how you structure your data, but that tends to unravel one way or another somewhere down the line. For example, if you try to scale your data up past a certain point or when something changes slightly. So again, investing that little bit of time up front to make sure you’re storing the data the right way and you have the right hardware capabilities is quite an important thing.

In terms of analysis in the media, I think we’re getting into an interesting place now that the tools are there to do a fantastic job. We can get the data into a meaningful shape and start visualizing and analyzing it. The key then is to start having a point of view on the data. Historically, you might have a particular way at looking at the data or reporting on it that doesn’t necessarily give any insight to a point of view. Tableau gives you that flexibility to take the next step to something that’s more actionable and ultimately more insightful.

As far as Tableau in general, what has really struck me, coming at this fairly fresh, is the amount of development that goes into it. Even just thinking back to September, the degree it has moved forward in that time has been fantastic. Things like data extracts are now fantastically quick, which is really important. When we’re showing to clients, it’s no good if it takes a minute to load a new view of the data. They’ll lose interest, and they won’t see the contrast you’re trying to make between x and y.

Being able to show data on maps and how those have evolved is also fantastic. And now Tableau is being rolled out on the iPad. All these things really engage clients and make our lives a lot easier as analysts. It takes data from being something that is really dry—where you have lots of bar charts, lots of tables, and you pick out the one number that was interesting—to something where you’ve got lots of people engaging and starting to ask more questions. Depending on how you look at it, that’s either a really good thing or a lot more work. It’s a good sign that you’ve got people really taking an interest, and you can show them visualizations that really make the penny drop in a way that a table of numbers never would.