Time to Talk: Alteryx
So I've recently had the experience of diving into Alteryx for the first time. It’s been something that we've looked at, but we just didn't have the time or opportunity to really go for a deep dive into the world of Alteryx. Recently, as part of the last Iron Viz feeder for Tableau Conference 2017, I finally got the opportunity to use Alteryx in anger for the first time.
It's a revelation. Seriously I know that sounds a little cliche, but it is a real breath of fresh air. The general experience of data prep and pre-Tableau analysis has been connecting or importing data into Excel and then doing some work there, before passing it into Tableau for the next stage of analytics. Or using SQL or creating a simple application to do the transformations necessary. All tedious and to a greater or lesser extent time consuming and in a business context costly to maintain and keep functioning.
Excel has a big problem though, well it has many problems but there are two big ones for the purposes of data prep and ETL. Excel crawls to a halt the more rows and the more columns you add into it, and Excel can become a pain to maintain proper auditable and repeatable processes for ETL. Most clients realise this but I'm not sure that everyone is aware of the difference Alteryx can make, or are willing to spend money on another tool.
So let's break it down and take a look at some common tasks for ETL that I came across in this particular use case and ones that I know would be an issue in a business.
Mapping different groupings and classification; This is one of the common tasks that you may want to do and one that can end up as a big drain on your Tableau install, You can, of course, do cross database joins in Tableau but you may find that the more mappings and other joins you are doing, the more time your Tableau extracts take to refresh. By doing your mappings in Alteryx you can offload the work of refreshing extracts from Tableau and improve your end user experience. This also moves us onto our next point.
Removing unnecessary information. It may seem sensible to keep as much information in your data sources as possible, but realistically the more columns your dataset has the slower you will find your Tableau workbooks to be, so removing the columns you know you don't need can help speed up your workbooks. Additionally, if you are preparing data sources for self-service, it's best to keep the data sources limited to what you really need and not overload your users.
Pushing the data to a server. Alteryx has the option to create Tableau extracts but it also includes the option to publish the outputs to a variety of different locations including publishing back to the server the outputs of the workflow. This means that Alteryx can sit as the interface from several operational data sources and a Tableau data warehouse.
You can do it all again and again. Alteryx allows you to save and publish workflows to a server and then schedule them to run on an automatic cycle simplifying complex ETL tasks and allowing the task to be examined and understood by interested parties. The ETL process is not a hidden and opaque black box. But open to auditing and validation from stakeholders.
Now there are a lot of ETL tools on the market but Alteryx excels for the same reason that Tableau does in the visual analytics space. It is a tool that does not require its users to have complete technical mastery of programming or SQL, putting the power to analyse and prep data in the hands of the business experts. With a drag and drop interface sitting on top of a highly sophisticated set of very configurable tools you can create and maintain even the most highly complex ETL process.