Engine Works
Under the hood of Alteryx: tips, tricks and how-tos.
Everyone talks a big game about “fast time to value.” And what’s faster than pizza delivery? A reasonable wait time for pizza delivery is about 30-45 minutes, which is pretty speedy.
But what's the wait time for data insights about the pizza? Not so speedy, if you’re not using an analytics stack that makes the process seamless from start to finish.
Like pizza, data insights are better when they’re fresh. Wait too long and the pizza isn’t as edible, and the insights aren’t as actionable.
When it comes to delivering timely insights, the Alteryx Analytics Cloud Platform really does enable faster time to value. Combine it with other cloud solutions like AWS, and you’re making the process even more efficient.
We don’t just talk the talk, though. The proof is in the pudding, or in this case, the pizza! So we tried it out with a simple pizza workflow of our own.
Let’s say you’re a business analyst for a pizza chain in Colorado (we’ll call it Hypothetical Pizzeria) and you want to examine last year’s performance to plan for the coming year. You have a feeling your sales aren’t doing too well, and your Chief Revenue Officer wants a revenue report faster than you can order a size large Barbeque Chicken.
To get the job done, you’ll need to get the data, transform it into a polished, consumption-ready dataset, then generate insights that will tell you what’s going on with revenue. Let’s get cooking!
Connect to AWS to get fast access to the data you need
If you’re a pizza chef, you wouldn’t want to spend hours looking around for ingredients before you can start making the pizza. Same goes for data.
Open your Alteryx Analytics Cloud Platform interface to quickly connect with a source like AWS and access your datasets. Use the Connections tab to create a new connection to the data source – in this case, our hypothetical pizzeria uses Amazon S3 and Amazon Redshift. You can also see your existing database connections show up on this tab.
Check out the Data tab to see the datasets you own, across sources. Neat, right? Here we can see the four files we need for our pizza revenue analysis.
The Alteryx Analytics Cloud Platform makes it ridiculously simple to set up connections and organize your datasets, all in your browser.
Build a workflow in Designer Cloud to prep the dataset and conduct exploratory analysis
You have the ingredients, now it’s time to get cooking. Open Designer Cloud and drop in the Input tool to add the four CSV files in from the S3 bucket – Designer Cloud makes it super easy to search for the datasets you want.
Then start exploring and transforming your dataset. Designer Cloud covers the tool categories you’ll need to cleanse, parse, blend, transform and analyze to your heart’s content. Just drag and drop away, just as you would in the classic Designer experience on desktop.
For this workflow, we did some prep, joined the datasets, added a column for “Revenue” (since we are making a revenue report, after all) and exported the final, cleaned-up dataset to Amazon Redshift.
If you’ve played around in Designer Cloud, you might notice something about the Interactive Results Grid: It refreshes the data each time you add a tool. So you don’t need to run the whole workflow to get a preview of the changes! You can quickly see whether you’re doing what you intended to do, like making sure your pizza has the right toppings before you bake it.
Hmm, speaking of toppings. Which pizza flavors were most popular at Hypothetical Pizzeria? Let’s check it out.
Apparently, the Hawaiian pizza flavor is in the top five, which is baffling.
So Designer Cloud is great for answering your data questions like “Who on earth enjoys pineapple on pizza?” And it’s also great for prepping a dataset for downstream activities like training a machine learning model or creating a report. Which is where Auto Insights comes in.
Pop that dataset into Auto Insights for smart, speedy intel
To export your final dataset from Designer Cloud, drag in the Output Data tool and choose where you want it to go. In this case, Hypothetical Pizzeria uses Amazon Redshift as our data warehouse for cleaned-up, consumption-ready datasets.
All you have to do next is upload the dataset to Auto Insights, then select what kind of Mission you want to create. If your datasets are the ingredients and Designer Cloud is the cooking, Auto insights is when you make the final delivery.
Since we want revenue insights for our Chief Revenue Officer, let’s choose the “KPI Deep Dive” mission template and select Revenue as our KPI.
Give it a few seconds, and voila! Auto Insights tells us what’s going on at Hypothetical Pizzeria.
Auto Insights doesn’t just tell us what happened. It also tells us why. Many a dashboard will let you know that your sales are decreasing. But you still need to figure out what could be causing the decrease. Auto Insights uses AI to read your dataset, detect meaningful patterns in the data, and identify trends that might be causing a change.
Auto Insights can show you something unexpected, that you might not have looked into if the AI hadn’t pointed it out. For instance, I might have thought store location was the only factor to examine. But Auto Insights is telling us here that it thinks Size Large pizzas are the top cause for the decline in revenue.
It also looks like our vegetarian pizza flavors are doing better than average. Who knew the “Vegetables + Vegetables” would be so successful!
The last thing we need to do is hit send. With Auto Insights, you can schedule the Mission to send via email in a clean, polished summary.
And that’s how you can deliver insights faster than the amount of time it takes to deliver a pizza!
AWS, Designer Cloud and Auto Insights are powerful tools to get a fast win – while also creating something impressive and genuinely impactful. And it all happens right in your browser, from start to finish (I did this all on my Mac).
Try it out for yourself! Get started learning Designer Cloud and Auto Insights right here in Community. And if you want to run a hypothetical pizzeria and play around with a fun dataset, you can find the files I used on Kaggle.
Resources:
