And if you’re lucky, your organization will have a dedicated data team whose job it is to work with you and get your data in order. But many organizations don’t have dedicated data teams; or worse, they do have dedicated data teams but they don’t have enough time to focus on what you need, because this quarter the CEO said we’re focusing our resources on Sales initiatives. Whatever the reason is, data teams (like software engineers) are and probably always will be a scarce resource. When I was on the data team at DigitalOcean, I think the most common Slack message I sent (statistically) was “Sorry, I don’t have any bandwidth for this right now.”
So what’s a non-coding marketer (or sales leader, or designer, or product manager, or operations lead) to do?
The original Alteryx product: visual pipelines#
Alteryx actually started as a research project at UC Davis, and was incorporated as “SRC” in 1997. The types of products they were developing then were really all over the spectrum, with one theme: they helped business teams organize and analyze their data. The “modern” Alteryx era really started in 2006, when the “Alteryx” product was released (which obviously eventually became the company name).
They raised their first round of funding in 2011, and then raised a lot more before going public in 2017. But the company announced in December 2023 that it's being taken private by a group of investment funds. But enough history. What's the product do?
The core Alteryx product helps non-technical people just like you move and clean their data, so they can get it ready for analysis. It’s basically a visual workflow builder:
You drag different operations onto the canvas, and use the GUI to tell Alteryx what you want to do with your data sources. This is a marketing workflow that pulls in two data sources – advertising clicks, and sales leads from the data warehouse – and cleans and merges them. Let’s take a look at a few specifics.
The advertising click data has the timestamp of each click recorded in a weird way. This workflow splits that column into two, separating each by the space (‘ ‘) in each.
The split column functionality in Alteryx
After doing some cleaning on the sales leads too, the workflow merges the two datasets:
The data merge functionality in Alteryx
And then after that, it summarizes some aggregated data and exports that to some Excel sheets so the team can figure out who to run promotions against.
What makes this tool powerful is that it doesn’t require any code to use – everything is graphical. And though there’s definitely a learning curve, it’s very doable for functional teams themselves (in our case, marketers) to do what would traditionally be “data work” by themselves.
There’s a pretty big catalog of things that you can do in the workflow designer tool: filter rows, create formulas, sort data, pivot data, join two or more data sources, extract text, you name it. And once you’ve defined a workflow, you can save it and re-use it for any new data that comes in, schedule it to run at specific times, etc.
Alteryx since then: making use of your cleaned data#
Outside of the workflow designer (which, to my understanding, is the primary Alteryx product), Alteryx provides a bunch of other products that help non-technical teams get use out of their data. One way of framing it is that the workflow designer helps you get your data in the format you need it in: these other tools help you use that data and get value out of it. Let us peruse a few products.
Auto Insights#
Yes, you guessed it. Auto Insights gives you automatic insights.
You pick a dataset and a metric that you care about – let’s say total spend for a sheet metal manufacturer – and Alteryx runs some automatic analysis to help you figure out what’s changing and why. In this example, it looks like spend decreased by 20% this month, mostly driven by lower expenditures in engineering and construction. I didn’t need to write any queries to get this info, which is neat.
Machine Learning#
Alteryx ML lets you build Machine Learning models on top of your data without writing any code. You start by picking your dataset – in our case I uploaded the database database. You then choose which column you want to “predict” and what kind of model you want to use.
It then picked a type of algorithm for me, and told me how accurate the predictions were:
Not bad! (Although in practice, it probably is pretty bad)
This is where things get a little confusing when it comes to the target audience. A lot of the configuration you need to select when using Alteryx ML – like the type of model you want to use, looking at evaluating how accurate your model is, etc. – requires a good deal of domain expertise. I understand what it means because I have a Data Science degree, but I’m not sure it’s obvious to the average Joe. But I digress.
Honorable mentions#
Out of respect to you and keeping this post brief, here’s a truncated view of some of the other Alteryx products:
- Location Intelligence: a giant map that you can overlay your data on, draw shapes, zoom around, things like that
- App Builder: a way to build interactive apps (with filters, dropdowns, etc.) that use your data
- Reporting: something like a dashboard builder, where you can add tables, images, maps, and the like
And there’s even more! It’s a pretty giant suite of tools.
The astute reader will digest this post and think: Justin, I feel like you’ve described tools before in this very newsletter that do exactly this. And you’d be right; the analytics space is a crowded one, with hundreds of overlapping tools. And the lack of specificity on marketing websites only compounds the problem.
Could you use Alteryx in place of Looker? Probably. If you’re using Airflow, do you need Alteryx? Maybe. Aren’t there like 10 startups working on visual ML model builders? Also yes. Analyze at your own peril.
