Turning messy data into organised data for analysis is a solved problem if you're a data engineer - but you shouldn't need to be a data engineer to understand your employees. Just like we're using embedded finance to provide flexible access to wages in our Financial Wellness product, we've embedded advanced data processing into our Insights product.
I’ve worked closely with data for many years. I’ve run machine-learning teams, business intelligence teams, and built software to make data more accessible to the people who can make best use of it. In my experience, “build the thing” tends to be pretty straightforward!
We know what we want to achieve and know what metrics we want to show. Massaging the data so it conforms our expectations is the tricky bit - getting it from a → b and making sure it's got the attributes we expect, with the quality we expect.
With our tools, we’re attempting to achieve three things:
- If you use multiple workforce management systems, or the same one with different configurations - it should be easy to pull this data in and data from those systems should behave consistently.
- As little configuration as possible - all you need to do is tell us the “shape” of your business, we handle all the query-building and processing behind the scenes.
- You can see useful numbers immediately without having to build anything, write any code, fiddle with a graph-builder.
There are a few ways this can go wrong, that we aim to solve in our tool.
Integrations are maintained by us
An integration is when we copy data into our database to process and analyse it. We need to maintain integrations to support instant withdrawals in our Financial Wellness product: so we’re experts at it. When they change, we’re a step ahead - you won’t notice a thing. There’s no need to download an Excel sheet, and no need to delete/rename data points that might have changed since the last time you ran an analysis. This data changes often - any friction in getting the data means it doesn't get in front of the right people in a timely manner 💩.
You sign in, connect the accounts, and we make the data fit - from any workforce management system. Using departments over here and multiple accounts over there? No problem, we can make it behave cohesively for analysis. Once connected, that data turns up every day and gives you fresh data to make informed decisions.
Data’s rarely perfect, but that shouldn't matter
Humans are producing the data - and we make mistakes. We usually see around 2-4% of records are corrupt in a way that would make analysis of those records impossible. These errors are things like a “hired date” being after a “quit date” 😵💫
That shouldn't stop you analysing the data - perfect is the enemy of good! So Data Coach just quarantines them and lets you know they need to be fixed to have the most accurate picture. In some cases, Data Coach just uses a different value - a created date instead of a hired date, for example.
Fixing those issues shouldn't be hard
There are billion-euro companies and products built around fixing the Data Coach quarantined records problem. AWS Glue Databrew, Google Dataprep - these are great products. But they’re built to service every use-case - and you need to be able to write code (or be willing to pay someone) to do that for you.
We have the advantage of knowing exactly what kind of data we want to show in our tool, and exactly where it’s coming from. So we just do it for you, and tell you exactly what you need to do to fix the issue. Those issues can be broken down into the same shape you described your business in, so the issues can be assigned to the person who knows how to fix it.
Don't let data wrangling get in the way of understanding your employees!