Intro

I’ve spent a year and a half at Treety running Product & Data - and occasionally marketing, managing the entire tech team, strategy, research, design, UXR, and so forth. It’s a start-up, you wear many hats.

Picking just one key project to highlight my time there wasn’t easy. After all, I was managing half the employees and making key decisions for the entire company. So I went with largest impact that stood out the most.

For context, Treety was not the most resourced company at that time- between relatively small funding and team size, it was a challenge to build great things. And with inheriting a legacy product that was designed for a slightly different audience and use-case, it made the PM role very hands-on and very focused on making things happen against these constraints.

Fortunately, we live in 2020s, where most things have been built by someone else.

The challenge

Treety started with a relatively modest feature set: being able to request, aggregate and calculate metrics across each asset and fund - but not across time.

So for clarity, requesting data is just asking the portfolio companies of a fund to fill in a form.

Aggregating metrics is adding up or averaging all the value to get the metrics for the fund overall.

Calculating metrics is using some other metrics to derive a new one - i.e. dividing CO2 emissions by company valuation to get CO2 intensity.

Adding time to data is never easy, especially with a custom solution. For once, you move from a 2D table (metric x asset) to a 3D table with lots of metadata (metric x asset x time + time added, change log & value properties/attachments). Now all the calculations and aggregations are infinitely more complex - especially as you might receive different metrics at different time periods and frequencies. On top of that, you need to somehow visualize this complex data.

Without this feature set, we couldn’t reasonably compete for larger clients who relied on quarterly or monthly reporting and needed an in-depth visualization system to build reports.

The project

The solution we came up with was pretty simple:`

  1. Start by standardizing the dataset → fill in all missing dates and break-down monthly/quarterly metrics into their daily parts.