Product

Ergest Xheblati is a data architect and author. I asked him to be our first guess in our 4-part Metric Tree Masterclass because in his LinkedIn posts he always appeared thoughtful, well-researched, and balanced. And when I finally had the chance to meet him I was delighted that he lived up to my expectations and more.

I wanted to start this series with a conceptual and well-researched introduction for two reasons:

It helped set the bounds and define what we would (and wouldn’t) be covering in the series
It reminded us of why metric trees are so valuable instead of of simply because we’ve seen them talked about online.

Watch the full conversation:

Or get the summary:

In our conversation, Ergest and I cover:

the origins of metric trees
what separates metric trees from similar concepts like KPI trees (if anything)
what impact can metric trees make for a data team and for the wider business?
related concepts like problem trees, and The Theory of Constraints
where you should look for a metric tree ally
effective ways to introduce metric trees to business stakeholders

Key Takeaways

You can listen to our full conversation in the audio clip at the top of the page.

Don’t get caught up in what you call your metric tree. The simplest way to think of a metric tree is as a systematic understanding of your business. There are two fundamental things your tree needs in order to be a metric tree: a spatial layout representing relationships between metrics, and easily refreshed data so that the tree may serve primarily as a diagnostic tool for the business.
Metric trees are primarily useful for what Ergest calls ‘debugging the business’. This often means identifying the root causes of significant changes (e.g. a drop in revenue or website conversion). However, they can be extended to many other applications as well, such as: identifying growth levers, solving problems (e.g. how should be best invest $10K in our marketing budget), improving the design of experiments, and connecting people and teams across the company around a common understanding of the business.
Metric trees can have a significant impact on data teams too. They can help shift you out of being a service provider and provide clear signals to the business. They can also save data teams a lot of time investigating root causes instead of starting from scratch each time metric changes. Metric trees can even serve as mechanisms to make sure you are answering the right question. You can cross-check the tree when you receive a request from the business to make sure you’re understanding their context and working on the biggest problem, not just what they request.
Metric trees have many origins and related concepts. Abhi Sivasailam, founder of Levers Labs, and creator of SOMA, was inspired by the DuPont tree, which broke down ROE into its component parts. Ergest highlights other related concepts like The Theory of Constraints by Eliyahu M Goldratt, The Problem Tree, and the Profit Tree.
When considering how to start your metric tree, Ergest recommends finding an ally and suggests starting with Finance. They will have a vested interest in understanding these key metrics broken down across the business and will be relatively unbiased in their approach. And when thinking of where to start your tree, Ergest suggests it’s always best to start at the top - ideally with the company’s North Star metric, but it can also be done within individual departments - and work your way down.