8min chapter

Zero Prime Podcast cover image

E25 [DEEP DIVE]: How Metric Trees Are Transforming Businesses with Vijay Subramanian

Zero Prime Podcast

CHAPTER

Harnessing Metrics Trees for Organizational Success

This chapter explores the power of metrics trees in optimizing organizational management and decision-making. It emphasizes their role in connecting data and business goals, enhancing collaboration, and leveraging generative AI for effective analytics.

00:00
Speaker 2
And you've written about the potential for Metrics Trees to serve essentially as a system of record, supporting a range of applications around metrics like monitoring and anomaly detection, as well as experimentation and even some of these business review, planning, goal setting, use cases that you mentioned. Can you share a bit more about what your ultimate vision is for how metrics trees can transform an organization?
Speaker 1
I think you've pretty much articulated it, but yeah, because metrics is what are they really capturing? They are capturing the core business processes, right? Whether it's the growth model through acquisition, to retention, whether it's the consumer engagement model for a consumer company, it's how these metrics all ladder up to financial metrics, like revenue, margin, even EBITDA and whatnot. So I think the trees are capturing the business process, which is why I talk about them as potential systems of record, because they're really pushing the abstraction up from tables, columns to metrics to now really a metric tree, which is really the business, right? Is really the business model. And once you have them, a lot of things become, I would say, easier and even elegant. Like let's say you want a root cause, why a output metric at the very top is behaving a certain way. You have the relationships all the way underneath, right? You have the relationship model. So root causing just falls out by traversing the tree, decomposing, asking, you know, doing some analysis, doing some math, running some algorithm, you have the answer in almost instantly, the software can provide to say, well, you're down on revenue by 5% because it's coming from this specific customer growth problem that you're seeing in this specific channel, right? So the debugging is almost instantaneous. And again, like you even pointed to the other applications, right? So like detecting anomalies automatically, because we have the full tree, we don't just have one dashboard, we can see how changes in one thing affects the other. So detecting anomalies, root causing experimentation, pre-post analysis, I launch a feature, I want to where which metrics are being affected, because it's very rare that only one metric gets affected. Often you see one do better, the second one do worse, you want to know what the net impact of that is, right? So, all that just falls out very naturally. And this is all by the way, just I'm talking purely about diagnostic analysis on respect to data, not to mention what it can do in terms of a proactive or a predictive sense, right? Because you're like 15 days in and you want to know what's going to happen end of the month. You want to forecast that. You basically have the relationship of all the metrics. You can forecast that. You can ask yourself, if I move this metric at the bottom of the tree by 5%, how does it cascade all the way up to revenue? You can ask that question. A lot of things that we'll probably take today, a team to pull some data, build a model in a spreadsheet and maintain it, it pretty much falls out the moment you map the data to a metric tree. So I'm pretty excited about that as becoming a fundamental piece of infrastructure, a system of record on which if you do these applications so much, so many common workflows just fall out very naturally.
Speaker 2
Yeah, the root-causing stuff is obviously very interesting. And I know there's been several people at Data Council over the last year or so that have been talking about such benefits of metric trees as well. So I think that's particularly exciting when you start to develop this organized kind of opinioned way about the ontology of the business, if you will. These are some of the side effects, the happy accidents, if you will, that follow such a process and seems to be quite powerful. So you wrote a piece on Medium right after Rent the Runway IPO where you reflected on how important it is for both data nerds like us to get out of the technical silo and interact and learn with the business side of things. So, I was wondering, do you feel that metrics trees are a good tool in the street guard? I mean, obviously you do, but maybe expound upon how they help bridge the gap between some of the technical folks and the business folks.
Speaker 1
I mean, this has been the biggest learning for me in my career. Now I'm an HD, an OR, and I started my career thinking very technical, building models, solving technical problems. But the thing I've realized over time is different functions just have a different mix of personality traits or attributes, if you will, like builders, operators, advisors, three things I think about a lot. So engineering is a builder function, customer support is an operator function primarily. I think data is one of those interesting functions in a business where data and finance teams are advisory functions. We're trying to inform how to make the best decision possible. But what's interesting about data, I will say, like in the last seven, 10 years, as I mentioned, as we've made a huge shift to the cloud, a new plethora of tools have emerged to make data modeling easier and data operations easier. I definitely noticed that data teams and data leaders have, I wouldn't, maybe I'll say regressed. They've regressed on the advisory side of the equation, on the operator side of the equation, and being more like, well, I'm going to just build these things and I'm going to enable you, you business user to figure out what to do, right? I mean, here's all these dashboards. Here's all these pieces I've built together. All my data models are here. All my fact tables are here. And the dim tables are there and now you go have at it. I think that's not bad necessarily. I think it's good. It's good to have a builder mindset, but I do think there's a bit of regression in that they've lost some of the advisory and the operational elements in their job. Because ultimately, a data team is not rewarded for producing more dashboards. I mean, that's not the goal of the data team. It's rewarded for optimizing the business. We're really utilizing data to make an impact. So yeah, I do think data folks, yes, retain the builder mindset. Think about not repeating yourself. Think about building blocks and all of that, but definitely get out of your technical box. Really, you have to understand the domain. You have to understand how marketing works, how they think, what the incentives are, sales, finance, operations, product, the functions you're supporting, you really have to think about that functions and really, you know, you know, think like an operator, I mean, and that's, and if you do more that more and more, I think you will be more successful in your career. You'll be very effective. And sort of bringing it back to the metric tree. I mean, I think metric tree is, in my opinion, the right framework for this. Because it is the system of record, to use that phrase again, that connects the data to the business in a very explicit way. So it's like it brings the language of understanding a common understanding between data teams and business teams. that can be very powerful for both parties, not just the data team.
Speaker 2
And so to summarize, what benefits do companies get from adopting metrics trees? So
Speaker 1
one, you get, as I said, violent alignment across the data, exec, and business teams on how the input metrics connect and ladder up to the outcomes for the business. Two, it supports the scientific process of understanding what drives what, and you keep refining these input output equations. So these trees are actually fluid. And three, in a very practical sense, every data operation in the business is just easier, faster, richer, monitoring performance, root causing, business reviews, feature reviews, pacing, forecasting, planning, all the way leading up to really optimizing the business process.
Speaker 2
And of course, in today's world, we can't not talk about the implications of AI. So... Oh, I was wondering when that's going to come. You were waiting for it and you got it. So how does the recent rise of generative AI impact your thoughts on the space? Well,
Speaker 1
I'm sure the investor, you're probably more plugged into what the different folks are working on. But from where I sit, I see folks trying to solve the problem of asking questions and getting answers on data using generative AI, which has basically been called text to SQL, right? Because the business team doesn't know SQL, they don't know the data structures. I'm sort of skeptical that that is actually possible in a general sense without a very, very good backend, a good metrics layer or a good metrics or semantic backend. But moreover, what I would say is if you model your business through metric trees, you have the systems of record, it gets very powerful because now Genet can do what is good at, which is it can utilize this rich metrics backend, which already runs and produces data and explains what's going on. And what it can focus on really is generating narratives, because ultimately, narrative is a very critical output of a data team, sort of telling stories around what is happening in the business. So I think what Geneic and Truly Shine is probably framing the narratives more so than doing the heavy lifting around the back end, the data, the metrics, and the calculations. I think that's probably better left to a different kind of software. This is my humble opinion.
Speaker 2
CB Well, your insights have been really valuable, Vijay, and I've appreciated you taking the time to share with us your thoughts on the space, wishing you the best of luck with Trace, and I'm sure we'll see you around the data community sometime soon. Thank you so much for the time. This has been a pleasure. Thanks for listening to the Zero Prime podcast. I hope you enjoyed our deep dive into the metrics world with VJ. Follow VJ on X at V-J or learn more about Trace at hellotrace.io. If you like hearing from engine your founders on the cutting edge of enterprise startups and developer tools, please leave us a review on your favorite podcast app and subscribe to the show. We'll see you next time.

Get the Snipd
podcast app

Unlock the knowledge in podcasts with the podcast player of the future.
App store bannerPlay store banner

AI-powered
podcast player

Listen to all your favourite podcasts with AI-powered features

AI-powered
podcast player

Listen to all your favourite podcasts with AI-powered features

Discover
highlights

Listen to the best highlights from the podcasts you love and dive into the full episode

Discover
highlights

Listen to the best highlights from the podcasts you love and dive into the full episode

Save any
moment

Hear something you like? Tap your headphones to save it with AI-generated key takeaways

Save any
moment

Hear something you like? Tap your headphones to save it with AI-generated key takeaways

Share
& Export

Send highlights to Twitter, WhatsApp or export them to Notion, Readwise & more

Share
& Export

Send highlights to Twitter, WhatsApp or export them to Notion, Readwise & more

AI-powered
podcast player

Listen to all your favourite podcasts with AI-powered features

AI-powered
podcast player

Listen to all your favourite podcasts with AI-powered features

Discover
highlights

Listen to the best highlights from the podcasts you love and dive into the full episode

Discover
highlights

Listen to the best highlights from the podcasts you love and dive into the full episode