Saving lives with smarter decisions: Operations Research in animal shelters

Ritish Chugh:

I think my advice is I think treating analytics more like a trust problem rather than just a technology problem. I think it’s at the end of the day, whatever tools the data and analytics organization is really building, it’s for the leaders and humans. I mean, the easier we can do to make that easily digestible, as well as something that has validations built in, is trustable, as well as at the same time is consistent, as well as has transparency built into it. I think it can make organization life a lot easier, as well as even increase the trust for the analytics organization to be the front runner in terms of building technology solutions.

Ashley Klimp:

Organizations have more data, more dashboards, and more sophisticated analytics tools than ever before, and yet many still struggle to make confident decisions quickly. Sometimes the problem isn’t the data itself, it’s what happens between the insight and the action, unclear ownership, competing versions of the truth, slow review cycles, and analytics that sit beside the business instead of helping to shape it. In today’s episode of Resounding Human, the Informs podcast, I’m joined by Ritish Chug, senior analytics engineer with Airbnb, to talk about what it takes to move from producing analytics to building an organization that actually knows how to use it. Ritish, welcome to the Resounding Human Podcast.

Ritish Chugh:

Thanks. Thanks a lot, actually. Thanks

Ashley Klimp:

For having me. So you’ve worked across several industries and organizations, helping teams build analytic solutions that support important business decisions. From your experience, what’s the biggest challenge organizations face when trying to become truly data-driven?

Ritish Chugh:

Yeah, that’s a great question. So I think it starts from within. It’s also one of the major challenges that I’ve seen in organization, even major organizations is although everyone spends a lot of time and effort in having the best tools and having the best infrastructure and even having the best talent in engineers, but there’s not enough emphasis which is given on building the right foundations of data. And there’s a concept called as governance. It is normally taken a little lightly or it’s added after the fact where some solution is built out and you add that on top. But in my opinion, that needs to be really built in. It needs to be included right from within. It needs to be embedded as part of your data strategy. I think that is the fundamental difference between some organizations which are thriving with data and some organizations trying to play catch

Ashley Klimp:

Up. So many companies have obviously invested heavily in data platforms, dashboards, and now AI tools, yet I think we still hear about organizations struggling to make timely decisions. Why do you think that is?

Ritish Chugh:

I think it’s based on trust. I think at the end of the day, it’s a human who’s making a decision, even though that is served to that person through a machine or to a bunch of algorithms and whatever that is. It is also related to a lot of gut feeling and that trust really needs… It’s a cultural thing. It needs to be built within the organization. And the way it can be done is something that is consistent and something that is more transparent, something no matter how many times you write a manual query or you’re getting that information from a dashboard or these days through AI, it needs to tell one single coherent story. The metric has to be the same each time and across geographies, no matter what it is. So that’s what getting consistent answers each time with any person who tries to get it is the way that you can build trust with leaders and your peers.

Ashley Klimp:

So Ritish, when you’re designing an analytics solution, how important is it to understand who will ultimately make decisions with that information?

Ritish Chugh:

It is very important to know your audience, I feel. I would say in my experience of working with different leadership, they have their own way of digesting information. Some folks like to really dive deep into their data, for example, finance people, they don’t really like the charts and graphs too many on their dashboards. They really want us to give us the data. We want us to dive deep. We want to double click and see how the trends are going, and we also want to be able to download that information and do our own analysis with it. So that’s where it becomes really important to know your audience. Similarly, a lot of executives, they don’t want to connect the dots. They want the information to be given to them so that they can act on it quickly and make the right decisions for the stakeholders. So that’s what it is.

I’ve seen a little difference in the way different folks consume information, so it is very important to know your audience who you are building it

Ashley Klimp:

For. All right, Ritish, it’s time for our Ask AI segment where I have posed the question to AI, in this case ChatGPT, and read the question and the response and then would love to get your thoughts on it. So the question I posed, “What’s more dangerous for an organization having bad data or having good data that nobody trusts?” And AI responded, “Bad data can lead to bad decisions, but good data that nobody trusts can lead to delayed decisions, no decisions or decisions based on instinct instead. The real value comes when people understand where the numbers came from, know how much confidence to place in them, and know what action should follow.” So Atrish, which problem would you rather inherit, bad data or lack of trust in good data?

Ritish Chugh:

That’s a good question. I think bad data I feel is more of a infrastructure problem, and I feel like it’s better to have good data than nobody trusts because at the end of the day, trust needs to be built and trust can be built through transparency as well as repeatability. So at least if we have the good data, if we have something that is reproducible, it is just a habit that can be inculcated within the organization. So like I mentioned earlier, one of the major ways to build trust with the organization is having your data, democratizing your data. Secondly, having it traceable to each metric where it really came from and having solid controls around as to who updates that metric, when that metric is updated to or demand controls like get controls, version controls so that it is completely auditable. Even if you are at a time of IT audits or any kind of stocks audits, it can be completely traceable and it can be tied to what business information or objective you’re trying to track with this metric.

So having that all information available, I think it’s very critical to build that trust. So I feel like having good data with nobody trust can still do better than having bad data at all because it can lead to disastrous results in terms of if anyone acts on that information on the company’s behalf.

Ashley Klimp:

So obviously we’ve talked a lot about trust. This seems to be one of the biggest factors in whether analytics actually gets used. In your experience, what helps organizations build confidence in their data and analytics?

Ritish Chugh:

I think one of the major things is having the right governance. I think I stressed on this word way too often because I think what it really means is having ownership. So I think various people or folks within the organization needs to have ownership around if there’s a certain metric that is being referred, it needs to have complete traceability as to what it means. It has one single fundamental definition which is used across the company. So meaning if you’re talking about revenue, for example, revenue means one single thing. It doesn’t mean that sales team considers revenue to be everything that is booked to them, doesn’t mean it’s provided, whereas finance might consider it to be revenue at a time of offering the service, means the service is yet to be offered. So similarly, all the assumptions need to be considered together and it needs to have one single coherent definition that company wide can be used.

Second is repeatability, like I mentioned earlier. Now you have your AI models training on this information, and one of the things which I’ve seen in the past is these days where companies are actually competing to roll out AI across the company, but they don’t have solid fundamentals or foundations of this information. It can lead to a lot of problems in terms of AI failing fast, meaning AI can confidently give you the wrong answer and make you convinced for it that this is the right answer. But the way everyone can get the same information, consistent answer is having the right solid foundations where it tells one coherent story. So like I mentioned earlier, having the right traceability, reproducibility, as well as transparency is extremely important to build trust.

Ashley Klimp:

So speaking of AI, this is obviously something that makes it even easier to generate insights and recommendations, but as you mentioned, that’s sort of a double-edged sword. How do you think AI changes a trust equation? Does it make building trust even more difficult?

Ritish Chugh:

These days, I think it is harder in the beginning and I think it’s like a curve, you can say it has a harder adoption in the beginning and it tries to wither down as in when you have built enough amount of trust and it leads to consistent answers. But again, with AI, like I mentioned earlier, it can give you a wrong answer confidently fast if you don’t really have the foundation strong. So no wonder it is a very exciting place to be in right now, and it is so easy to really just converse with AI and gives you the answer without really depending on a human analyst to build that for you. And that’s where the problem comes in, because at least in terms of a human, they can crosscheck and give that to you, whereas AI just gives it to you without even validating.

So that’s where I feel that kind of, I would say, balance needs to be maintained and having the right investing in terms of your foundation will really help you scale AI even faster and then the organization, I feel.

Ashley Klimp:

All right Ritish, for students and early career professionals who are listening, what’s one lesson that you wish someone had shared with you about making analytics impactful and not just technically correct?

Ritish Chugh:

Yeah, that’s a great question. I think the most important thing I feel is the idea of a great storyteller. So you understand data, you can query it in the information, and at the end of the day, you are working with humans who rely on that data to make their decisions, and decisions are very important decisions for the company. So it’s very important in that case is that to be able to connect the dots and being able to produce your output in a very simple and easy to understand digestible format. So that’s how it distinguishes a regular data analyst from a storyteller is one that someone who really works with their stakeholders to really understand what they’re trying to track and how they want to achieve that rather than working on the other side where you just build something and hand it over to your stakeholder.

I think that’s where the difference really comes from. So my advice to earlier career students as well as early career professionals is learn and be curious, understand about data, understand about the fundamentals, learn SQL, master it, and engage with all the big data tools as well as visualization tools, and also ask about what business really wants to do with this information, what they’re trying to track, how it can help them grow, and what can you do better to remove any kind of redundancy with it? So that would be my advice.

Ashley Klimp:

Excellent. Very good advice. All right, Ritish, we’re onto our rapid fire question section where I’m going to ask a series of questions and just say the first thing that comes to mind, your first instinct, okay?

Ritish Chugh:

Okay.

Ashley Klimp:

All right. Bless you. Dashboard or conversation?

Ritish Chugh:

I would say conversations.

Ashley Klimp:

Okay. Speed or certainty?

Ritish Chugh:

Certainty.

Ashley Klimp:

Yeah, that one’s pretty obvious. What’s one analytics buzzword you’d retire if you could?

Ritish Chugh:

I would say data governance.

Ashley Klimp:

What do you think is the most underrated skill for an analytics professional?

Ritish Chugh:

Storytelling. I think it’s very, very important for any data profession.

Ashley Klimp:

All right, I’m going to ask you to finish this sentence. A mature analytics organization…

Ritish Chugh:

Has the right data foundations and knows what questions to ask to the business.

Ashley Klimp:

Excellent. All right, one final question. If you could leave our listeners with one piece of advice about creating organizations that make better decisions, what would it be?

Ritish Chugh:

I think my advice is I think creating analytics more like a trust problem rather than just a technology problem. I think it’s at the end of the day, whatever tools the data and analytics organization is really building, it’s for the leaders and humans. I mean, the easier we can do to make that easily digestible, as well as something that has validations built in, is trustable, as well as at the same time is consistent, as well as has transparency built into it, I think it can make organization life a lot easier, as well as even increase the trust for the analytics organization to be the front runner in terms of building technology solutions.

Ashley Klimp:

Today’s conversation is a reminder that analytics maturity isn’t measured by how many dashboards an organization has or how sophisticated its technology stack is. It’s measured by what happens next. Can people trust the information? Can they understand it? Can they act on it quickly? And is analytics helping shape the organization’s priorities rather than simply reporting on them? Ritish, thank you for joining me and sharing your perspective on what it really takes to bridge the gap between data and decision

Ritish Chugh:

Making. Thanks. Thanks a lot, Ashley. Thanks for having me. It was a pleasure.

Ashley Klimp:

And thanks to everyone listening to Resound on the Human. Be sure to subscribe and share this episode with someone who has ever sat in a meeting where everyone had the same dashboard and somehow still have five different answers. Thank you so much.