Trusting the Data: How to Turn Insights into Action | E25 – Hard Refresh – An ISSA Podcast

Sam. Welcome everyone to another episode of Hard Refresh. And on this one we have an incredible guest today. I'm going to emit no words in his introduction because I'm not going to do the introduction. He's going to introduce himself. Tim, take it away.

Hi all. My name is Tim Moriarty. I'm a co founder of Data Insights Consulting. I've been a data nerd my whole life and I've been in the data engineering data science space for over a decade now, so hoping to provide some insights into that environment.

I'm so excited about this. I was telling Kayden and Tim is probably one of the smartest people I have ever met in my life. I have the pleasure of working with him as we were building out udis and we have. Every time he speaks I'm like, what is this magic that this man is saying? Like, it's just so incredible. I'm so excited because this conversation is so incredibly timely and there's so much value that Tim has from a data scientist perspective and just the build the things that he's done. So I'm so excited. I know he's also part of the tech and innovation committee, so you're going to see him a lot more with Dave as well. But I'm so, so pumped for this session.

High praise.

It's going to be a lot of fun. But before we dive into this session, you know, Tim, I'll be honest. I thought you were going to go the AI roasting route with your tech background. But Tim has chosen to do two truths and a lie. So let's dive in.

I got to keep you on your toes. All right.

I know.

So two. Two truths and a lie. I a three time state champion. I've had a personal phone call with the greatest defensive player ever, Lawrence Taylor. And I have a Power Bi dashboard that generates my workout routine.

Holy smokes. Okay, okay.

Can you define the first one?

Yeah.

What? State championship?

Yeah, the championship is in weightlifting.

I'm gonna say he does have a Power Bi dashboard for his workouts because I've seen what he's done with Power Bi and so that's like, I feel like a cakewalk for him. So. Yes. And then he's had a conversation with. Who was this now? Lawrence Taylor.

Lawrence Taylor, Defensive end for the Giants. Long time ago though.

Okay. I'm gonna say that is true. And don't take my advice, Kaden, because if history has any type of weight, I'm always wrong.

I wonder if the fact that he has a Power Bi dashboard and the weightlifting three time champion are correlated. So I would say that the second one is actually false.

Okay, you lead. All right, tell us.

All right, so Karina was actually right. I'm only a two time state champion, not a three time.

Always in the details.

Oh my gosh. Holy.

Always in the details.

That's impressive though. Holy. Okay, so this state champion, when you were in high school or like recently, what's happening with that?

Yeah, so I'm the 2013 Florida High School Weightlifting individual state champion and also the team state champion. So the two medals that you see back there are from that state championship, actually.

That's awesome.

So you're a jock and a nerd.

Yeah, that's a combo. Usually people like find one out and then are shocked on the other like, oh, you're actually smart. Wait, you're strong.

Wow, that is a powerful combo. Watch out. Love it.

That's.

That's cool. Well, we always get caught in the details. So there's a, there's probably a life lesson in there somewhere, you know.

Yeah, always.

That life lesson is not going to be taught today on our episode. So we're going to dive in. We have some quick control. Karina's going to tell us about an upcoming event and then we'll dive into clearing the cash.

Yep. So we do have the webinar that we hosted live. So if anyone missed it, reach out to us. We can share the link below on this episode for the webinar on client retention that Sager Billy Dana and Mike Fitz did. And it was incredible. Great feedback and overall such strong content. So do check that out. And then we have a CLX that we're hosting in person in Vegas, but we also have one in September. Don't we have another one that we're doing? Yeah, right.

Not we aren't doing the September event anymore. So the committee is, is focused on planning the CLX in, in person to make it a home run. So if you haven't decided to come to Issa in Vegas, not a very smart decision. So time to make a smart decision and plan on coming. There's going to be a lot of things happening at Vegas this year, as there always is, but put it on the calendar. It's always worth attending.

Yes. So, and I think that is it besides veo, of course, which we've already done the VEO takeover, so I won't bore anyone with that. But just kidding. Please come to veo. Make sure that you support the Hispanic community. Very important. Coming up, 30 days away and momentum continues to build and our very own Kaden is emceeing the gala shameless plug.

Alongside my right hand man.

It's gonna be Sam the man.

Sam. Sam. I am not me. But anyways, yes. Even though VEO is not a Innovation and Technology Committee affiliate per se, it is Issa, and definitely one worth promoting as our last episode was taken over by that. So those details are in all of the good social channels. They have been distributed, so please look out for those. But let's dive into our topic today. We have Tim and Karina. You mentioned he is the smartest person you know.

Yeah, yeah.

And that is very fitting with the topic we're about to dive into today, one that probably all three of us are pretty passionate about. Tim just has the upper hand on being like, you know, in the back end of code and data science that you and I probably don't spend our days in. So with that, we're going to be talking about data's impact now, Tim, the amount of data that's growing and existing in the digital world today. I was at a conference last week where someone said it doubles every couple hours now, like the amount of data that is being built and stored. And I don't know if that's long term sustainability issues or not. That's not what we're here to talk about. But what I want to talk about is trusting in data, which I actually, I just did a. A religious talk on Sunday about data, interestingly enough, about trusting data and AI and people hiding behind data and all of that. But we want to take a different approach today and we want to talk about your deep experience in trusting analytics and how that gets built out. So let's start off with what we could go very deep. We're not going to, because our, our audience isn't like that deep. But in this surface level, it's like, let's take analytics and dashboards, for example. In your experience in connecting data points, you mentioned Udis, Karina. What is it that gets ignored compared to what actually drives action when it comes to data?

Yeah, so I think there's two really big things. You know, it's common nomenclature, which, like engineers always miss because they're so caught up in the back end. And my math is correct. Why do I need to actually have the terminology right? And then the other thing is not just insights, but actionable insights. So it has to be able to actually drive a business decision. Those are the two biggest things. So I could be 100% right in my calculations and 100% right in everything. That I'm doing. But if I call this a can and you call it a container and we're not in agreement on that, it doesn't matter what I say, it doesn't matter how right I am, I'll never get the message across to you. And that is what is usually missed in these dashboards, especially as data gets more complicated. Get caught up in the complexity of the data minutiae and not the simplicity of what the end result is supposed to be, which is driving actionable insights for that end stakeholder.

So a follow up question to that. What has been the most successful data insights that actually produces action? Like think of typical business metrics for enterprises or different companies you've helped in your work. What are the things that the industry, the cleaning industry as a whole should be looking at as a metric that actually makes you take action or shares something to make a business decision off of?

Yeah, I think a couple of big ones here. Retention is a huge one. You know, it's great that you're getting new clients and you're getting new revenue, but are you retaining the revenue that you're bringing in? And it's, it's, you know, it's dollars you already have. So let's optimize the strategy to keep those jobs going because that is one of the big things that I'm seeing in this industry is there's a focus on growth without a focus on retaining what we've already captured. So I think that's one of the biggest insights is where are we weak on retention and how do we improve that retention?

I think you touched on a really important part in terms of like not just retention but expanding, peeling back the layers of for example, the nps. And I've really understood the power of NPS score. And the input is the, the power of the input produces the output. And what I mean by that is like that you capture the NPS in different ways from an in person to a data collect, to a survey to a phone call. And to Tim's point, like that type of data that does start to paint a picture not just from a client retention of like hey, we're doing great, but also of warnings of saying like, hey, this is something to be cautious about, the NPS is tracking low or something. This allows you to have those guardrails and really have a clearer line to retention and also expansion because you can say that the NPS paints a picture of the growth of that client and hopefully the expansion of that portfolio. Should you have that ability to be a national Partner, right?

That's exactly right.

Thank you.

Like, just validated, you know, nothing, Nothing else to say. Karina, mic drop. So, what there. In my experience, there is a certain hesitation that comes with going all in on data and trusting a data source. What do. What do some of those foundational trust issues stem from, do you think?

Yeah. So, you know, I as, you know, a consultant, I'm coming in as an outside vendor. So it's a huge thing to build trust. So there's always inherent biases, and one of the biggest ones is the survivorship bias. So that term comes From World War II planes that came back. They re armored the planes based on where the bullet holes were. Seems logical, except the planes that were shot down were shot at different spots, so they actually needed to armor where it wasn't hit because that was where the planes that they lost were hit. So biases like that are coming in. It's, you know, I've done it this way for 15 years. Why would I change? And I also have a gut intuition of what I think the data should be. So those are usually things that inherently make it difficult. Is that gut intuition of, like, you know, I've got a brain, I've got this super advanced, you know, algorithm that I store up here, and it's telling me that my gut intuition says that the number should be 12%, and you're telling me it's 22%, why? So the way you know that you get around the bias and the trust is one, you need to have a really, really good discovery on, like, what you're trying to do and what you're trying to maintain. Two, you need sources of truth. You know, what is the actual source of truth? This. Let's build trust. So, like, I have validated down to the penny on your transactions for this period in time. You can see, let's go through it. But now, what I found, I found discrepancy and what you interpreted 12 months ago versus what I'm seeing now. Here's the discrepancy. Here's the three reasons supporting it. So, like, when you're going against a stakeholder's bias, one, you need to build trust by confirming some of their old data. So find data that they, they already had, confirm it, show them that. And then come with a disagreeing factor and have supporting reasons on why you disagree and open the conversation with them to be like, okay, so this is where we're at. Why, why is this? Is there a caveat that I missed? Or was the reporting different in the. In the structure 12 months ago and the, you know, inherent assumptions were wrong. So that's a really big thing is building trust. And the first thing to build trust is at least validate some of the previous data and then move forward to where you disagree. I love that.

Can I ask you that is. Go ahead, Karina.

I was just going to say, Tim, this, I'm so interested in this aspect of it because I think data governance, we're working on this right now again. But like, from auditing cap standpoint, what would you recommend is a good auditing process if you are going to invest in doing this single source of truth and it really relies so heavily on data. What's a good auditing process for companies to have that they can understand that they can ensure that there is data validity in these systems?

Yeah, so there's this concept called the Gartner model of analytics, which has like four stages and there's the descriptive, diagnostic, predictive and prescriptive. So when we're in that diagnostic stage, what you want to do is you want to make reports and have those reports agree with each other, you know, because a lot of times you have the foundation of what businesses are reporting on, whether it's finance, operations, you know, customer service. At some point, there's a KPI that they all agree on. So make sure that your OPS data and your finance data on the core KPI are agreeing with each other. And you could set up flags on that too, where it says, you know, you know, ops versus finance, you know, marker for revenue. Do they agree? If they don't, then we've got a problem because then our core foundation is wrong. And that's really where you build it. Like get those diagnostic reports right. And then let's start moving on to the descriptive. Then we move to the, you know, predictive. Let's predict what we're going to do. And then the prescriptive, where I've did my diagnostic phase and now I know I can prescribe solutions based on the data. And it really is, you know, how do you trust the data? Well, yeah, you hit it over and over and over again and make sure that you're getting the same result.

Yes, a hundred percent. Because that's the reality is, is that as soon as you have a KPI that says, hey, one system source is saying this and the other system source is saying this and they're misaligned somewhere something's wrong. It could be in both systems. It can be one or the other. So the, the foundational layer of the data coming in and where it's being stored. Is it accurate, Is it complete? That's, that's huge. What a. So with that, I want to dive into maybe an example. I, I think of one that. A very, very light example. I was working with a school district back east and people complained certain things were or were not happening. And they were able to use data to defend everything against the claims against the board of education for the cleanliness of the school district data. Super powerful. Do you have stories like that, Tim, where you can bring some insights or some, some lessons in what data has actually done for, for some of your customers or people that you've worked with?

I'll bring in a couple of different ones here. So I'll bring in one where there was, you know, both, both have positive outcomes. One was highly emotional, one was not highly emotional. So I'm going to start with a highly emotional example. I was working with a larger client who had staff on site, was in the medical industry. So a lot of data. It's really important to be able to access your data at a transactional level quickly. You know, a lot of reports are analytically grouped. We want to be able to, you want to be able to drill into the transactions as well, quickly. And here's a great reason why. This client had a bonus structure for their employees based off of a database. It was originally through Excel spreadsheets. So, you know, the people that were getting bonus were kind of frustrated because they used to be able to manually enter and now it's off of this database that they, you know, can't manipulate the numbers, per se. They had one client who, you know, was like, no, I released 500% of what we're saying here. My bonus should be higher. And when we went into their tracking sheet versus what was in the database, the rep actually went back through and realized, oh, no, I, I entered an extra digit. So, you know, it was highly emotional because the manager was like, no, this is one of my best employees. I know they wouldn't lie to me. And eventually the rep was like looking back at their sheet and said, no, no, my fault. I, I entered a number wrong. And once we got through that once, it was like, wow, it caught one of the best employees errors. The client was sold. The second example was I was working with a smaller consulting firm and a tangential industry, home industry, the home remodeling industry. And they were doing a lot of sales, but they had recurring. They were doing, they were tech sales in this industry. So they had like, you know, bigger recurring revenue Packages and the sales reps just kept trying to sell the largest and largest packages. And eventually, you know, I put together a customer lifetime value model for them. And I was able to show the CEO and the director of sales that it was actually the mid tier clients that were making the revenue. And, you know, the account executives didn't want to go with it. I mapped it out for them. I was like, guys, if you consider your callbacks like you're actually making more money on these mid tier ones. And to convince them, you know, I kind of had to do a little bit of a unorthodox solution. I made a bet with the head account executive and I said, all right, listen, you take this for two months. You know, we got, we got the callbacks coming in with, with this, we got a two month and a six month evaluation. @ each evaluation, if I'm wrong, I'll take you out to Fogo to Chow. He took me out to Fogo to Chow because I was right. So those were two situations and we actually just had one where we were going against old data versus new data. We found among 5 million transactions, we found that we were off by about 550 bucks, able to quickly find where it was. Somehow in an ETL process, a single GL transaction was dropped. And we were able to find that within the first three days of that being dropped across 5 million different transactions. So those are just a couple of examples that, you know, when you find those discrepancies that are so small at the transactional level and they buy a huge amount of trust.

Too.

Yeah. I think the granularity of that approach too, though, Tim, speaks to the trust that you organically build with data. Like if you start through that, what was the first stage of the model that you said it.

Was? The, the descriptive. What is.

Happening? Exactly. So it just feels like it's almost like a. You start to date and then you get engaged and then you get married. Right. That's how it feels. You start to really have this relationship with data and then you rely a lot of trust and then you can shape decisions based on it too. So I think that's one of the things that I've taken away is because ultimately what happens when you talk about data, and not just data, but tech in general, there's a lot of emotional intelligence. Kate and I talk about this all the time. There's a lot of emotional intelligence and cultural mind shift or paradigms that you have to deal with when people are being so closely, are working so closely with tech and to have that human behavior element as part of this in the conversation of learning how to trust the data, work with data. It just, it's very critical to growing the organization, I.

Think. And, and Karina, you brought up a great point. Emotional intelligence is huge here. Especially engineers are stereotypically not great with that. One of the big things that you'll always find, especially if you're coming in consulting or coming in from a different department and you're doing data analysis, is people are always going to be fearful that you're replacing their job. So you need to highlight to them, like, what is the worst function of your job right now? Let me take that off your plate. And just like, I'm enhancing your job, I'm not replacing it. That's a huge thing when coming.

In. Well, but I think too, like, a lot of times that I was, when I was working at Pioneer, we would start to report on inspection metrics and present it to the client engagement team. And there was so much just. People are getting very defensive, right? They're like, well, you know, that's. That score's not right or something happened or they weren't there. And it's like, guys, it's not. We're. We're tacking the problem. The actual data metrics of the deficiency. It's not about that. You didn't do your job, right? That's not. What we're trying to, like, talk about is actually look at the actual metric that you're measuring, whether it be a star rating, whether it be the type of task deficiency. Let that tell the story. Attack that and not the person. Tim, to your point, like, let's take the person part out of this unpacking, but understanding that the person does play an element of, like, handholding and emotional intelligence. But if you break some of that wall down and show them like, this is not a. Not only is it not going to replace your job, but it's not a personal attack to you that you're not doing your job well. We're just trying to grow. And the data is going to help us do.

That. Exactly.

Yes. This. This reminds me of. There's a bunch I started thinking of, but have you guys seen the movie Hidden.

Figures? I actually have not seen that.

One. Oh, yes, I.

Have. Yeah. So it's, it's a, you know, there's. There's scenes in it where they're like, they're trying to compute all the math for launching the, the rocket, NASA's rocket, and getting the astronauts up in space and all of this. And I think of all of the data, like this is really before computers really had, you know, the level that they have today, obviously and they're trying to like run the machines against the main actress and her ability to do the math. And it's like, no, the machine's actually wrong. You know, I think of where we've come from from that obviously that was a movie that is, is not necessarily all real life and all of the, the scenarios that happened with it, but I just, I think of that where, where we've come as a society when it comes to data in building trust. Back then it was like, that's just. Anyways, it's what I thought of it was just like, nope, I'm going to trust her over this.

Machine.

Right. The machine got the data wrong. We aren't necessarily in that today as much as we were then. And I think of AI especially as it's going to start processing things at a much faster rate, being able to drive insights. That's kind of the last question and then we'll wrap up today's episodes. Tim, is that where do you see AI playing a role in this whole scheme of data and integration and trusting.

Data? I think AI is going to open the door a lot for people who understand the concepts of data but don't want to learn the syntax of code. So, you know, I'll be the first to say it and LLM right now kicks my ass at syntax, but I kick its ass at context. So where I see it and where I see people adapting is, you know, you're never going to be, you're never going to have robots working for you, but you could have cyborgs working for you. And what do I mean by that? It's a lot easier to navigate LA with a GPS than it is without. And it's the same concept here. It'll be a lot easier to navigate data with an LLM assisting you versus you by yourself or the LLM by yourself. It's going to be a hand in hand partnership of humans and AI moving.

Forward. I think that's the best illustration anyone has given in all of our talks.

Karina. I love.

It. I told you he is a wealth of knowledge. We need to have him back on the episode. I think if we need to do like a Tim takeover of Hard refresh to do the entire.

Show, I love it. Well, Tim, that brings us to a wrap. What's the one piece of final advice, the one golden nugget? What's the one thing you want people to remember about data analytics and.

Integration. You know, know the question that you're trying to ask before you find an answer. And what I mean by that is there's a famous example of some statisticians going on football, and they found the number one cause of punts was fourth down because they didn't know the context that they were doing the analysis on. So know the question before you provide the.

Answer. That's a funny example. I love.

It. Yeah, that is a great one. Love.

It. That's good. Well, awesome. Tim, it's been a pleasure to have you on the show today. Thank you for your time and for our audience. Hopefully, you caught something that was super insightful for you to take action to go find the right data. That's a wrap on Hard Refresh. Thanks.

Everyone. Thank you.