Alter Everything

MaddieJ · ‎12-07-2020

Steve Mann, Managing Director from Propel32 joins us for a plethora of tips for analytics process efficiency! He reminds us why it’s important to use simple examples when communicating your insights, and why basics like low hanging fruit can be much more than just basic.

Panelists

Maddie Johannsen - @MaddieJ, LinkedIn, Twitter
Steve Mann - @Propel32, Propel32 website, LinkedIn

Topics

Transcript

Episode Transcription

MADDIE: 00:00	Welcome to Alter Everything, a podcast about data science and analytics culture. I'm Maddie Johannsen and today I'm joined by Steve Mann, Managing Director of Propel32.
STEVE: 00:11	I'm a proud resident of New York City and a native of Perth, West Australia. Hope my family don't hear me say it in that order.
MADDIE: 00:20	Steve has a background in finance, has lead teams at PwC, and is currently focused on using analytics and M&A, mergers and acquisitions. As Steve is in New York, he's had a very close look at the devastation that COVID-19 has had on the city and decided to give back. Propel32 partnered with NYC Service and New York Cares, the largest volunteer network in New York City and used Alteryx to provide actionable insight into their data at a time of true crisis. So throughout our conversation, you'll hear Steve talk about the importance of analytics basics and why simplicity should never be underestimated. Let's get started. I'm interested to hear, then how it became kind of happenstance for you to move to New York?
STEVE: 01:10	So happenstance for me-- and Alteryx actually has a part to play here. So I spent my whole career up until launching Propel32 at PwC. I was working up in Toronto. I was up in secondment. And I was doing my normal sort of M&A related work and started playing around with analytics tools. Where I was looking at my career and thinking, "Well, I love doing this work. I love analysing businesses and helping clients in an M&A environment, but if I've got to look at these data problems for the next 40 years, I'm going to be pretty unhappy." So started to play around with analytics tools that would help me get around those problems and ended up spending more and more time solving commercial problems with these analytics tools and sort of pivoting from the traditional work that I was doing. Ended up getting a tap on the shoulder from the firm to see if I was interested in coming down to New York to build a team focused on analytics in M&A and while again it wasn't on the agenda before that the appeal of coming down to New York from a professional perspective as well as a personal perspective, it was too hard to say no to. So bit the bullet. Moved down to New York two and a half years ago now and really feel ingrained or starting to become ingrained in New York and haven't looked back. So I'm really grateful for the opportunity to come and do that. It's opened up a world of opportunities, both professionally and personally as well.
MADDIE: 02:46	So yeah, Steve, with your years of experience leading teams at PwC and now as a Managing Director at Propel32, I'm sure you've seen analytics change over your years, and I'm wondering if you can share with us how you've seen that change happen or what the change has been?
STEVE: 03:06	Yeah. Certainly. And I would say so in my field, I'm really focused on analytics in M&A. So a lot of my clients are either private equity clients or corporate clients who are looking to buy or sell a business, and they're looking to understand as much as they can about that business in a very short space of time. If they're selling a business, demonstrate to a potential buyer where the growth opportunities exist and get those buyers really comfortable that those opportunities are real with a really strong analyses. With that in mind, the concept of analytics is one that has always been of interest. I think in the last four or five years its gone from something that's interesting to something that people appreciate can actually provide a competitive advantage in a deal setting. So the pressure to get it right or the pressure to get the ball rolling has definitely picked up. I think many people in this space, or many clients in this space, are trying to work out the best use cases. I think a lot of us in analytics can appreciate that analytics can mean a lot of different things and I think the challenge is finding the best use cases that have the greatest impact in a deal setting. And look analytics M&A is tough. It's really short timelines. You have very limited access to management. Often, the data or existing infrastructure isn't set up to analyse the business in the way that the business is going to be analysed for the purpose of a transaction. The stakes are really high. Someone's going to buy or sell the whole business. So it's a high-pressure environment.
STEVE: 04:53	And equally, there's a need for simplicity because a lot of people need to understand what's going on so really complicated solutions aren't going to work. So it's a tough environment. And because of that, where I've seen the highest and best use cases or highest and best use of analytics is really spending a lot of time on developing a really powerful descriptive analysis that's sitting on top of an enhanced dataset that's had a lot of scrubbing that has been pulled together from a number of different data sources and that has had simple things fixed. I'm sure many of us have felt the pain of a same customer name that's entered four different times. Simple things like that, that can cleaned at speed so that we have an enhanced dataset to drive a powerful business analysis that can be delivered within weeks. They're the highest and best uses in an M&A environment from what I've seen. And I think the market is starting to dip their toe in and try different things. And I think as we go along in this journey, we will continue to evolve to more sophisticated use cases. But where I see the market right now is a desire to find those use cases and those who have figured it out are able to create a competitive advantage by finding new insights that otherwise would have gone unseen in a very short period of time.
MADDIE: 06:33	Yeah. And a couple of things that really struck me as you were talking, I love that you mentioned kind of the-- you kind of sort of alluded to the chaos that can be out there when it comes to applications or platforms or trying to pick the best data strategy or how do you even get started. And I think that that can definitely be overwhelming. But as you mentioned one of your focus areas is just really trying to drill down narrow the scope and be nimble and be able to deliver those insights that are super impactful for people who are, as you said, looking to purchase an entire business or sell a business or anything like that. Really big picture stuff but really trying to focus I think is really important and I love that you bought it up. So when it comes to actually working with your clients, I'm curious if you have a specific approach to helping them find insights with analytics? How do you really get started with the mindset of this is what we're going to come in and do for you?
STEVE: 07:44	Yeah. I think the reason we ended up doing that and spending so much time upfront on aligning on what we're trying to do is have felt the pain-- again, I describe analytics as a double-edged sword because it has so many use cases and there's so many options, and so without that clarity, I think [inaudible] it introduces risk of spending time where we revisit in a few weeks' time or through the project and not meeting the overall objective because there's lots of good ideas and lots of good use cases and it's almost hard to say no to some of those. So that having that clear plan just keeps everyone on track.
MADDIE: 08:34	For sure. And let's talk about some of the maybe most forgotten about or underestimated steps or concepts in analytics. Is there any low-hanging fruit that you think analysts should be keeping more top of mind?
STEVE: 08:51	Low-hanging fruit, I would say that's a phrase I use a lot, I think. The low-hanging fruit in my mind and from what we see is focusing on, as I've mentioned, sort of really powerful descriptive analyses that can be explained in a really simple way to business users that can drive a different outcome with that analyses. I think there's this inherent trade-off that we always acknowledge between precision and simplicity. And for those that get excited about analytics and the world of possibilities like I do sometimes or a lot of the time the ability to add precision, continually add precision, is exciting but appreciate that often the developer or the builder of the analysis is not the intended user and so stepping back and being in the shoes of the intended user and thinking about, "Hey. If I was the business user and I was using this new tool or I was trying to leverage this new analyses, do I understand what's it doing? Do I understand how it's built up? Can I act on it with conviction? Can I communicate to others what it does and convince them to make a different decision because of it?" I think the role of simplicity sometimes can easily get overlooked when thinking about analytics applications. So I think simplicity can often get underestimated particularly for those-- again, my hands up here. I'm one of them that gets excited about the world of possibilities with analytics.
MADDIE: 10:31	So, yeah. And something that you mentioned earlier about how spending time upfront is super important for making sure that you don't waste any time. And that kind of reminded me of sometimes where if you get into a debate, or you're in an interview and you get on a tangent, and eventually you totally forget what your main argument was or what the question was. How do analysts make sure that they don't get on these tangents and how often should they check-in with themselves to make sure that they're still asking the right questions?
STEVE: 11:07	Yeah. Look, good question. I think it's definitely not a one-size-fits-all. But as a part of our sort of planning, we kind of try and introduce at least a number of steps in there where we get to step back and get out of the weeds and think about, "Okay. Are we still on track here?" Thinking about those steps are, I think, step one is, do we actually have a clear hypotheses or game plan that we're shooting at? And that game plan should include checkpoints around if things are going on track, X, Y, and Z should happen at these stages. So step one is actually having the plan and having all the stakeholders aligned to that plan. I think the next piece is really getting comfortable with data quality. I think where I've seen the most amount of time being wasted is building a great analyses but not having a great dataset that fits into it. So the data integrity checks upfront, can I trust it? And often it's okay if the datasets not perfect. The challenge I always sort of give to the team is, "It's okay that it's not perfect. How can we fix it?" There are ways to work around it. And even if that means that we don't end up in a-- with a perfect dataset as long as we can articulate the limitations and impact of those limitations to our client, we still often end up in a better place than they would have been without the benefit with our work. So identify the data weaknesses. Work around them where we can. Where you can't, document and communicate those and sort of make sure the end-user understands the limitations.
STEVE: 12:48	Next step is, okay. We now know what data we have and the limitations. Let's re-evaluate what we thought we were going to build now having the benefit of knowing what data we have. And again, re-engage with stakeholders and business users before you go and build to make sure that those limitations would be acceptable and get their input around them. And then once we go and do the build I think there's a couple of interim steps to make sure the stakeholders have some input as we're building out the outputs or building out the analyses. Keep it simple. Don't add too much flexibility and functionality upfront. Do the bare minimum. Get that in front of them. Make sure that that aligns to what they were expecting. Get their feedback and then document your assumptions or limitations that they can use going forward. I think at that point usually a lot of our work would be in the hands of business users and sometimes if we step back, we've probably had three or four touchpoints to make sure we're not going too far-off course. But in the instance where the output or deliverable is something that is maintained on an ongoing basis, I think including in the plan ongoing checkpoints to make sure that you're not only testing new data as it's going through but also getting periodic user feedback is a good part of the process.
MADDIE: 14:07	So you mentioned your focus on M&A but you also mentioned earlier during your intro that you did work for New York Cares. And I'm curious if you could tell us more about that campaign that you did for them and then also maybe tell us a little bit about that campaign through the lens of that strategy that you just laid out for me.
STEVE: 14:31	Yeah. So maybe a little intro on New York Cares and the Volunteer Coordination Taskforce. The New York Cares and the task force coordinate thousands of volunteers in New York in times of emergency. The taskforce was created during Hurricane Sandy. And they provide critical services, like meal delivery, to vulnerable residents of New York. So the taskforce had thousands of volunteers who are out doing sort of boots on the groundwork to help the community. So how we got involved, we were sitting here in New York looking around and seeing the sort of challenging times our community was sort of facing back in March/April and wanted to find a way to give back. So we started to reach out to not-for-profits and through or network to figure out how we might be able to use our skillset to give back to the community. So ended up getting in contact with New York Cares. And then as we sort of stepped through the strategy of the project, first step was how to workshop with them and talk through their overall organisational goals and what they're doing on a day-to-day basis. Talk through how COVID was impacting their operations. And could clearly see changing demand patterns based on what was happening with COVID. So demand on food security and the impact of isolation amongst the community was having a drastic impact on the way they went about their work and the demands of the community. Equally, they had a great network of volunteers who were wanting to do great work.
STEVE: 16:05	So we stepped back. Heard what their business problem was. The team and I sort of went back and prepared our themes document as we usually do. Had a think about the problems that we thought we could solve and quickly sort of identified that a geographic analysis based on the types of projects that they were undertaking was going to be helpful for them. They actually had a fairly rich dataset, but it was in a large, cumbersome file and it was hard to use. So they had a rich dataset. It didn't necessarily talk to what was going on with COVID. And we also understood their business was neighbourhood-based. So they kind of had a neighbourhood-by-neighbourhood lens. So we prepared our themes document. Had a couple of options around areas we could help prioritise those. The first being this geographic analysis. And the questions we were looking to answer there were, where are the areas that require incremental or extra service over this time and how was COVID impacting demand by each zip code? So that they could help effectively re-allocate their resources and volunteers around those changing demand patterns. So talked them through the themes and the analysis that we could do and where we thought our time was well spent. Got their input. Tweaked a few things. Aligned on the analysis we were trying to build. We all agreed that this geographic analysis was highest and best use.
STEVE: 17:37	We went away. Built out a pilot using their data. There was a bunch of publicly available information on COVID cases in New York by zip code that we were able to leverage as well. Develop a pilot. Go through the process of sort of testing it with them on how they would use it on an ongoing basis. And then within a couple of weeks, they have a web-hosted dashboard that their teams able to use to help identify areas where they need to reallocate their resources. And then what we're doing now is on a periodic basis updating that analysis to make sure that they can continually make adjustments around resource allocation. I think where we were able to-- where this tool really had a big impact for them was to be able to identify specific zip codes and specific programs in those zip codes that needed some more attention. And given the simplicity of the tool, anyone in the organisation can use it. So it's not like there was a significant amount of training involved. Now, it's a tool they can use on an ongoing basis to answer those common questions around, "Okay. We've got a lot of-- we've got a lot of resources. We've got a lot of volunteers. Where do we re-allocate them to on an ongoing basis?"
MADDIE: 18:56	Yeah. You mentioned pulling in some third-party data. What questions do you ask yourself when you're choosing the right third-party data sources to enrich that analysis?
STEVE: 19:10	I ask myself all the questions I know my clients are going to ask me. I've been asked enough now to have a list I can almost rattle off. It's what is the incremental insight that this dataset will provide and what will I be able to do with that incremental insight versus without it? Helps me work out, does it matter? Do I care or is it just nice to have? Does it look pretty or is it actually useful and how useful? Next question is, how complicated is it going to be to bring in? This goes back to the simplicity versus precision problem. If it's going to mean the level of effort to develop this project is times 2 or times 3 but it's only a 20% increment on the insight, it might not be worth doing that right now. So what's the impact? How hard is it to bring in? And then once I'm comfortable that it's a worthwhile effort, some pressure tests on the quality of the data. So what is the source? Where does it actually come from? Because we often find with third-party datasets, it can be assumed that it's right, but its true source can be quite varied. So really understanding where the underlying source came from. How reliable it is. What can I cross-reference it with? How can I make sure it's right or I can get a sense of how wrong it might be? How much does it cost or the level of effort to acquire it and how regularly it's updated? They're the common questions we go through all the time and they're the ones my clients ask me, so I try and be prepared as much as I can.
MADDIE: 20:45	That's so cool. Yeah. I mean, it sounds like you're really making sure that that is the right resource. I think that it is-- it can be easy to find some sort of data online or just datasets that are widely available, especially when it comes to public issues like this. But yeah, making sure that it's coming from the right place, how often is it updated. Those are some of the questions that might not be as well-known or as-- it might not be as natural to ask yourself those questions. So thank you for sharing those.
STEVE: 21:19	No problem.
MADDIE: 21:19	And then you also mentioned that it was really important to you to provide New York Cares with a dashboard that would be really easy to read and be really easy to consume. So when it comes to packaging insights for New York Cares or your clients or really anybody, what are some tips that you have for packaging those insights so that they can be easily consumed and acted upon?
STEVE: 21:49	Simple stuff. I would say put yourself in the other person's shoes. I think having the benefit of building this analyses I know what it's like to get excited about what's possible. And again, this whole simplicity versus precision. But make it simple to start. You can always add precision and add new views as time goes on. And so I would say keep it really simple as a starting point. I would say don't overthink the visualisation. And I'll admit, I'm biased here because I grew up in Excel spreadsheets and financial models. But I always find the benefit of being grounded in the numbers first or the data first because if you have to prove it to someone, if you have to get someone comfortable that the analysis makes sense, at some point, they're going to want to see the underlying data. So you're better off getting yourself comfortable in a simple format first and then moving to figuring out how to tell that story in a simple visualisation. I think other simple things, really simple things, talk slowly and clearly when you're communicating these things. They're really complicated problems. It's hard to grasp all of the concepts that you're talking about. Use simple diagrams and examples. I try and talk about simple examples, "If I bought a can of Spite every month for the last three months and in this latest month, I bought two cans of Sprite," using those simple examples like that help people conceptualise or understand the challenging concepts. And then, I don't know. I think yeah appreciating that simplicity.
STEVE: 23:27	I had a think about this when I'm sitting here, and I've got a canvas in my room that my last team bought me that's got a bunch of quotes on it that in hindsight, I probably ask all the time. And it's a bit of a pattern, but I think it helps package things in a good way. The question - I'm reading them verbatim here - is, do the numbers reconcile? Clients will ask. If you can't get comfortable with the numbers in the first place, you're not going to go and further than that. So do the numbers reconcile? Give me the analysis in a table, so you can prove it to someone that it works. So the next quote I've got here is, what would you do with the benefit of the analysis versus what you would do without it? So helps pressure test. Is it going to create any incremental insight from what the business user already has? That's a good pressure point, pressure test, I should say, to make sure that we're focusing in the right places. And then the last one I always challenge my guys is, if you had the benefit of this analysis, if you had your last $100, or you had to make a decision on what you would do here, what's the decision you would make? And go into the conversation with your client with that mindset because that's the decision that they've got. So you should go in helping them use the tool or use the analysis to answer the question you know they've got. Go all the way to the end. What would you do if you were them with the benefit of this analysis? I think that gets me comfortable that we're providing incremental insight and value if we can get through those questions.
MADDIE: 25:06	Definitely. And that just brings such a smile to my face thinking of there being a poster, or as you said, a canvas. Which, by the way, is that Australia speak for a piece of artwork?
STEVE: 25:18	It is. Sorry. [laughter] It is. Yeah, a piece of artwork that's got a bunch of quotes on it. Sorry.
MADDIE: 25:26	No. That's great. That's really cool. Yeah. No, it makes me smile thinking of that hanging in your home or your office or wherever you're at. And I wonder how many listeners out there, how many other people of the community have something like that too in their home. That's really cool. So last question that I have for you. It's so great to hear about the time that you took to help New York Cares and create this dashboard for them so that they could utilise their data in a more actionable way to help more people. And I'm curious what advice you would give to other analysts and data scientists looking to help other non-profits like New York Cares?
STEVE: 26:06	Do it. Do it. Look, we didn't where to start and it felt at times, I'll admit, didn't know who to reach out to or how we might help, but it's been one of the most rewarding things we've done. And what I'll say is, you'll likely have a really big impact and you can do that with a relatively small commitment. And for us, we're all in now. We're going to be working with the New York Cares guys going forward. Really excited about what we've been able to do and how we can help them going forward. And I would say, find something that you're passionate about and reach out to them directly. My guess is that you'd be welcomed with open arms. Not only great people but they appreciate the impact that you can have. And so yeah, don't be scared. Reach out and just get involved. It's a really good feeling.
MADDIE: 26:58	Great. Yeah. That's super inspiring. I hope that a lot of our listeners out there or people who have analytics expertise will seek out these opportunities, especially when it's the holidays and people are looking for chances to give back like this. It's perfect timing.
STEVE: 27:13	Yeah. I totally agree. It's one of the highlights for our year, that's for sure.
MADDIE: 27:20	Thanks for listening. What are some of the go-to questions you ask yourself to make sure you're still on track with your analysis? Share it with us by leaving a comment on the community at community.alteryx.com/podcast or share it on social media using the #AlterEverythingPodcast. And don't forget to subscribe to Alter Everything on your favourite podcast app and share with a buddy who you know is deep in the analytics game. Catch you next time.

MADDIE: 00:00 Welcome to Alter Everything, a podcast about data science and analytics culture. I'm Maddie Johannsen and today I'm joined by Steve Mann, Managing Director of Propel32. STEVE: 00:11 I'm a proud resident of New York City and a native of Perth, West Australia. Hope my family don't hear me say it in that order. MADDIE: 00:20 Steve has a background in finance, has lead teams at PwC, and is currently focused on using analytics and M&A, mergers and acquisitions. As Steve is in New York, he's had a very close look at the devastation that COVID-19 has had on the city and decided to give back. Propel32 partnered with NYC Service and New York Cares, the largest volunteer network in New York City and used Alteryx to provide actionable insight into their data at a time of true crisis. So throughout our conversation, you'll hear Steve talk about the importance of analytics basics and why simplicity should never be underestimated. Let's get started. I'm interested to hear, then how it became kind of happenstance for you to move to New York? STEVE: 01:10 So happenstance for me-- and Alteryx actually has a part to play here. So I spent my whole career up until launching Propel32 at PwC. I was working up in Toronto. I was up in secondment. And I was doing my normal sort of M&A related work and started playing around with analytics tools. Where I was looking at my career and thinking, "Well, I love doing this work. I love analysing businesses and helping clients in an M&A environment, but if I've got to look at these data problems for the next 40 years, I'm going to be pretty unhappy." So started to play around with analytics tools that would help me get around those problems and ended up spending more and more time solving commercial problems with these analytics tools and sort of pivoting from the traditional work that I was doing. Ended up getting a tap on the shoulder from the firm to see if I was interested in coming down to New York to build a team focused on analytics in M&A and while again it wasn't on the agenda before that the appeal of coming down to New York from a professional perspective as well as a personal perspective, it was too hard to say no to. So bit the bullet. Moved down to New York two and a half years ago now and really feel ingrained or starting to become ingrained in New York and haven't looked back. So I'm really grateful for the opportunity to come and do that. It's opened up a world of opportunities, both professionally and personally as well. MADDIE: 02:46 So yeah, Steve, with your years of experience leading teams at PwC and now as a Managing Director at Propel32, I'm sure you've seen analytics change over your years, and I'm wondering if you can share with us how you've seen that change happen or what the change has been? STEVE: 03:06 Yeah. Certainly. And I would say so in my field, I'm really focused on analytics in M&A. So a lot of my clients are either private equity clients or corporate clients who are looking to buy or sell a business, and they're looking to understand as much as they can about that business in a very short space of time. If they're selling a business, demonstrate to a potential buyer where the growth opportunities exist and get those buyers really comfortable that those opportunities are real with a really strong analyses. With that in mind, the concept of analytics is one that has always been of interest. I think in the last four or five years its gone from something that's interesting to something that people appreciate can actually provide a competitive advantage in a deal setting. So the pressure to get it right or the pressure to get the ball rolling has definitely picked up. I think many people in this space, or many clients in this space, are trying to work out the best use cases. I think a lot of us in analytics can appreciate that analytics can mean a lot of different things and I think the challenge is finding the best use cases that have the greatest impact in a deal setting. And look analytics M&A is tough. It's really short timelines. You have very limited access to management. Often, the data or existing infrastructure isn't set up to analyse the business in the way that the business is going to be analysed for the purpose of a transaction. The stakes are really high. Someone's going to buy or sell the whole business. So it's a high-pressure environment. STEVE: 04:53 And equally, there's a need for simplicity because a lot of people need to understand what's going on so really complicated solutions aren't going to work. So it's a tough environment. And because of that, where I've seen the highest and best use cases or highest and best use of analytics is really spending a lot of time on developing a really powerful descriptive analysis that's sitting on top of an enhanced dataset that's had a lot of scrubbing that has been pulled together from a number of different data sources and that has had simple things fixed. I'm sure many of us have felt the pain of a same customer name that's entered four different times. Simple things like that, that can cleaned at speed so that we have an enhanced dataset to drive a powerful business analysis that can be delivered within weeks. They're the highest and best uses in an M&A environment from what I've seen. And I think the market is starting to dip their toe in and try different things. And I think as we go along in this journey, we will continue to evolve to more sophisticated use cases. But where I see the market right now is a desire to find those use cases and those who have figured it out are able to create a competitive advantage by finding new insights that otherwise would have gone unseen in a very short period of time. MADDIE: 06:33 Yeah. And a couple of things that really struck me as you were talking, I love that you mentioned kind of the-- you kind of sort of alluded to the chaos that can be out there when it comes to applications or platforms or trying to pick the best data strategy or how do you even get started. And I think that that can definitely be overwhelming. But as you mentioned one of your focus areas is just really trying to drill down narrow the scope and be nimble and be able to deliver those insights that are super impactful for people who are, as you said, looking to purchase an entire business or sell a business or anything like that. Really big picture stuff but really trying to focus I think is really important and I love that you bought it up. So when it comes to actually working with your clients, I'm curious if you have a specific approach to helping them find insights with analytics? How do you really get started with the mindset of this is what we're going to come in and do for you? STEVE: 07:44 Yeah. I think the reason we ended up doing that and spending so much time upfront on aligning on what we're trying to do is have felt the pain-- again, I describe analytics as a double-edged sword because it has so many use cases and there's so many options, and so without that clarity, I think [inaudible] it introduces risk of spending time where we revisit in a few weeks' time or through the project and not meeting the overall objective because there's lots of good ideas and lots of good use cases and it's almost hard to say no to some of those. So that having that clear plan just keeps everyone on track. MADDIE: 08:34 For sure. And let's talk about some of the maybe most forgotten about or underestimated steps or concepts in analytics. Is there any low-hanging fruit that you think analysts should be keeping more top of mind? STEVE: 08:51 Low-hanging fruit, I would say that's a phrase I use a lot, I think. The low-hanging fruit in my mind and from what we see is focusing on, as I've mentioned, sort of really powerful descriptive analyses that can be explained in a really simple way to business users that can drive a different outcome with that analyses. I think there's this inherent trade-off that we always acknowledge between precision and simplicity. And for those that get excited about analytics and the world of possibilities like I do sometimes or a lot of the time the ability to add precision, continually add precision, is exciting but appreciate that often the developer or the builder of the analysis is not the intended user and so stepping back and being in the shoes of the intended user and thinking about, "Hey. If I was the business user and I was using this new tool or I was trying to leverage this new analyses, do I understand what's it doing? Do I understand how it's built up? Can I act on it with conviction? Can I communicate to others what it does and convince them to make a different decision because of it?" I think the role of simplicity sometimes can easily get overlooked when thinking about analytics applications. So I think simplicity can often get underestimated particularly for those-- again, my hands up here. I'm one of them that gets excited about the world of possibilities with analytics. MADDIE: 10:31 So, yeah. And something that you mentioned earlier about how spending time upfront is super important for making sure that you don't waste any time. And that kind of reminded me of sometimes where if you get into a debate, or you're in an interview and you get on a tangent, and eventually you totally forget what your main argument was or what the question was. How do analysts make sure that they don't get on these tangents and how often should they check-in with themselves to make sure that they're still asking the right questions? STEVE: 11:07 Yeah. Look, good question. I think it's definitely not a one-size-fits-all. But as a part of our sort of planning, we kind of try and introduce at least a number of steps in there where we get to step back and get out of the weeds and think about, "Okay. Are we still on track here?" Thinking about those steps are, I think, step one is, do we actually have a clear hypotheses or game plan that we're shooting at? And that game plan should include checkpoints around if things are going on track, X, Y, and Z should happen at these stages. So step one is actually having the plan and having all the stakeholders aligned to that plan. I think the next piece is really getting comfortable with data quality. I think where I've seen the most amount of time being wasted is building a great analyses but not having a great dataset that fits into it. So the data integrity checks upfront, can I trust it? And often it's okay if the datasets not perfect. The challenge I always sort of give to the team is, "It's okay that it's not perfect. How can we fix it?" There are ways to work around it. And even if that means that we don't end up in a-- with a perfect dataset as long as we can articulate the limitations and impact of those limitations to our client, we still often end up in a better place than they would have been without the benefit with our work. So identify the data weaknesses. Work around them where we can. Where you can't, document and communicate those and sort of make sure the end-user understands the limitations. STEVE: 12:48 Next step is, okay. We now know what data we have and the limitations. Let's re-evaluate what we thought we were going to build now having the benefit of knowing what data we have. And again, re-engage with stakeholders and business users before you go and build to make sure that those limitations would be acceptable and get their input around them. And then once we go and do the build I think there's a couple of interim steps to make sure the stakeholders have some input as we're building out the outputs or building out the analyses. Keep it simple. Don't add too much flexibility and functionality upfront. Do the bare minimum. Get that in front of them. Make sure that that aligns to what they were expecting. Get their feedback and then document your assumptions or limitations that they can use going forward. I think at that point usually a lot of our work would be in the hands of business users and sometimes if we step back, we've probably had three or four touchpoints to make sure we're not going too far-off course. But in the instance where the output or deliverable is something that is maintained on an ongoing basis, I think including in the plan ongoing checkpoints to make sure that you're not only testing new data as it's going through but also getting periodic user feedback is a good part of the process. MADDIE: 14:07 So you mentioned your focus on M&A but you also mentioned earlier during your intro that you did work for New York Cares. And I'm curious if you could tell us more about that campaign that you did for them and then also maybe tell us a little bit about that campaign through the lens of that strategy that you just laid out for me. STEVE: 14:31 Yeah. So maybe a little intro on New York Cares and the Volunteer Coordination Taskforce. The New York Cares and the task force coordinate thousands of volunteers in New York in times of emergency. The taskforce was created during Hurricane Sandy. And they provide critical services, like meal delivery, to vulnerable residents of New York. So the taskforce had thousands of volunteers who are out doing sort of boots on the groundwork to help the community. So how we got involved, we were sitting here in New York looking around and seeing the sort of challenging times our community was sort of facing back in March/April and wanted to find a way to give back. So we started to reach out to not-for-profits and through or network to figure out how we might be able to use our skillset to give back to the community. So ended up getting in contact with New York Cares. And then as we sort of stepped through the strategy of the project, first step was how to workshop with them and talk through their overall organisational goals and what they're doing on a day-to-day basis. Talk through how COVID was impacting their operations. And could clearly see changing demand patterns based on what was happening with COVID. So demand on food security and the impact of isolation amongst the community was having a drastic impact on the way they went about their work and the demands of the community. Equally, they had a great network of volunteers who were wanting to do great work. STEVE: 16:05 So we stepped back. Heard what their business problem was. The team and I sort of went back and prepared our themes document as we usually do. Had a think about the problems that we thought we could solve and quickly sort of identified that a geographic analysis based on the types of projects that they were undertaking was going to be helpful for them. They actually had a fairly rich dataset, but it was in a large, cumbersome file and it was hard to use. So they had a rich dataset. It didn't necessarily talk to what was going on with COVID. And we also understood their business was neighbourhood-based. So they kind of had a neighbourhood-by-neighbourhood lens. So we prepared our themes document. Had a couple of options around areas we could help prioritise those. The first being this geographic analysis. And the questions we were looking to answer there were, where are the areas that require incremental or extra service over this time and how was COVID impacting demand by each zip code? So that they could help effectively re-allocate their resources and volunteers around those changing demand patterns. So talked them through the themes and the analysis that we could do and where we thought our time was well spent. Got their input. Tweaked a few things. Aligned on the analysis we were trying to build. We all agreed that this geographic analysis was highest and best use. STEVE: 17:37 We went away. Built out a pilot using their data. There was a bunch of publicly available information on COVID cases in New York by zip code that we were able to leverage as well. Develop a pilot. Go through the process of sort of testing it with them on how they would use it on an ongoing basis. And then within a couple of weeks, they have a web-hosted dashboard that their teams able to use to help identify areas where they need to reallocate their resources. And then what we're doing now is on a periodic basis updating that analysis to make sure that they can continually make adjustments around resource allocation. I think where we were able to-- where this tool really had a big impact for them was to be able to identify specific zip codes and specific programs in those zip codes that needed some more attention. And given the simplicity of the tool, anyone in the organisation can use it. So it's not like there was a significant amount of training involved. Now, it's a tool they can use on an ongoing basis to answer those common questions around, "Okay. We've got a lot of-- we've got a lot of resources. We've got a lot of volunteers. Where do we re-allocate them to on an ongoing basis?" MADDIE: 18:56 Yeah. You mentioned pulling in some third-party data. What questions do you ask yourself when you're choosing the right third-party data sources to enrich that analysis? STEVE: 19:10 I ask myself all the questions I know my clients are going to ask me. I've been asked enough now to have a list I can almost rattle off. It's what is the incremental insight that this dataset will provide and what will I be able to do with that incremental insight versus without it? Helps me work out, does it matter? Do I care or is it just nice to have? Does it look pretty or is it actually useful and how useful? Next question is, how complicated is it going to be to bring in? This goes back to the simplicity versus precision problem. If it's going to mean the level of effort to develop this project is times 2 or times 3 but it's only a 20% increment on the insight, it might not be worth doing that right now. So what's the impact? How hard is it to bring in? And then once I'm comfortable that it's a worthwhile effort, some pressure tests on the quality of the data. So what is the source? Where does it actually come from? Because we often find with third-party datasets, it can be assumed that it's right, but its true source can be quite varied. So really understanding where the underlying source came from. How reliable it is. What can I cross-reference it with? How can I make sure it's right or I can get a sense of how wrong it might be? How much does it cost or the level of effort to acquire it and how regularly it's updated? They're the common questions we go through all the time and they're the ones my clients ask me, so I try and be prepared as much as I can. MADDIE: 20:45 That's so cool. Yeah. I mean, it sounds like you're really making sure that that is the right resource. I think that it is-- it can be easy to find some sort of data online or just datasets that are widely available, especially when it comes to public issues like this. But yeah, making sure that it's coming from the right place, how often is it updated. Those are some of the questions that might not be as well-known or as-- it might not be as natural to ask yourself those questions. So thank you for sharing those. STEVE: 21:19 No problem. MADDIE: 21:19 And then you also mentioned that it was really important to you to provide New York Cares with a dashboard that would be really easy to read and be really easy to consume. So when it comes to packaging insights for New York Cares or your clients or really anybody, what are some tips that you have for packaging those insights so that they can be easily consumed and acted upon? STEVE: 21:49 Simple stuff. I would say put yourself in the other person's shoes. I think having the benefit of building this analyses I know what it's like to get excited about what's possible. And again, this whole simplicity versus precision. But make it simple to start. You can always add precision and add new views as time goes on. And so I would say keep it really simple as a starting point. I would say don't overthink the visualisation. And I'll admit, I'm biased here because I grew up in Excel spreadsheets and financial models. But I always find the benefit of being grounded in the numbers first or the data first because if you have to prove it to someone, if you have to get someone comfortable that the analysis makes sense, at some point, they're going to want to see the underlying data. So you're better off getting yourself comfortable in a simple format first and then moving to figuring out how to tell that story in a simple visualisation. I think other simple things, really simple things, talk slowly and clearly when you're communicating these things. They're really complicated problems. It's hard to grasp all of the concepts that you're talking about. Use simple diagrams and examples. I try and talk about simple examples, "If I bought a can of Spite every month for the last three months and in this latest month, I bought two cans of Sprite," using those simple examples like that help people conceptualise or understand the challenging concepts. And then, I don't know. I think yeah appreciating that simplicity. STEVE: 23:27 I had a think about this when I'm sitting here, and I've got a canvas in my room that my last team bought me that's got a bunch of quotes on it that in hindsight, I probably ask all the time. And it's a bit of a pattern, but I think it helps package things in a good way. The question - I'm reading them verbatim here - is, do the numbers reconcile? Clients will ask. If you can't get comfortable with the numbers in the first place, you're not going to go and further than that. So do the numbers reconcile? Give me the analysis in a table, so you can prove it to someone that it works. So the next quote I've got here is, what would you do with the benefit of the analysis versus what you would do without it? So helps pressure test. Is it going to create any incremental insight from what the business user already has? That's a good pressure point, pressure test, I should say, to make sure that we're focusing in the right places. And then the last one I always challenge my guys is, if you had the benefit of this analysis, if you had your last $100, or you had to make a decision on what you would do here, what's the decision you would make? And go into the conversation with your client with that mindset because that's the decision that they've got. So you should go in helping them use the tool or use the analysis to answer the question you know they've got. Go all the way to the end. What would you do if you were them with the benefit of this analysis? I think that gets me comfortable that we're providing incremental insight and value if we can get through those questions. MADDIE: 25:06 Definitely. And that just brings such a smile to my face thinking of there being a poster, or as you said, a canvas. Which, by the way, is that Australia speak for a piece of artwork? STEVE: 25:18 It is. Sorry. [laughter] It is. Yeah, a piece of artwork that's got a bunch of quotes on it. Sorry. MADDIE: 25:26 No. That's great. That's really cool. Yeah. No, it makes me smile thinking of that hanging in your home or your office or wherever you're at. And I wonder how many listeners out there, how many other people of the community have something like that too in their home. That's really cool. So last question that I have for you. It's so great to hear about the time that you took to help New York Cares and create this dashboard for them so that they could utilise their data in a more actionable way to help more people. And I'm curious what advice you would give to other analysts and data scientists looking to help other non-profits like New York Cares? STEVE: 26:06 Do it. Do it. Look, we didn't where to start and it felt at times, I'll admit, didn't know who to reach out to or how we might help, but it's been one of the most rewarding things we've done. And what I'll say is, you'll likely have a really big impact and you can do that with a relatively small commitment. And for us, we're all in now. We're going to be working with the New York Cares guys going forward. Really excited about what we've been able to do and how we can help them going forward. And I would say, find something that you're passionate about and reach out to them directly. My guess is that you'd be welcomed with open arms. Not only great people but they appreciate the impact that you can have. And so yeah, don't be scared. Reach out and just get involved. It's a really good feeling. MADDIE: 26:58 Great. Yeah. That's super inspiring. I hope that a lot of our listeners out there or people who have analytics expertise will seek out these opportunities, especially when it's the holidays and people are looking for chances to give back like this. It's perfect timing. STEVE: 27:13 Yeah. I totally agree. It's one of the highlights for our year, that's for sure. MADDIE: 27:20 Thanks for listening. What are some of the go-to questions you ask yourself to make sure you're still on track with your analysis? Share it with us by leaving a comment on the community at community.alteryx.com/podcast or share it on social media using the #AlterEverythingPodcast. And don't forget to subscribe to Alter Everything on your favourite podcast app and share with a buddy who you know is deep in the analytics game. Catch you next time.

This episode of Alter Everything was produced by Maddie Johannsen (@MaddieJ).
Special thanks to @andyuttley for the theme music track, and @jeho for our album artwork.

Alter Everything

Episode Guide

76: Analytics simplicity: A guide for staying out of an analytics black hole

Panelists

Topics

Transcript

This episode of Alter Everything was produced by Maddie Johannsen (@MaddieJ).
Special thanks to @andyuttley for the theme music track, and @jeho for our album artwork.

Alter Everything

Episode Guide

76: Analytics simplicity: A guide for staying out of an analytics black hole

Panelists

Topics

Transcript

This episode of Alter Everything was produced by Maddie Johannsen (@MaddieJ).Special thanks to @andyuttley for the theme music track, and @jeho for our album artwork.

This episode of Alter Everything was produced by Maddie Johannsen (@MaddieJ).
Special thanks to @andyuttley for the theme music track, and @jeho for our album artwork.