Worldwide spend on cloud infrastructure hit a staggering $169 billion in 2022 with Microsoft and Amazon Web Services (AWS) combined, accounting for 62 percent of the cloud market share. In its first quarter of this year revenue from AWS totaled $21.35 billion, representing almost 17% of Amazon’s overall revenue. What lessons can we learn from within AWS about effective cloud financial management? How can FP&A teams better understand this complex line item? We are joined by Jeff Duresky, Senior FinOps Commercial Architect, Amazon AWS. He is a finance leader who supports the cloud leader’s largest and most strategic customers – and has also held strategic FP&A roles at companies including Capital One.
“As a finance professional the more you understand cloud, the better the insights you’ll be able to drive” – Jeff Duresky
In this episode
- Jeff’s worst budget experience (aka professional service accrual releases).
- The link between cloud financial management and FP&A
- trading fixed expenses in cloud
- Smart decisions related to cloud expenditures and management
- Common cost language – how to get it right
- Secrets to cloud forecasting
- Promoting the best finance and engineering team conversations around the cloud
- Finding reductions in costs for cloud users
- Core metrics to assess and unit costs (cost per gigabyte, cost per instance)
- First steps to begin forecasting cloud costs
- My biggest finance strategic moment
- The last finance thing I googled
- The importance of taking notes in a career in FP&A
Notes
How to establish and drive a forecasting culture
How to talk about cloud with a non-cloud audience (business partnering)
Follow and connect with Jeff Duresky on LinkedIn https://www.linkedin.com/in/jeffduresky/
FP&A Today is brought to you by Datarails, Datarails is the AI-powered financial planning and analysis platform. Keep your own Excel financial models and spreadsheets and benefit from AI for data consolidation, reporting and planning.
Paul Barnhurst:
Hello everyone. Welcome to FP&A Today, I am your host, Paul Barnhurst, aka the FP&A Guy. FP&A Today is brought to you by Datarails, the financial planning and analysis platform for Excel users. Every week we welcome a leader from the world of financial planning and analysis. Today we are delighted to be joined by Jeff Duresky. Jeff, welcome to the show.
Jeff Duresky:
Thanks, Paul. I’m extremely happy to be here today.
Paul Barnhurst:
Well, we’re really excited to have you. And I like the Amazon shirt, if anyone wants to guess where he works. You, you, you got it right. He’s supporting Amazon Cloud Financial Management during his career. He is worked at several different companies in multiple finance roles, obviously, including FP&A and Cloud Finance management. He did his undergrad from Virginia Tech and then he earned his MBA from the University of North Carolina. He has a strong finance and ITslash kind of data analytics background. And so we’re really excited to talk to him about cloud finance and several other subjects today. But where we’re gonna start is just giving you a minute to tell us about your background. If you could take a moment and tell us a little bit about yourself.
Jeff Duresky:
Yeah. So, you know, as Paul said, I did my undergrad at Virginia Tech in IT supply chain management, operations management. So I started off in procurement roles that rolled into business analyst FP&A, moved into more pure finance roles at a few companies supported regulatory supported internal audit. And then finally the, where I was prior to AWS was at Capital One supporting their data center exit business case. So Capital One’s been very public about moving from data centers onto AWS public cloud. And I supported the business case and then moved into cloud finance. And now I’m with AWS on our optics team. And we’re a team that supports some of AWS’s largest, most strategic customers for cloud efficiency, for optimization of their cost and the culture of cloud financial management or finops.
Paul Barnhurst:
Great, thanks for that introduction. So this is a question we like to start off with everybody ’cause anyone who’s done FP&A has that horror story around budgeting. So what’s maybe the worst or most challenging budget experience you ever had?
Jeff Duresky:
Sure. So I’ll, I’ll say a term that will probably bring chills to a lot of FP&A professionals, but professional service accrual releases. So I inherited a budget during a reorg and just had to keep the budget flat. I didn’t have any, any context behind what had been loaded prior. We had these large professional services accruals the first month or so trying to understand variances to that and just kept on taking the variance, plugging it to the last, you know, couple months of the year. ‘Because I knew I needed to understand what was going on with these accruals. Began digging in, realizing how many of them had contract end dates that were six months old, eight months a year old or more, or they had been fully invoiced and we still had the accruals. So you have to, to dig in, start signaling to leadership.
We may have some accrual releases on this, but we need to make sure that we are a hundred percent correct reaching out to the vendors. Have we been fully invoiced clean all of that up and then had to make sure that when we loaded the budget the next time we timed the release accruals with our outlook. So we didn’t have material variances because of timing. And so you, you realize just the importance of having a tracking mechanism on some of these very broad professional services. And then the ownership, who, who owned that? Was it the project manager? Was it the PO? Should we have gone on PO date the accounting teams? So that was a, was certainly a challenging project.
Paul Barnhurst:
Yeah, I can completely relate. As you were telling that, I was thinking about a role I came into where they had just stood up a shared service center and they were operationalizing a company that had been a portfolio company, a holding company. And so they brought in something like 20 different organizations all to one. And there were all kinds of fun stuff with the balance sheet and things not being accrued and professional services. And it was just a nightmare the first few months. And so I know what you’re talking about. Okay, do we have stuff accrued, right? Do we have, you know, do we have enough? Do we have too much? Is there a bad guy? Fortunately for, unfortunately for us, a lot of ’em were bad guys. <Laugh> not good buys, but it was definitely a learning experience. So what was your takeaway from that? What’s probably the biggest thing you learned from that experience?
Jeff Duresky:
The, the biggest thing was signaling early that there is something going on. I don’t have enough data right now to necessarily say when this is gonna happen, what the outcome is, but making sure that senior leadership was aware of a work stream that was going on, and then starting to figure out who are the right people we need to get into the room. If it was a broad professional services contract, who are the ones that are gonna be able to say, we are complete with this. The work is done, all invoices have been signed off. And that was something that just didn’t exist in one area of for, from a tracking mechanism.
Paul Barnhurst:
I love how you said signal early, but that leads me to a follow-up question. How do you manage that when you don’t have all the facts but you signal early, right? Because immediately it’s like, well, what’s the risk? How big of a problem is this? Any advice you could offer that you learned on how to manage that process? ’cause I totally agree with you. We need to signal early, but often we don’t have enough information.
And so it is trying to find that right balance of how much information you have versus how early you signal.
Jeff Duresky:
I think it, it comes down to orders of magnitude and materiality. So if you’re a, if you’re a giant company, you have a cost center that you’re working with that is in the tens or hundreds of millions of dollars, is this an issue that is 50,000, 500,000 or 5 million thinking of that, that orders of magnitude mm-hmm. <Affirmative>. And you should be able to get a sense of where you think it may land. So that’s where you start to signal that, you know, it’s not gonna be low. It’s probably not gonna be the 5 million, but half a million to one and a half million. That’s my my quick guess. And as soon as I know more, I’m gonna keep you in the loop.
Paul Barnhurst:
I, I like that. And I think that’s a really good approach is one, materiality and two, do a rough sizing. Yes, you may be wrong, but at least you’ve given ’em an idea versus holding off and fighting it out. It’s much bigger than you originally thought. And they’re like, well, why didn’t you tell me earlier?
Jeff Duresky:
So absolutely no one’s gonna, no one’s gonna hold your feet to the fire of if you were wrong, as you’re trying to signal early to make sure that there’s a situational awareness is gonna be far worse if you wait too long and it is material. And then why didn’t you let us know, at least with that rough sizing,
Paul Barnhurst:
Fully agree that I, I’m a hundred percent aligned with you on that. So next question here. You know, you’re currently focused on cloud financial management. Can you talk to our audience about what that is and how that role ties into FP&A?
Jeff Duresky:
Absolutely. So when you think of the cloud, basically what you’re doing is trading fixed expenses. Either your data centers, your servers for your variable expenses and then paying for what you need. So how it comes into your procurement or your finance is you go from centralized procurement, centralized purchasing, to now decentralize purchasing with potentially millisecond billing. You may have hundreds or thousands of engineers. Each one has the autonomy and the permissions to spin up resources at any point for the project that they’re working on. But you still need to be able to report on the aggregate cost and usage for your cost center, for your organization, however you ring fence your ring fence, your costs. And then also you wanna make sure that you have the sense of ownership that those groups have an ownership of the costs and how that aligns to your general ledger.
So the, the chargeback or showback process of cloud cost, who owns them? Who’s setting the budget, who’s doing the forecasting? And then also, you know, you go one step further from a finance perspective, understanding cloud enough that you can help your engineering, your tech teams for potential optimization. Are they using the right service for the workload or for the, for the business model that they’re working on. And so it encompasses a number of different aspects of not only working with tech, working with your finance, accounting, procurement while also making sure that you’re able to post your issues to the general ledger and then everything reconciles at the end with your monthly invoice.
Paul Barnhurst:
So if I’m hearing you right, it feels like, you know, with cloud finance, obviously you’re managing the cloud expenses, which is that trade for fixed CapEx, you know, to opex that whole piece there and letting someone else manage manage that and being able to spin it up and pay for what you use. But I’m hearing you say it’s also a more technical role than like a traditional FP&A. You gotta be able to at least speak to the engineers and be able to look at things and help them make smart decisions in how they’re managing their cloud resources. Is that fair to say?
Jeff Duresky:
That really gets to the, the culture of cloud finance or the culture of finops within your organization. So if you are in the public cloud, if you’re using cloud, someone is performing all of those activities, someone is doing monthly variance report from a finance perspective, someone on the architecture, the engineering side is choosing the, the service for the workload that they’re working on. The cultural aspect is how much are they talking? And then how well do they understand everybody’s role within the broad umbrella of cloud finance. And so from a, from a finance perspective, it may be a side of the desk job. For r a lot of companies others have dedicated cloud finance teams.
Paul Barnhurst:
Sure,
Jeff Duresky:
However it happens to be the better. As a finance professional you can understand cloud, the better insights you’re gonna be able to drive. And also with your monthly variance analysis, with your forecasting and budgeting, understanding what the engineers are doing, at least from a high level perspective is gonna help you be clearer within your forecast and within your assumptions of that forecast.
Paul Barnhurst:
You know what it is like 13 different spreadsheets emailed out to 23 different budget holders, multiple iterations, version control errors, back and forth updates you never really feel in control of the consolidation and collection process. Yep, I’ve been there. Stop. Breathe. Datarails is the financial planning and analysis platform for Excel users. Datarails takes data from all your company’s disparate sources. No organization is too complex, consolidating everything into one place, secured in the cloud. Now all your data finally talking to each other, everything is automated. Back into your report in Excel. Cashflow FX conversion, intercompany transactions now automated and UpToDate drill down and variance analysis in seconds. Don’t replace Excel, embrace Excel, turn your Excel into a lean mean FP&A machine. Find out more@www.datarails.com.
So I’m curious, what attracted you to cloud financial management, right? That’s pretty specialized area within finance within FP&A. So what, what led you to focus on that area?
Jeff Duresky:
Like a lot of roles that people have you fall into it <laugh> it, it, the cloud really didn’t exist when I graduated undergrad. And so just, you know, cloud finance as a specialty is relatively new. As I mentioned earlier, when I was hired at Capital One to work on the data center, exit business case cloud was one piece of the entire business case as you look to move from data centers into the cloud and you can, you know, read all of the, the public blog posts about that journey and when they eventually close their remaining data centers. So managing that from a very broad perspective. Bandwidth opened up once data centers are closed and they’re all in on the cloud. And at that time, capital One had stood up a cloud finance team dedicated just to that one general ledger line item, cloud costs.
And I just dug more into that I to understand the cost, to have a sense of ownership of how do we support the forecasting and budgeting. And what I really liked was the ability to see the work that you do impacted almost immediately. And, you know, example, because you have this, you know, millisecond billing on some of the, some of the services or second billing, you can change the service and see the cost immediately or there are prepaid discounts AWS savings plans, reserved instances if we’re using AWS terminology where you can prepay for a discount on some of the services, you see the cost impact as soon as that purchases. So if you’re looking at a daily cost, you can see a cost decrease immediately and it’s fantastic to see the actions that you take reflected immediately in the data, which you don’t always get with a, a finance perspective when you’re not looking at daily costs.
Paul Barnhurst:
Sure. I would argue in most places you’re not able to see that immediate impact. You know, restructuring, obviously if you take out a bunch of heads you see it the next month, but those are the type that none of us enjoy. Right. <laugh>, right. When it’s a big restructuring, yes, we see the savings, but that doesn’t mean it was a fun experience where here like, oh, I encourage this and I immediately see the decrease in usage or the average cost for, you know, whatever you’re measuring. So that, that makes sense to me. I can see where that would be rewarding. Because that’s not the typical way you get to see your impact. Sometimes it is, but most of the time it takes a fair amount of time. So I like that. I can appreciate that. So, you know, next question here. You wrote an article about improving cloud forecasting and in the article you talk about how establishing a common cost language is crucial to, you know, cloud financial management and working in that area. Can you elaborate on that a little bit and tell us more why that is?
Jeff Duresky:
Absolutely. So, you know, within, within cloud and the common cost language, you can either have, how do you speak about cost from a grouping perspective or ring fencing your cost? Is it an account perspective? Is it tagging So metadata on the services and that’s how your company thinks of costs or the pricing model that you’re looking at for the cost as well. And so you can have your on demand, so that’s what you’re paying for as you use it or as I mentioned, some of those commitment base discounts where you commit to it and then you prepay or you’re committing to the, to the service in exchange for a discount. And within a w s there are so many different ways to look at your cost. You can look at it from an unamortized and amortized a blended cost, what is your, how are you grouping or filtering the costs?
And so making sure that as you’re working with your tech team from a finance perspective, when you say we spent a thousand dollars last month, what is your grouping? Was that your unblended cost? You’re unamortized is that you’re amortized because you have a prepaid commitment that you purchased all upfront. And so being very clear with what that thousand dollars happens to mean. And then additionally, you may have contractual credits or within your chargeback model you may have some internal billing mechanism making sure everyone is aware. And so the the worst thing that would happen is, you know, from finance, you may want your tech teams to think all in on demand public pricing costs and not thinking about any contractual discounts, any savings plan benefit, just focus on what they control. And then centrally from finance, you’re gonna apply the discount factor after if they apply all of that and don’t tell you, and then you apply, you’ve now double cut or double counted any benefits. And it comes back to you weren’t on the same page from the beginning of your forecasting exercise of what costs that you expect it to load into your general ledger.
Paul Barnhurst:
Got it. So it sounds like a lot of it comes from all the extra things that can go on, you know, one the different ways you can buy it, you could prepay and so you’re ti you might be advertising a certain amount over 12 months, but that’s different than what you may have used in that month. And then you also have different contractual discounts you may get, whether that’s volume based, usage based, however that may be. And so you may need to make sure everybody’s applying the same assumptions and on the same page. So you’re talking apples to apples, you don’t get double counting or in some cases no counting Right. That type of thing. .
Jeff Duresky:
And, and when you’re, you’re using the, either the AWS tools or the the third party tools as, because there’s so much data that you can look at how you’re grouping those costs becomes very important. Especially if you do have those, those prepaid discounts and the number of conversations I have where numbers don’t tie out and you realize they were looking at, you know, unblended not amortized or vice versa. It just, it it’s a common occurrence.
Paul Barnhurst:
Yeah. I think we’ve all been there before. Maybe not, you know, your situation but where numbers don’t tie and you gotta get in and dig. Okay. What’s different in your assumption or the way you’re looking at it from the way I’m looking at it? ’cause Everybody’s confident, their number’s, right? <Laugh>
Jeff Duresky:
Absolutely
Paul Barnhurst:
Rarely does someone come into the call, here’s the number I have, I know it’s wrong, tell me why I have experienced that. But rarely is that the case, right? So it’s often a reconciliation exercise of what did you do that I didn’t do? So obviously we all deal with that. So how, how does finance help create that common cost language? Kind of what are the steps or what advice would you offer to make sure you’re able to keep everybody on that same page around how they look at expenses?
Jeff Duresky:
When you’re looking at it from a, from a forecasting exercise, one of the things that we’ve seen companies have success is just being very deliberate with documentation of this is how forecasting is going to go. Not only from what is your common cost language, we expect these costs, these are the filters you need to look at when you’re using whatever tool for the cloud costs. But also from a timeline perspective, what I’ve realized is how many people in tech just don’t understand a finance calendar? And so you may say, I need it at the end of the month. What you actually need it is business day minus two is when you need to load it. So you need it by business day minus four. So you can roll it up and say this is the high level before we make adjustments before we load. So you really need a business day minus five because you need time for them to have questions <laugh> and they think they have until the end of the month.
And so understanding what are the timelines from finance just having that awareness. The amount of people like I had no clue I thought I had until the, the end of the month and everyone’s in finance like, well the end of the month doesn’t mean the end of the, the last day of the month. You know, this is our, our closing calendar. And so the more documentation of how you’re approaching your forecast, how you’re gonna look at material workloads, you know, potentially saying we want to do a deep dive on our top five applications accounts and then apply generalized assumptions for the rest of ’em. Because we can’t do a deep dive on everything. We don’t have the bandwidth making sure everyone is aware of this is our approach, these are our timelines, this is our cost language. And so there’s no ambiguity of what both sides need to prepare.
Paul Barnhurst:
Great point. And I love the example you gave of end of month versus that means workday minus five and the business is like, wait, no, I thought I had till 12 o’clock on the 31st even though it’s a Saturday, I was gonna work on it over the weekend. No, no, no, I need it by Monday. That’s what I mean by end of the month <laugh>, that’s the 25th. Right. You know, those type of things like been there before. I think that’s a common challenge obviously, and I’m sure you’d agree. Not unique to cloud,
Jeff Duresky:
Not
Paul Barnhurst:
At all. It’s something we all deal with in finance, but it’s a great example of that importance of making sure we have a common language that we’re speaking to. I appreciate that. So when it comes to cloud forecasting, you know, I’ve heard you need a higher level of technical knowledge. We talked a little bit about that previously, but maybe can you talk a little bit more about why that is and you know, what level of knowledge do you think people need or is helpful when you’re dealing with the cloud role?
Jeff Duresky:
I think in any role within finance, the better you understand the function you’re supporting, whether it’s a business unit, a particular function within a business unit, the better that you’re forecasting and your variance variance analysis gonna be. You have context behind the numbers. Numbers in of themself are meaningless, it’s the why behind it. And so cloud somewhat amplifies that. There’s just far more moving parts. You know, because it is consumption based, you can help your tech teams understand the data that they’re looking at. They may be a bit myopic on their particular workload, their particular account, where within finance, a lot of times we can zoom out and then we can provide a much broader view. But then you, from finance, you wanna understand what are the, what are the big drivers. An example of that would be storage. And so, you know, storage is a simple concept, but there are multiple general purpose storage or other storage classes within the, the category of storage.
And each one is gonna have a different pricing lever. And so if you’re storing all of your data in a general purpose storage bucket, that’s hot storage, it has the highest performance, but you may not need that for all of the data. There’s certain storage classes that could drive up to 95% savings on there. And so being able to have an analysis that you’re storing a petabyte worth of data, it continues to grow. But I don’t see any retrieval costs in there. So is this the right storage class? So you can take storage and then from a finance learning perspective, understand what are the pricing and billing mechanisms of storage, you know, a simple Google of storage classes or ways to optimize storage. And you can bring those trends to the business where they may only be focused on the app performance and not the co the cost aspect.
Paul Barnhurst:
And you know, when you said that around storage, it made me think of the old days of, all right, are we gonna back this up right on a hard drive? Are we putting it to a tape each night? Because that’s a lot less expensive. It’s a, you know, we don’t need it very often. Very similar in the cloud. I’m sure
Jeff Duresky:
A hundred percent same
Paul Barnhurst:
Idea.
Jeff Duresky:
Yes. Now
Paul Barnhurst:
That totally ma totally makes sense to me. It’s about how often are you retrieving it, what level of compression do you need? All kinds of different things that go into that.
Jeff Duresky:
And so the trends that finance can help understand, or even looking at areas where maybe one business unit is doing a great job with their storage, taking that data and showing it to another business unit, what learnings can we share? And that goes to that cultural aspect of cloud financial management sharing best practices within your organization to become more efficient in the cloud.
Paul Barnhurst:
Yeah. And and it feels to me, you know, as I listened to all this, we talked at the beginning, you know, the, the technical knowledge is really understanding the business, right? You’re not being an engineer, you don’t have that deep knowledge that the engineer’s gonna have, but you need to be able to talk to ’em. You need to have those educated conversations, which should be the case for any FP&A role when you support the business, you really should pride yourself on learning the business. I always like to say, what’s one of the number one things someone can do to be better in FP&A and finance. It’s learn the business. You’ll get more benefit than learning one more Excel formula that you may or may not use in your career.
Jeff Duresky:
100%.
Paul Barnhurst:
Although now we have the whole Python library, so we can learn a lot of stuff in Excel, but that’s another story. So I dunno if you saw that yesterday, but I did, you just announced they add Python to Excel.
Jeff Duresky:
I need I need to get some of my teammates on that. Some of the, some of the projects they’ve done is nothing short of magic. Just how, how technically adept they are. And so I’m I’m curious to get their thoughts on it.
Paul Barnhurst:
Yeah, I’m curious to hear some of these more technical people as well. I don’t know Python, I’ve never learned it. You know, I’ve seen some basics. I know Excel well, but I’ve heard of all these use cases and it’s just fascinating to see how quickly things are changing right now. Yeah, it
Jeff Duresky:
There’s, there’s always the, the constant evolution and just understanding what else is out there and then how can you scale those learnings. Mm-Hmm. <affirmative> know how can you take something that you built in Excel, use this, you know, the Python library, whatever it happens to be, and now you can scale it out to countless users to do the, the business as usual, the you know, run the engine work much quicker to focus on the insights. And that’s where it’s gonna hopefully unlock some of that bandwidth for more exciting analysis.
Paul Barnhurst:
E exactly. The hope is it allows us to focus more on the analysis and the decision-making, the strategic that all these changes we’re seeing, whether it be Python and Excel Chat GPT and generative ai, whatever it might be, that they make us more efficient. We can really be the business partner of the business needs instead of as sometimes I like to say the data junkie, because we’ve all had those days where it’s like, all right, I’m done dealing with this data.
Jeff Duresky:
Again, the the context behind the numbers is what you’re trying to get to. Not just, we grew 3%. It’s the why behind it is why you’re there
Paul Barnhurst:
A hundred percent. It’s that, that why, and then what do you do about the why? Right? That next level of, okay, well I understand what happened, so what, why does that matter now? What, what do I do? And so I really, I like that. And I, I totally agree with you. So next question here. With cloud forecasting, why is it critical that finance works so closely with engineering? Why isn’t it something we can’t just, you know, talk to product, get some general ideas, run some numbers and move on?
Jeff Duresky:
It’s the dynamic nature of cloud. And as we mentioned, all of those, all of those pricing levers that you have, choosing the right service. And a good example is, you know, if you think of, of forecasting, and we say, we’re just gonna take the trailing 12 month average grow it at 3%, and, and we’re done. And we’ve seen customers where they were somewhat inefficient in the cloud. We’ve worked with them, they’ve done some of those prepaid commitment base, seen a decrease in cost, they’ve chosen better storage classes for their storage, they’ve seen storage classes fall off a cliff. And so their cost profile, maybe the first six months is really high and then it’s dropped significantly. If you just, you know, take that blended average, oh, you, this was this much in grow, you are gonna be materially off. And so understanding how my infrastructure is running now, is that what it’s gonna be in the future?
And so, you know, you need to have the trends of it, understanding what the trends, but because you can put something in place and change your trajectory almost immediately, that needs to come into play. And then also just with testing and product launches, having those incorporated into the forecast, just like you would with any other product, any other line item that you’re fos forecasting. But it may be, we’re gonna test for three months, we need to have a bubble, and then we’re gonna turn it off so we can’t continue to forecast very high, you know, test or non-production environments. And so the conversation of what we have now is that truly what we expect versus 3% growth and, you know, call it a day, I think becomes more important.
Paul Barnhurst:
And, and I would imagine, you know, you being for Amazon, you got thousands or millions or whatever the number is of different companies, I’m guessing it’s probably in the millions of using cloud, right? All the applications and everything. So I would imagine it gets pretty complex from your end to try to forecast what all your cloud costs are gonna be. I’m curious, is it one of more complex models you’ve had to work with in doing cloud? Like how, how challenging does that get to do the modeling?
Jeff Duresky:
And so for me, I’m actually a customer facing role, so I’m helping some of AWS’s largest customers with their forecasting models
Paul Barnhurst:
Got it
Jeff Duresky:
Versus doing anything internal to to AWS. So I’m trying to help the, the customers understand how are you forecasting, what are the touch points between tech and finance? Does finance feel like they have at a seat at the table? And does tech feel a sense of ownership? You see that with you know, different companies where tech will say, I know finance put in a number. They never talk to me. Finance says tech didn’t give me a number. I had to look at our last 12 months, and it grew, you know, 2% month over month. That’s what I used. I didn’t have any better information. And so helping companies understand the benefits of those interaction, those touchpoint is gonna then help their forecasting process and be much closer to the actuals is the hope.
Paul Barnhurst:
You’re seeing all kinds of different models and different issues and every customer has slightly different contractual terms and so you’re really trying to help them better forecast.
Jeff Duresky:
Absolutely.
Paul Barnhurst:
Yeah. That, that has to be a challenge. It’s kind of basically a consulting role like someone who’s a fractional C F O going into a bunch of different companies and trying to help them forecast their business and answer those questions.
Jeff Duresky:
Right. And, and with my team, we have a broad perspective, a very tech heavy, very finance heavy. And when we’re meeting with a lot of the customers, what we’re trying to set up for success is have both representatives in the room. And so us being able to tee up a conversation and see the customer, then talk about it, see the, the finance and the tech realize, oh, we have the same problem, or this is how we’re gonna approach it. And so we’re able to be a lever to help drive operational change within those companies. And so that’s been something really rewarding coming into, into this role because we can see what other companies have done, what have been successful, and use those experiences to then help another customer drive change and drive efficiency.
Paul Barnhurst:
Sure. And I, and I expect you see that in a couple areas. One, helping ’em get more accurate in their forecast, but I think even more rewarding is sometimes helping them think about things of how they can save cost. Absolutely. How they can be more efficient, you know, seeing that one that has a 30% reduction in cost and they’re so grateful for the insights you helped bring them.
Jeff Duresky:
Yeah. Those are, those are the great ones. When we can look at a particular service and, and say you’re using older instances. If you go to a newer instanceis a 10% savings. Look at our pricing on, you know, the AWS website just by moving to a newer ser a newer instance, you can drive 10% savings, scaling that out, helping them understand what the effort is gonna be, and then when it’s implemented, seeing the cost drop is fantastic.
Paul Barnhurst:
Got it. Thanks. So I’m curious, one question that led me to a question. Do sometimes you feel like you’re stuck playing mediator? Are there some where engineering’s like, well, finance didn’t tell me this, and you kind of have them airing their laundry as you’re trying to help ’em. Do you run into that? Sometimes
Jeff Duresky:
We have seen challenges that have arisen when you have those, when you have everyone in the room. And a lot of times it’s not a unique challenge
Paul Barnhurst:
Sure.
Jeff Duresky:
For example, I’m, I’m working on a blog post right now about reasons why your general ledger may not align one-to-one with your with your cost and usage report. And it’s gonna go into how your corporate calendar is, how your amortization policy at your company versus how AWSdoes it in the data. And so, you know, we’ve seen where tech is saying the numbers didn’t align to it and being able to say, and that’s okay. These are reasons why they’re not gonna be material. So you can focus on the true drivers of the cost versus any differences that don’t reconcile in that month. And so that’s where our experience comes in.
Paul Barnhurst:
Yeah. because I’m sure there’s lots of times the general ledger and the report are not the exact same, it’s accruals, whatever other reasons, the prepaid portion, you know, different things that engineering isn’t gonna understand. And so it’s like, well, finance, why didn’t you book the number that matches the report? It’s like, well, because there’s all these adjustments and trying to help them understand that. Yeah. I, I could a hundred percent see that for sure. So, you know, next question here. What are some of the key metrics you look at in, you know, this, this area of cloud finance? What are some of those key metrics to help you understand what’s happening that you really like to focus on?
Jeff Duresky:
Unit costs are, you know, we use that for so many different areas within finance and cloud is no exception there. When you can look at what is your cost per gigabyte, what is your cost per cost per instance, hour, whatever metric that you’re looking at. And that’s where you’re then able to see what are your trends, see the impact if you’ve purchased a savings plan, if you change the different service class that you’re using. And that’s where you can see the operational efficiency of the cloud. And so you, what you wanna show is maybe costs have grown and in a vacuum you’re trying to understand why did they cost your unit metrics are show have actually decreased. And you’re saying we are using more in support of X feature, but we’re doing it at a cheaper basis because of these changes that we’ve made to our, to our ecosystem.
And you know, example customer, we’ve talked about storage. It’s an easy one to have where worked with one that we were able to show when they implemented one of the other storage tiers, their volumes grew by 10 or 15%, their unit cost dropped by 50%. And so it was this huge you know, huge story of one business unit. Your volumes are growing, look at that unit cost, how can you scale it to the other ones? And you’re becoming much more efficient with the choices that you’re, that you’re making. And then the next step probably the, the holy grail of costs are gonna be those, you know, business metrics. So if you have the data to be able to say, this is our cost per click, cost per customer cost per campaign, and that’s where you’re needing to bring in not only your cloud cost, but your business knowledge, how good is the data that you have to support those ones? And it’s, it’s very much a challenge. We, we see some companies that do it really well. We see others that, you know, they’re huge companies and they just haven’t been able to make that connection yet for the business metrics. But they know that’s where they want to be to really tell the holistic story of the, of the cloud.
Paul Barnhurst:
I think you said good data and challenge in the same sentence. That’s just a given. So you know that that was what I thought as I heard that part around the data, but all of what you said makes sense. And obviously, you know, like almost anything, a lot of those key metrics are around the unit economics.
Jeff Duresky:
Absolutely.
Paul Barnhurst:
Really understanding what the cost per unit is, whether that’s click product, whatever it might be, there’s gonna be different things that, that makes a lot of sense to me. So next question here on cloud, and then I have a couple other questions beyond that, but what advice would you offer to someone listening to our show today who is just starting, you know, forecasting their cloud costs? What advice would you offer them?
Jeff Duresky:
I think just starting off with, with forecasting, say that you’ve been volun-told you own cloud costs, <laugh>, you, you know, you have, you know, a full business unit, you do the full reconciliation and now cloud is just one line item within your, within your four pul full portfolio, which I think is, you know, pretty common and a lot of people are probably nodding their heads to, to that one. The, the documentation of the processes and the assumptions is just key. And it that probably applies to, to everything. As you’re coming in new to, whether it’s a technology, a role, just being able to sit in meetings, hear what everyone’s saying, take notes, play it back to make sure you have that understanding. And then the, the documentation within cloud it’s gonna be who are the key people that you need to work with it, what is it the application owner?
Is it the product manager? Is the account owner depending on how you’re ring fencing your cost and what is the materiality? And then how dynamic are those costs? So sitting with them and just understanding for your top applications or accounts, what do they expect for the next six months, 12 months? Is there anything that is not captured in the current run rate that is gonna be material to the forecast as you look ahead for the next six to 12 months? And another thing that I’ve found is a lot of times it’s hard to get people to commit to a percentage. You know, if you ask ’em, what do you think your growth is is gonna be hem and haw, kind of wishy-washy on a number. And so going into some of those conversations of, I’ve done this analysis, I’ve looked at your cost, here’s what I’m seeing, and I’m projecting a 10% growth based on what I’m seeing, how does that sound?
Does that make sense? And giving someone a number to react to, well, 10% seems a little high, let’s lower that. That is a much easier conversation. It’s a little bit of them telling you where you’re wrong, but you’re documenting, you’re giving ’em something to, to react to. And then you can have that within your assumptions, you know, spoke with so and so. This was the trend, it was growing at 4% after discussions, we think it’s actually only gonna grow at 2%. That’s what we’re loading into the, the budget. And so doing the, the little bit of the research before and then giving people a number and assumption to react to can help have more productive conversations is what I’ve found.
Paul Barnhurst:
Great. Thank you. I appreciate that. And that’s good advice. And yeah, not just for cloud, but a across the board, there’s some really good advice there when you’re doing forecasting and working with your business partners. So this is a fun question we like to ask everybody here before we get into our get to know You section. So can you tell me about a time in your career when you experienced what we call a strategic moment, you know, i e a strategic insight that later empowered you to drive change or to help improve the organization? Yeah,
Jeff Duresky:
So I’ve mentioned within cloud you have your payer account and then you can have, we’ll call linked account. So linked accounts that roll up to your payer account for you know, for your billing, your invoicing, and you can have hundreds or thousands of linked accounts. And so within my, my cloud finance life over the last couple years I was working on a project where I was noticing a cost increase at the account level. And so as you dig in more, you know, I see hundreds of new accounts coming on every month and each one has very minor costs, but every month there was 200 more accounts than there were the next month.
And so at that point, you need to realize something’s going on. There’s a reason why these accounts are are being created. And so I wanted to understand what is the what is the project mm-hmm. <Affirmative> start working with, you know, some of the engineers, some of the product owners, and playing back the cost. I noticed we’re increasing a lot of the accounts. Yeah. It’s for a, a project that we’re gonna migrate some of our workloads over the next six months or more. So we’re staging them right now. Makes perfect sense. That seems like a good plan. Well, when I look at the, the cost within that account, I see you’re running service A, B, or C and you know, the total cost is, you know, five, $6 a day on that. Yeah. It’s, it’s immaterial, you know, $5 a day. So now I’ve taken this information $5 a day, less than a cup of coffee, but we’ve got 365 days, we’ve added hundreds of new accounts, you’re telling me we’re gonna continue to add new accounts and we’re not doing anything with them for at least six months.
And so you take this, these little bits of information, which in isolation seem like they’re immaterial, you boil ’em up and the numbers are just eye-popping on what you could potentially say are wasted costs based on everything that you’ve just learned, you know, leading up to that. And so, you know, it became realizing in aggregate, these are gonna be a large cost. What are we doing with the accounts now? Are there ways that we could optimize what are going on with those services? And so we were able to, to look into the services, realizing it was a monitoring expense for something we didn’t need. We were able to turn off the service cost, drop off a cliff, it can be turned off when the accounts are needed. And so you just realize that no one had the full picture of it, or they may have kind of had in the back of their mind, here’s what the project, but it was a, a zoom in to get the information and then zoom out for what the impact of that information was. To see the broader scope of this, this overarching project and what it, it made everyone realize was there needed to be a more formal program. What are other costs that may catch everyone off guard with this? Are there other areas where mm-hmm <affirmative> we’re, you know, costs are being spun up or, you know, one policy may have impact across all accounts. What is that gonna gonna be? And so it was one of the reasons where a formal program ended up being set up for this, this project.
Paul Barnhurst:
Got it. Thank you for sharing. And I can see how it made a big difference when you aggregated it.
Jeff Duresky:
Yeah,
Paul Barnhurst:
Right. In isolation, it looks small.
But many small things before you know, it become a big issue. And sometimes people don’t have the full picture to understand that. So that’s a huge insight that drives, you know, big savings. So I really appreciate that example.
Jeff Duresky:
And that’s where finance, you know, can really, finance can really play a big role in that, that people can become focused on their workloads, especially in the, in the tech world, you only have your one application and so being able to step back and see the entire landscape, see trends that are broader than just one account or application or however you ring fence your cost can be huge from a storytelling perspective.
Paul Barnhurst:
Sure. No, that makes sense and I appreciate that. And so this next section is one of my favorite sections is our Get To Know You section. We kind of like to ask these questions almost like rapid fire where you get 30 seconds to answer each one. So we have four questions here we’re gonna answer. And the first one is, what is something interesting about you that not many people know? Something I wouldn’t find online, like something that makes you unique?
Jeff Duresky:
So anyone that’s worked with me knows that, you know, I like getting out to, to the gym at lunch or at some point, you know, I try to try to do something around my lunch break, but what people probably don’t realize are the amount of emails I send to myself when I’m out in the gym because a thought pops in my head of something I’m working on. So my inbox is just filled with a thought or a comment that <laugh> I needed to step away from a project. Something comes to me, let me email it to myself real quick so I don’t forget it. Otherwise I’m gonna be repeating it in my head. I’ll come back from that and I may have five or six emails from me. It’s just like a one sentence thing of look up this cost, or here’s a good line to use in this deck I’m working on.
Paul Barnhurst:
I like that. Thanks for sharing. That’s a fun one. So next one, if you could meet one person in the world, dead or alive, who would you meet and why?
Jeff Duresky:
Such a broad a broad question. So I’m gonna go with recency bias for me, I just finished the book, the Splendid inand the Vile by Eric Larson. It’s all about Winston Churchill during the Blitz. And he’s just such a polarizing figure when you look at, you know, failed campaign in Gallipoli or you know, how the, the Bengal famine and how he handled that. But at the same time, you know how the country looked to him for support during some of the darkest times of World War II. And then he was also known, you know, for his biting wit, such a excellent conversationalists. So, you know, reading that, reading the descriptions of some of the dinners he had would would be a fun fly on the wall for for a dinner conversation with him.
Paul Barnhurst:
I’m sure that would be, it’d be very fascinating. That’d be a fun one. So next question here, what is the last thing you Googled, you know, looked up on YouTube or used chat G P T or whatever generative AI tool you want about how to do something in finance FP&A or Excel?
Jeff Duresky:
445 accounting schedules. So I, I, as I mentioned, I’m working on a, on a blog of reasons why your, your data, your cost and usage report may not tie out to your general ledger. So, you know, researching different amortization policies came across this and it, you know, unlocked something. I think I heard about it in business school, but I had no experience with and just realize like, wow, you know, this equal 13 week period, but what are the implications on that third month, five week period, especially as we’re we’re looking at the amortization of prepays within
Paul Barnhurst:
Sure.
Jeff Duresky:
And so that led me the rabbit hole of this entire accounting policy I knew nothing about.
Paul Barnhurst:
Yeah, that is definitely a rabbit hole and I know, you know, 445 is common in retail in some places. I’ve never really, I’ve never had to deal with it, but I can see where it can make a huge difference, right? You got a different week in every third month and you gotta adjust for that just like seasonality or anything else. So I appreciate that one. So next one. Favorite, favorite thing about Excel? Favorite function feature.
Jeff Duresky:
I think, you know, the default is gonna be, you know, a pivot table and SUMIFs and you can do about everything <laugh>. But Formatting features, just the ability to format to make long strings of numbers legible, especially from a finance perspective. When you wanna make sure people can easily read tables of data and you don’t realize, you know, if it’s kind of natural for you within finance, you get a thing of numbers and you format it instantly so you can read it when you see that are not formatted properly and you’re like, okay, where should the comma go? You know, just, it makes it so much more difficult and so having an eye for formatting can make a very complex chart much more manageable to to digest.
Paul Barnhurst:
Yeah, there’s a lot of great formatting features and definitely important when you’re presenting to the business. So we’re heading into our wrap up section. Have two questions left for you. So the first is,
Jeff Duresky:
What advice
Paul Barnhurst:
Would you offer to someone starting a career in FP&A today?
Jeff Duresky:
I had mentioned earlier about the forecasting and the documentation, but something that I’ve told a lot of people that I’ve worked with over the years that are new to a group or you know, new to FP&A in general, taking notes, you know, you can go into a meeting, you’re new to the company, hey, follow me into this meeting. Just sit there and listen and you have no context, you have no idea what is going on, taking notes, trying to make sure you understand this is what I think they spoke about. And then playing it back to either your direct or you know, some of your teammates, this is what I think was going on, did I understand all of that? Then it gives you an area to then understand where to ask questions. I got this point, what does that mean? And so those first couple weeks, if nothing else, you can take notes, show that you’re focusing and paying attention.
You know, it’s an important thing. You don’t zone out if you’re just not hearing, you know, words that you’ve never heard before. And then that helps, you know, with your, with your documentation, with your assumptions. As you get clearer in the role, you can go back and say, I remember these meetings. Now everything’s starting to make sense. So maybe you don’t take as good of notes when you’re six months on the roll. But coming into it, that’s been the easiest way, at least for me to try to get up to speed of trying to summarize what I understood and play it back.
Paul Barnhurst:
So just like we’re in college, take notes.
Jeff Duresky:
Absolutely. <laugh>,
Paul Barnhurst:
I like it that there’s definitely a good one. And studies have shown that even if you never go back and look at your notes, you remember more just by taking them.
Jeff Duresky:
I think the focus, that’s good
Paul Barnhurst:
Advice.
Jeff Duresky:
The focus. ’cause Otherwise, you know, you’re, you’re in there, you’re sitting, everyone’s talking a language that you don’t know because you’re brand new to this role and it’s very easy to sit there and just say, I don’t know what they’re talking about. Versus, oh, they said BAU, what is that acronym? Let me look that up later. Or whatever. You know, all of the acronyms that they’re gonna use at a new company. Kind of having that, that running list. I used to always have my notebook of acronyms and people to make sure I could reference that until it became second nature.
Paul Barnhurst:
Yep. Agree. Last question. If someone wants to get ahold of you, what would be the best way for them to do that?
Jeff Duresky:
Find me on LinkedIn. I love learning about how, you know, other companies that are, you know, in the cloud, whether they’re just beginning or you know, mature on the cloud. How are they approaching cloud finance? ’cause We’ve not seen, you know, one standardization and because it is so new, I think there’s a lot of learnings people can offer and share that become best practices for everyone else.
Paul Barnhurst:
Great. Thanks. We’ll make sure to put your name in the show notes and I know there’s a few articles you’ve sent, we’ll share those as well. Really enjoyed having you on the show today, Jeff, and being able to talk, you know, cloud finance is definitely an area I haven’t had to deal with in my career, so it was really fascinating to talk and learn a little bit more about that and I’m sure our audience will enjoy it. So thanks for being on the show with us today.
Jeff Duresky:
Thanks for having me. It was great.