This is episode three of Decoding the C-Suite, a five part miniseries from Recruiting Future about how leadership really thinks about AI, recorded with the leadership team at SmartRecruiters seven months into their acquisition by SAP.
If you’ve ever struggled to get AI investment signed off, this episode is for you. Tom DiDesidero, CFO at SmartRecruiters, explains exactly what he wants to see in a business case, why he backs short, measurable pilots over big commitments, and how to bring evidence rather than instinct. He also lifts the lid on the real economics of AI, from unpredictable inference costs to gross margins far below anything the SaaS world is used to, and makes a prediction about how the AI and SaaS worlds inevitably come together.
Key Takeaways
- AI breaks the predictability that the CFO's job is built on; usage-driven inference costs make forecasting harder than anything SaaS finance has dealt with in 30 years.
- CFOs back pilots over big commitments; short-term, well-defined tests with measurable outcomes are how AI investment gets signed off.
- Bring evidence, not "it feels like"; with AI research tools available, there's no excuse for a business case that can't support its own conclusions.
- Companies are measuring AI internally in ways they aren't reporting externally, and with no meaningful SEC guidance, that gap has to be reconciled.
- SaaS companies have the customers, AI companies have the capabilities, and each could destroy or complete the other; the two worlds coming together is inevitable.
Transcript
Matt Alder: [0:18] Hi there, and welcome to Decoding the C-Suite, a five-part miniseries from Recruiting Future about how leadership really thinks about AI. I'm spending the series with five senior leaders at SmartRecruiters, seven months on from their acquisition by SAP, decoding how each function around the leadership table is thinking about AI and what that means for anyone trying to drive change in talent.
Matt Alder: [0:44] Today, we are following the money. If you've ever struggled to get AI investment signed off, this episode is for you. Tom DiDesidero is SmartRecruiters' CFO, and in our conversation, he explains exactly what he wants to see in a business case, why he backs pilots over big commitments, and how AI is breaking financial models that have held for 30 years in the SaaS industry. He also shares some numbers about the real economics of AI that genuinely surprised me, along with a prediction about where the whole industry ends up. This is Decoding the C-Suite, episode three, The Money.
Matt Alder: [1:26] I'm going to start with a really big question. What would you say the biggest challenges are that AI is creating for CFOs?
Tom DiDesidero: [1:36] I think about it every day. The CFO job revolves around predictability and hitting goals, and the AI landscape is one where it really doesn't allow for easy predictability as one would hope. Then when you think about the underlying drivers to that, it's very expensive, and connecting that cost to the value that you're pricing to your customers is an ongoing observation. From an employee perspective, as we look to introduce and are introducing capabilities, we think about general and specialised use cases, and we also acknowledge very candidly that people are uncomfortable in the environment. So it's about acknowledging that point, and setting goals and driving behaviours that say, hey, it's okay to be uncomfortable. Failure is part of it as we learn to adopt, and ultimately you come out with good pieces too, and we embrace those.
Matt Alder: [2:43] I suppose also from a technology perspective, just in terms of the investments that you make in technology, AI must be making that really confusing as well.
Tom DiDesidero: [2:53] Yeah, so there's a bit of what I pay for as a CFO and what I sell to customers. I'll start with the pay example. It's much like any business case you would develop, in that you look to understand the problem you're solving and the cost with which to do that. The goals you set to decide to move forward with something are really important, to make them measurable and adapt as you go. So I'm very much a proponent, particularly when I'm paying for something, of the pilot focus. Give me something short-term, pretty well-defined, and then we're going to measure outcomes based on that before we commit further.
Tom DiDesidero: [3:40] Our customers look at it the same way. We went to market very quickly with our Winston product that others have spoken to, and I think part of that was making it credible and tangible. Okay, we're going to start here. We're going to start small and prove the value before we know you'll be comfortable in committing further. Those two principles are largely related, and that's how I think about it.
Matt Alder: [4:08] How do these times compare to what's gone before? Is AI the most disruptive thing that's come along, or can you think of parallels earlier in your career?
Tom DiDesidero: [4:18] When I first came out of school, email was just coming out. And then e-commerce, putting your card into the computer to pay for something. We were all uncomfortable with those things. As I think about running a business in this environment, there is clear value and opportunity as we move forward. But what is also similar is it's very expensive, and the cost of failure, both financial and opportunity costs, is significant. So you really have to weigh and be accountable for how you're measuring success progressively along the way.
Matt Alder: [4:53] And what's the tolerance for failure? Does that change because of everything that's going on?
Tom DiDesidero: [4:57] I think it depends on the type of company you are. We were a standalone company up until six or seven months ago, and the cash on hand and your balance sheet really steer how fast you can go on investments and the level of risk you can take. One of the great things about being part of one of the largest companies in the world now at SAP is there's more of a safety net to embrace the go big, go bold and play to win, knowing that you have more resources available to you to be more aggressive.
Matt Alder: [5:33] Let's dig into that and talk a little bit more about the SAP integration. Tell us more about that. What is the opportunity that it's creating, and how is that going to benefit the customers?
Tom DiDesidero: [5:46] In any acquisition, the integrations are challenging, and we are learning mutually how to work together, being a smaller company now part of a much bigger environment. I think the opportunity that others have spoken to as well is to take who we are, in that we're decisive and quick and innovative people by our nature, more of a startup mould, and when you put that inside a large hypermatrix organisation that has the benefits of scale and structure and resources, you can figure out how to make those two work through the integration process. They're challenging, but that ultimately is how these outcomes will be measured.
Matt Alder: [6:31] In terms of integration, you've spoken before on other podcasts about the importance of execution over theory. How is that looking at the moment, six months into this? How does it compare to the theory when you started, and what lessons have been learned so far?
Tom DiDesidero: [6:51] So when you have these pie in the sky opportunities, we all think about the shiny object. When you think about the execution point, going beyond theory, everyone tends to focus on the big prize, and the opportunity is exciting. The discipline that you must create, though, is establishing OKRs or goals that are short-term, measurable goals as part of the larger thing that you're developing, and keeping people focused on those short-term outcomes is really a priority, because distraction is something that you continually face. While well-intended, and maybe even good ideas come out of those distractions, because the cost of success and failure is so significant in this time, I always focus on what the next three to six months look like. Are we very clear on what those deliverables are? And you have to pivot and adapt as you go.
Tom DiDesidero: [7:59] So while we do set an annual goal like everyone does, the most important goals, I believe and we believe as a business, are the OKRs that we set that are more short-term oriented, in the first half of the year for example. That keeps you on pace, and when you see the success that comes from that, it gives you more comfort in making decisions, again with predictability in mind, so that you can continue to put your pedal on the gas and move forward.
Matt Alder: [8:25] You've talked about this the whole way through the conversation, but I want to really highlight that discipline that needs to sit behind innovation in a very unpredictable environment. You spoke to some of the things that are in place to do that, but what's the big principle there? What would you say to any organisation that has to innovate but has to keep that under control at the same time?
Tom DiDesidero: [8:50] So there's a level of, yes, be excited about the opportunity, but also match that with candour and realism. One of the roles that I get to play as a CFO is to encourage the excitement. It's what the customers are looking for in terms of our product evolution. It's where the great employees want to work. They want to work in environments where they're creating cool things and adding value. And then you kind of temper that with, this is great, but let's define measurable items along the way here to balance both. I would love to throw as much money as possible into where opportunity goes, but it has to be done in a responsible, measured way. And we're all facing that right now as CFOs.
Matt Alder: [9:44] And is there tension in that? How does that tension work?
Tom DiDesidero: [9:48] It's like any emerging product that you've developed as a company in any of your other roles. That's healthy. You want that. The worst thing that can happen is to be a yes person, where you have a leadership that tends to fall in line with whatever the board or the CEO wants to do. We have healthy conversations to figure this out as we go. There is, like anything, a lot of uncertainty, and you acknowledge that going in. We measure as we go, and we hold each other accountable for falling behind on things, if other competing priorities might be popping up. You have to say no to those things at times. It is one of the more interesting times in my career, I would say, when I look back on the last 30 years.
Matt Alder: [10:38] I want to flip this out to talk about employers, because with everything that's going on at the moment, employers need to invest in new technology, whether that's with existing providers or new providers. And talent teams, HR, talent acquisition, have not been good in the past at putting business cases together in a way that the CFO wants to see. What advice would you give to them in terms of speaking the language of the CFO and putting a business case together that's going to get them the investment that they need?
Tom DiDesidero: [11:17] So the principles of good business case development really remain the same. The problem we're solving for, how it applies to our product market fit, being able to clearly articulate the main drivers and assumptions so that they can be challenged, pressure tested, and ultimately arriving at a decision that we're going to move forward in a certain way.
Tom DiDesidero: [11:43] I encourage people to bring evidence, not "it feels like". I admit that's always a part of it. There are certain things I know and have evidence for, and there are other things that are more qualitative, or we'll see as we go, and that's a very natural part of agile development. I think that you now have resources available to you to do better research on these business cases, using Claude and other capabilities. Take advantage of that. Your ability to be more depthful and bring other resources to support your conclusions has never been easier. And I look for people to do that. I met with someone this morning and I said to them, I see you put together this proposal, how did you go about it? And he said, well, I actually put my initial draft through Claude. I said, there you go. It allowed him to supplement the ideas that were in his head very easily.
Matt Alder: [12:41] I think when HR and talent hear the word evidence, they think automatically that this needs to be hard numbers. This needs to translate into measurable value for the business. And sometimes that can be quite difficult in that area. Is that what evidence is, or are there other ways of thinking about it?
Tom DiDesidero: [13:01] So I think how ROI is being looked at in the AI environment is changing. The numbers matter, and they do. I look at some of the guidance coming out of the SEC, for example, or actually the lack of it. They're not pursuing real meaningful AI guidance on how they're measuring what's reported out publicly. As an investor, you get concerned about that, because you want to understand how companies are progressing or not with their AI capabilities. So that's a public market kind of perspective.
Tom DiDesidero: [13:46] I still come back to the basic virtues of show me enough of the math that we feel good about, and then I'll work on the grey or the softer parts of that, particularly if I have to go help sell it beyond our initial team. I also rely upon people's reputation and credibility as well. Someone who has a good history of success brings you something that is maybe more risky, at least on its surface, and you're more inclined to trust them. So think about how you present things and who's presenting them. Coupled with the research I mentioned, ultimately credibility leads to trust, and it gives you the ability to push the envelope, to stay aggressive on these emerging opportunities.
Matt Alder: [14:35] It's interesting what you're saying about the SEC there, because there's another level of risk with all of this, isn't there, in terms of the financial risk? We don't know what AI is going to cost or what the cost model looks like in the future. How do you think about and manage that?
Tom DiDesidero: [14:49] So I look at some of the vendors that we have, and I understand intimately what drives the cost of that part of our business. Where it's hard to predict is the usage, which really drives the cost. There's a cost per token, for example, that drives a lot of this. It's referred to as inference cost in our world. We are all struggling with keeping our predictive models up to date, because usage is very uncertain, and you really want to link it to the consumption that's driving it, what people are paying for. I think good finance organisations in this age really have to develop that skill quickly.
Tom DiDesidero: [15:35] In the old SaaS world we were in, it was a lot easier. Users drives fee, and it was well-established. We've been doing SaaS for what, 25, 30 years now. So this is a different way of running a business. The underlying drivers are far more variable than they've ever been. You have to acknowledge that and then work quickly to institute as much structure as you can around it. And I ask people, what's your degree of certainty, or how do you feel about what you're putting in front of me? They know I'm asking because I want to get to the right outcome. So it comes back to, this is what is solid, this is what isn't, and let's help with that judgment.
Matt Alder: [16:16] As you say, those SaaS business models have been in place for decades and decades. Do you think that they're going to fundamentally change moving forward?
Tom DiDesidero: [16:23] Yeah, I've been reading a couple of articles the last couple of days. It is going to change. You see investment now in AI services. Anthropic came out, I think it was yesterday, with some plans around that. They realised that in order to implement these very powerful tools, there are very real process and human impact changes to implement into workflows to ultimately be effective. So I see a whole opportunity around that. The world is now acknowledging that you need the people to make it work, to apply it into the business world.
Tom DiDesidero: [17:03] We're a very complex organisation at SAP, so the complexity there makes it harder to implement. But on the other side, a smaller company or an SMB business, they don't even know what AI is. They just know, maybe I should be looking at this. So there's an opportunity for people to help bring the capability into a form people can understand, digest, and then take their own piece of value from.
Matt Alder: [17:30] To come back to that ROI point, if you could change the way that the industry talks about ROI on tech investments or AI, how would you change it? What would you do?
Tom DiDesidero: [17:42] So it's a little bit obscured right now, as part of something much bigger in a business. There are no rules around reporting it separately yet. You see on earnings calls people speak to it kind of tangentially or anecdotally. Not to get too technical, but we look at gross margin as an old standard in the SaaS world. Being over 80% is always, yeah, you're in the green. We're seeing gross margins in AI, some are negative, but 25 to 60% is the range. That's very low relative to what we're used to. We're going to figure out that sweet spot in the middle. But what I am looking at internally is different from what I'm being asked to report externally, and that's the great connection that has to be made. If you're running your business in a certain way, looking at certain things, but you're not telling the public community that level of detail, that has to be reconciled. I'm looking to see some progress there.
Matt Alder: [18:47] When was the last time that we saw a shakeup like that? Was it when the internet launched?
Tom DiDesidero: [18:52] Yeah, I think the good parallel is the internet and the payments business. So gross merchandise volume, GMV, is the dollars you take in, but then your take rate, your revenue, is usually a fraction of that. It's understanding that volume versus revenue are very different things. Engagement is a much different, less regulated measure than, say, GAAP revenue would be.
Matt Alder: [19:18] I think one of the interesting things is when we look back at these previous technology revolutions, there's always this sense of invest, invest, invest, and the rules of business have changed, and those things don't matter anymore because this is a brave new world. And it's always come back to, actually, no, the rules of business still apply. AI is in a very crazy place at the moment. Do you think that it's inevitable that we will come back to the normal rules of business, or is this changing everything forever?
Tom DiDesidero: [19:47] So I had a conversation with a VC on this topic recently, and where I think this lands is, you have the SaaS companies over here, you have the AI players over here, and they both have what the other needs. AI doesn't have the customer relationships, but they could displace the tech very easily over time, maybe months, but relatively easily. SaaS needs the capabilities that AI is bringing to the market. Those two worlds will come together. No one's going to walk away from a multi-billion dollar or several hundred million dollar SaaS business and give it away. But they need the customers ultimately to monetise what their investment is. Those two worlds will come together. It's inevitable.
Matt Alder: [20:36] And final question for you. What does good leadership look like in this AI-driven world?
Tom DiDesidero: [20:45] Being comfortable with rapid change is critical. As leaders, you acknowledge that. You also are mindful to project calm, even though you may be going a million miles an hour in your head thinking through things. I think that's important. But I also think it's important to practise candour and realism with your employees. They're on the journey with you, and we need each other to make all this work. So I try to balance the optimism and the excitement with the, hey, I know this is hard, I know this is uncomfortable, it's uncomfortable for me too. When people see the humanisation of that, and they can identify that, hey, we're all in this together, that to me has always been a principle of good leadership. Now it's being done in an environment where decisions are far more expensive than what they've been, at least what we're used to.
Matt Alder: [21:49] Tom, thank you very much for talking to me.
Tom DiDesidero: [21:51] Thank you. Pleasure.
Matt Alder: [21:53] That was Tom DiDesidero on how a CFO really evaluates AI. In the next episode, we go inside the product itself. Shefali Netke leads product design at SmartRecruiters, and she faces a genuine dilemma in designing AI products for a market where some customers want to leap ahead, while most aren't ready. Her answer says a lot about where all this is heading. Subscribe or follow wherever you listen to your podcasts, and I'll see you in episode four.
