AWS Cloud Enterprise Strategy Blog

CFO Series: The Sniff Test

If you’re a CFO, you’re acutely aware of your fiduciary responsibilities. Over the long term, the cloud is likely to be a material element of your cost structure. The decision to move to the cloud must be made with appropriate care and diligence. To add complexity to the decision, much of your due diligence is based on projections that—given uncertainty and rapid change—are of limited decision-making value. The challenge is that in a fast-moving world, your duty includes…well, moving fast enough to keep pace so that the company is not disrupted. If the cloud really is as valuable as many suggest, then there’s an opportunity cost for every day you don’t take advantage of it, and therefore every incremental day of analysis might be destroying business value. How can you reconcile this with your fiduciary responsibilities?

Fortunately, you have other good sources of information to supplement your cost modeling. One is simply your executive sniff test: Does what you’re hearing make sense? Does the hype seem right, or does it sound like…well, hype? Or, on the contrary, does it seem right that your company is better off maintaining its own infrastructure in its own datacenter? As senior leaders we’ve tuned our intuitions to serve our judgment when the data is less exact than we’d like. We’ve also learned to use third-party information from industry analysts and peer references. Yes, all companies are unique—is yours unique in a way that makes moving to the cloud a good solution for others and not for you?

But there’s no reason for analysis paralysis anyway; cloud investment decisions are far less risky than they might appear. Because the cloud is pay-as-you-go, it’s possible (recommended, in fact) to make incremental investment decisions as you see the cloud demonstrate value. With that in mind, you should avoid overanalysis by supplementing your cost modeling with intuition, third-party information, and a good risk mitigation strategy. In fact, doing so may be the only reasonable way to assess the business case, because so much of the value of the cloud is tough or impossible to evaluate quantitatively: cost savings are just a small part of that value, while a larger portion usually comes from difficult-to-estimate improvements to agility, innovation, resilience, and competitive differentiation.

In this post I’ll talk about some of the benefits of the cloud you may be hearing about and make the case that they should pass your sniff test.

The Cloud Can Reduce Costs

Can it? That’s typically the question we’re trying to answer in our due diligence spreadsheets, right? But as I’ve said, those spreadsheet models are extremely sensitive to assumptions and often complex enough to obscure the real dynamics. And they probably exclude a lot of the real cost savings that come from the cloud. Your actual savings will depend on your success at managing those cloud costs (what we call cost optimization or FinOps). So, at a high level, why does it make sense that costs would be lower in the cloud?

Most obvious is the tremendous economies of scale Amazon Web Services (AWS) derives and can pass on to you.1 Because we operate at the scale of millions of customers in 24 regions and 77 availability zones, we can buy and build at extremely low cost. We’ve automated and refined our infrastructure provisioning processes and have become experienced in executing them. Yes, as you’d expect, we earn a margin—but we have competitors and, like our retail side,, we’re comfortable with low margins. Also, like, our mission is to be the world’s most customer-centric company. Since we believe that cost is extremely important to our customers, low prices are central to our strategy. Every service (product) team at AWS has a mandate to reduce its costs and pass the savings on to customers. As a result, we’ve reduced prices 85 times.

A second—sniff—reason to believe that we can reduce your costs is that your datacenter has unused capacity—capacity you’re paying for. Each time you add more infrastructure, you add capacity you won’t use, as a safety margin and because your needs fluctuate. According to a study reported in Forbes, 30% of physical servers are comatose—that is, they haven’t done anything in more than six months.2 We’ve found that in company datacenters, servers are used to about 15 to 20% of their capacity.3 In the cloud, however, you can adjust your infrastructure as your needs change, and therefore always keep it “rightsized.”

A third reason: you’ll be using cloud services we created by writing software, including everything from the mundane code that lets you provision and secure your infrastructure to the 175+ higher-level services such as machine learning and analytics we offer. The cost of that software development is spread over many customers and we can provide it to you with very low incremental costs—and therefore at a low price. Consider Amazon Rekognition, our machine learning service that recognizes objects in images. It would be expensive for you to build something like it—you’d hire people with PhDs in artificial intelligence, spend years of programming, and buy sophisticated hardware. Perhaps Rekognition isn’t relevant to your business. But many of our product offerings no doubt are, since they include things like databases, analytics, natural language processing, support for sensors and connected devices, virtual reality, augmented reality, and more.4

Our recent study of over 1000 AWS customers shows an average reduction of 27.4% in infrastructure spend per user and a 12.3% reduction in infrastructure spend per application. Yes, I know, as executives we’ve learned not to believe vendor-provided statistics. But it’s consistent with everything I’ve said and consistent with customer examples we can provide, from GE’s 52% savings on oil-and-gas computing infrastructure to LA County’s 60% savings on its call center systems.

True: everything I’ve just said doesn’t prove that you’ll save money by moving to the cloud. My point is that it’s extremely plausible. It should pass your sniff test.

The Cloud Can Reduce Your Risks

The cloud reduces risk for your company. For example, you can test new ideas before you commit large investments to them. You can spend less on new products you release until you’re able to grow their revenues. You can avoid large investments in fixed infrastructure that might turn out not to meet your needs later. And you can vastly improve your security posture (you do have to work at it, though!).

These claims might sound too good to be true, but actually you should recognize them as quite plausible. Let’s start with security. AWS was built from the ground up to be secure. We’re able to spend prodigiously and hire some of the best security talent available. We’ve had to make sure our services satisfy the needs of customers like the CIA, Homeland Security, Nasdaq, Barclays, DBS, Medstar Health, Novartis, and Bristol Myers Squibb. We’ve also made sure our services satisfy the needs of compliance frameworks from around the world. The work we’ve put into satisfying the needs of other customers benefits you as well.

When it comes to non-security risks, you’ll notice that in using the cloud you’re passing some of your risk on to AWS. We’re the ones who invest up front in fixed hardware and software that later may need to change with new technology or customer needs. You can change your infrastructure at any time in AWS—the risk of buying infrastructure is on us. AWS also improves your cash position because we’re the ones who lay out cash in advance for fixed infrastructure—you just pay as you go.

The cloud can clearly reduce risks in launching a new product. Since you pay for the amount of infrastructure you use, your costs are low when you first introduce a new product and only grow as your success in the market grows. Without the cloud, you’d have to gamble on infrastructure and other fixed costs to release your new product before you knew whether it would be successful.

True: everything I’ve just said doesn’t prove that your risks will be lower in the cloud. But there are good reasons to believe they will.

The Cloud Can Make You More Agile and Innovative

No, we definitely can’t guarantee this. It’s possible that you’ll develop or maintain a culture that resists new ideas and fights change. But it’s plausible that you’ll instead become more agile and innovative after migrating to the cloud. The cloud promotes innovation by reducing the cost and risk of trying out new ideas, speeding up feedback cycles for vetting and improving new products, and making advanced services like ML, AR/VR, and IoT available for immediate use at low cost. It also increases agility through elasticity (fast scaling up and down), pay-per-use, and the ability to change infrastructure on the fly.

It’s not easy to factor this into your business case for the cloud. Many organizations treat innovation and speed as if they were “nice-to-haves.” It’s hard to attach a dollar value to improving your agility and your ability to innovate. After all, you don’t know yet how you’ll use that agility or what innovations you’ll devise.

And yet, how can you not put a value on these things? Their value may very well overwhelm the value of your cost savings and risk reduction. One new innovation can create an entirely new market and a substantial revenue stream, and agility can mean the difference between bankruptcy when your business is disrupted and trillion-dollar market caps when you seize a new market opportunity. These benefits can dwarf the ones represented on TCO spreadsheets.

So, again, I can’t prove the benefit, but I propose that it can satisfy your sniff test. Why wouldn’t the ability to test new ideas before committing to them and fast and low-risk access to new technologies help you innovate?

Customer References

Many, many, many companies are using the cloud. It might be that your company is different, and the cloud won’t be effective for you. Maybe you should build your own reservoirs and water supply system instead of using the public ones. But that would seem strange, given the breadth of companies that are committed to using the public water supply and sewage system for their bathrooms—companies of all sizes, geographies, industries, ownership structures, sectors.

In financial services, companies like Capital One, Nasdaq, Barclays, Intuit, and DBS Bank find value in the cloud. In healthcare, companies like MedStar Health, Novartis, Johnson & Johnson, and Bristol Myers Squibb. 6,500 government agencies and 10,000 educational institutions. Coca-Cola, Under Armour, Unilever, and Kellogg’s; Toyota, Volkswagen, and BMW; McDonald’s and Dunkin’; Qantas, Cathay Pacific, Korean Air, and Ryanair; Warner Bros. and News Corp; FC Barcelona, Formula One, and the NFL; Comcast, Verizon, and T-Mobile; GE, 3M, Canon, Philips, Samsung, Siemens, and Pitney Bowes; Royal Dutch Shell and BP; Lululemon, Adidas, and Levi’s.

Some of these are probably companies whose judgment you respect. Intuitively, it would be strange if your company was so unique that this broad applicability didn’t include you. Really, telephones and plumbing are quite useful to more or less everyone!

Breadth of Impact

The challenge for you is to find cloud references that are meaningful for your company. It’s tempting to focus on companies in the same industry, but that only gives you a piece of the picture: it really tells you who’s gotten out ahead of you and may have a competitive edge because they were taking advantage of the cloud first.

Instead, you might want to start by identifying the questions that are most important to you and then finding references across industries that are relevant. For example, what are the specific risks you see in your cloud journey? Are you worried about compliance with a particular regulatory framework? If so, what companies using the cloud have the same compliance requirement? Are you worried about data sovereignty? What other companies would have a similar worry and how are they thinking about it in the cloud? Do you have extremely high transaction throughput? Who else does, and is the cloud beneficial for them? Are you worried about being able to establish financial controls in the cloud? Well, then, is there a company that does a particularly good job with that? How do they do it?

What are your goals? Agility? Responsiveness? A great example of a company that achieved both of these benefits is McDonald’s. Upon realizing that its customer preferences had changed (they wanted home delivery and primarily used mobile applications) and that the company faced a special challenge (the need to localize everything they did for the vast number of geographies they served), McDonald’s was able to—in just four months—develop and deploy a mobile home delivery app across those geographies. Even if you’re in a vastly different industry, if your goal is to be able to respond quickly to market changes, McDonalds is a powerful example.

The Burden of Proof: Comparing Sniffs

Before we stop sniffing, we should consider the case for maintaining the status quo—the case that you can do better in your own datacenter. That you can be more cost-effective and competitive, more secure, and more prone to grow revenues by spending time racking and stacking and wiring and powering and air-conditioning your own infrastructure. To maintain the status quo, you’d have to believe that:

  • You can gain the same economies of scale that we pass on to you
  • You can operate the infrastructure with the level of reliability that AWS can
  • You can secure your infrastructure as well as AWS, with its large investment, custom-designed hardware and software to increase security, and ground-up security design
  • Your infrastructure needs will not change over the next four to five years, despite the rapid change in the digital economy
  • You will not need access to the advanced technologies we offer as services today and the new ones we will offer over the next four to five years
  • You don’t mind investing heavily in your innovative ideas before they have proven themselves

Does that pass your sniff test?


1 Incidentally, we’re also able to pass on tremendous economies in environmental sustainability. It has been estimated that moving a workload into AWS on average reduces its carbon footprint by 88% thanks to our better capacity utilization, more efficient infrastructure, and use of renewable energy sources.


3 Werner Vogel’s 2020 re:Invent presentation

4 Note the difference between these services and traditional off-the-shelf software. AWS services are building blocks that you use to construct your own systems—you’re not constrained to what an off-the-shelf product can do. And you only pay for what you use, when you use it. When you buy a software product off the shelf, you’re once again paying for capacity you aren’t using in the form of features that don’t matter to you. Again, it makes sense that you save money when you don’t pay for capacity you aren’t using.

Mark Schwartz

Mark Schwartz

Mark Schwartz is an Enterprise Strategist at Amazon Web Services and the author of The Art of Business Value and A Seat at the Table: IT Leadership in the Age of Agility. Before joining AWS he was the CIO of US Citizenship and Immigration Service (part of the Department of Homeland Security), CIO of Intrax, and CEO of Auctiva. He has an MBA from Wharton, a BS in Computer Science from Yale, and an MA in Philosophy from Yale.