Cost pressures are coming, and it’s not clear whether most companies’ cloud programs are ready. With the macroeconomic environment becoming increasingly challenging and company leaders looking for ways to achieve higher business resiliency, CIOs and CTOs can expect uncomfortable questions about the costs of their cloud programs.
While cloud is able to generate substantial value when done right, many companies we’ve observed have seen their cloud spend grow as much as 20 to 30 percent each year. Without being more sensitive to costs and responsive to the economic pressures that companies are feeling, CIOs may soon find their cloud programs on the chopping block.
That would be an enormous pity and a huge setback for most companies’ competitive aspirations. Through a targeted set of practices, however, technology leaders can quickly cut as much as 15 to 25 percent of the costs of their cloud programs while preserving their value-generating capabilities.
Following are five things to help tech leaders get on that path.
Would you like to learn more about Cloud by McKinsey?
1. Stop unhealthy growth
Cloud cost increases can reflect healthy growth, such as growth in the user base, increased digital adoption, and the development of new digital capabilities. In many cases, however, those same cost increases also hide “unhealthy” growth due to poor stewardship, such as provisioning more resources than required, or immature consumption practices, such as forgetting to shut down instances that are no longer used.
Companies often don’t have a clear perspective on these healthy versus unhealthy costs. This can be particularly detrimental if customers’ spending habits shift, as they do during recessions, leaving companies uncertain about how to adjust their own cloud spend.
To ensure the transparency of cloud spend, companies should set up a consistent, high-quality, and comprehensive tagging and reporting capability (often automated) and put in place an allocation model that promotes accountability, such as charging business leaders, or at least making them aware of, the costs of their products or services that use the cloud. Companies should also introduce financial controls, such as actively tracked and managed budgets for the individual product teams that allow for healthy growth while ensuring that cloud spend is going toward business priorities and critical use cases.
2. Focus on simple fixes (there’s often a lot of value there)
When helping organizations manage their cloud programs, we have consistently uncovered a broad range of cost- and performance-optimization opportunities. The good news is that, in many cases, companies can capture these productivity improvements by taking relatively simple steps. The most common no-regret actions include releasing unused capacity, introducing scheduling and auto-scaling features, and aligning service levels to specific application requirements (for example, switching from memory-optimized to standard instances or using a serverless compute engine for containers instead of managing their own cluster). Companies should prioritize the ones that offer maximum benefit and deploy them rapidly across teams and cloud users while doing quick feedback loops—if the lever is successful for one app or team, it can then be scaled to the others.
One major public-sector agency, for example, was able to achieve around 20 percent savings by adjusting cloud services to better fit application needs, getting rid of assets it was paying for but no longer using, implementing some basic guidelines for tiering storage, and updating the instances to the latest version.
Once a baseline level of optimization is achieved, companies can sustain the results by training technologists on cloud best practices so they understand which actions they need to take to lower costs—and then empower them to do so, mandate that the FinOps team continuously scan for new cost-reduction opportunities, and track the results of optimization efforts.
Cloud by McKinsey Insights
3. Unlock cloud elasticity to stop paying for unused cloud capacity
In theory, cloud elasticity—the ability of cloud to scale up and down to meet a company’s in-the-moment needs, should lower costs because it enables companies to pay only for what they actually use. But many companies have multiple practices in place that keep them from using cloud’s elasticity effectively, such as rigid and often manual provisioning practices, technical debt that makes it impossible to build in elasticity features, and excessive use of reserved instances. As a result, companies are paying for cloud capacity that they’re not using.
A global telecom operator faced this issue due to the predominantly manual way the cloud infrastructure was scaled. While teams adjusted capacity up to respond to increases in traffic, they rarely readjusted down when demand was waning. This issue can have significant consequences during a recession, when customer demand weakens and businesses cut back on promotions or institute discounts, all of which can have a large impact on traffic and cloud usage.
Companies should work with their engineering team to identify inelastic applications and workloads—the usual suspects are those that were simply “lifted and shifted” to the cloud—and refactor them, starting with the ones with the largest footprint. In many cases, it’s relatively simple to set up standard autoscaling features. Investment in more advanced capabilities, such as containerizing workloads, can lead to even more efficient elasticity, but these must be carefully considered, as they often require more time and effort. In general, companies should avoid “lift and shift” in future migrations, unless there are strategic reasons for it, such as an exit from a data center.
4. Take another look at vendor agreements
It’s often the case that an organization was too optimistic about the expected pace of its cloud migration and is now stuck with spending commitments that can be hard to meet in an economic downturn. Many organizations fail to initiate renegotiation until 12 to 18 months before their contract expires, by which point it’s often too late to negotiate effectively.
In reviewing vendor contracts, companies should ask if they would sign the same contract today. If the answer is no, they should try to renegotiate. Companies are sometimes able to negotiate trade-offs—for example, deeper discounts for reduced flexibility or pushing back timetables on meeting set targets. Companies that are approaching or in the process of contract renewals should consider incorporating provisions that would enhance flexibility, such as being able to step commitment levels up or down based on agreed trigger events, or revising the parameters of the services credit to be able to use it all in a single year or to spread it out over multiple years.
5. Don’t stop cloud migrations; just be smarter about them
A common misconception is that organizations can reduce costs by slowing down cloud migrations and working within their on-premises environments, which they’ve already paid for. However, compared to cloud environments, on-premises data centers require continuous operating support in the form of labor, utilities, leases, and licenses to maintain systems, manage refresh cycles, and combat outages. Cutting costs in any of these areas can also lead to expensive issues. Furthermore, thoughtful and targeted cloud migrations not only help lower costs but also position the business to grow more quickly once the downturn has passed.
Organizations should prioritize migrating workloads to the cloud that generate value, such as those that enable critical business initiatives (for example, customer-support automation), use hardware that will shortly need to be replaced or upgraded, or have sizable operations overhead.
Organizations have adopted cloud thanks to its promise of increased flexibility and performance at a fraction of traditional IT costs. The current challenging economic environment puts this promise to the test. To weather it successfully, companies can make clear and simple moves to cut costs, build value, and improve productivity.
