Erik Peterson, our CTO and Founder here at CloudZero, recently hosted a webinar-style interview with Matthew “MJ” Jackson, DevOps Lead at Drift.
This interview took a deep dive into the methods Drift used to take control of their cloud spending.
If you own, manage, or otherwise have any stake in a SaaS company that heavily utilizes the cloud, you’ll want to read on to see the highlights of this conversation. MJ drops some incredibly useful wisdom that might help you stop your own cloud cost snowball and bring your bill down to a manageable level.
Here’s How Drift Reduced Cloud Costs — And Encouraged Innovation — With CloudZero
1. Drift built a new culture of cost efficiency based on visibility
Peter Drucker famously said that “Culture eats strategy for breakfast.” Erik agrees—but would tweak that sentiment slightly to say “lunch and breakfast!” “It truly conquers all.” And he’s not wrong. If a company doesn’t have a culture in place that encourages teams to keep costs at the forefront of decision-making, that company is going to struggle to control costs even if their strategies are otherwise sound.
But just as a bad culture can take the whole ship down, a good culture can change everything for the better. Building for cost efficiency should ideally be a top priority for engineers; eventually it should become second-nature, something that engineers take into consideration without having to consciously think about it.
Another of Erik’s favorite phrases is: “Every engineering decision is a buying decision.” He makes an effort to hammer this point home since it provides the foundation for the most impactful cost optimization efforts.
Drift’s leaders were well acquainted with these principles. They knew that to fine-tune their cloud costs, they must get engineers to take charge of cost optimization. The tricky part was how to make that happen.
Erik says: “A lot of people ask me, ‘How can I get people interested in this stuff?’ and I say, ‘Well, have you given them the ability to see the effects of their actions?’”
In other words, Drift’s first objective was to improve cost visibility and provide its engineering team with the data they needed to make good buying decisions. Given enough data and time, this would shift the organization’s entire culture when it comes to cloud costs.
This is a principle known as shifting left. MJ shares the details of how this works at Drift:
“All of our engineering teams are talking to their leadership on a regular cadence—a couple weeks, once a month—about what they’re building, the systems they own, how they’re running, what their usage metrics are, and what their error rates are. Having them just mention their costs at that level and at that cadence is crucial because it makes it easy for them later to target that cost.”
2. Drift’s engineers aim to control AI costs by making educated building decisions
Erik points out that companies are currently proud of how much money they funnel into AI. But at some point, he theorizes, there will be a sort of “AI apocalypse” concerning high costs.
Drift leans heavily on AI and automation to stay at the cutting edge of their field. And because AI is such a costly technology, Drift’s monthly cloud costs are a direct reflection of its engineers’ choices.
MJ credits much of their success in lowering AI costs to the cultural shift he discussed earlier:
“The concept that engineers like optimizing things — if you just tell them about cost they’ll want to make it go down — is a great one.” Later, he adds: “If I could go back and tell myself ten years ago that engineers would think like that, I would just frankly not believe it.”
3. Using this cultural shift as a solid foundation, Drift found a strategy to prevent monthly sticker shock before it happens
“You have to start thinking about cost before it matters,” MJ says. “You need to have an awareness of your constraints before they’re binding constraints, or when they bind they’re very uncomfortable surprises.”
“Nobody likes requirements at the end of the project,” Erik agrees. “The ship has sailed!”
The key, therefore, is for Drift’s engineers to understand the costs they’re generating in as close to real-time as possible, when they still have the opportunity to change course. If they were to wait for the end of the month and receive the bill for their cloud costs, those costs would already be fixed in stone.
“The biggest way [CloudZero] has changed things on a fairly fundamental level is that our engineers are more conscious and they build better architectures,” MJ says. “Because whenever they’re proposing a new [project], there’s an idea that it will cost about this much, at about this scale, and this is what will be driving that cost.”
Since engineers can predict with reasonable accuracy how each infrastructure decision they make will affect costs, they can make the smartest choice possible at every given fork in the road along the way.
This includes even the most broad decisions that can change the course of an entire project, such as whether to build a solution in-house or purchase a ready-made solution off the shelf. “When you think about cost all the time, the ‘build or buy’ decision gets a lot more clear-cut for people,” MJ explains.
“When you think about a machine as having a fixed, predictable cost, and you think about how much time you spend supporting similar machines, you’ll know, ‘All right, hosting it ourselves will cost money.’ It will cost real money, and you can quantify that.”
In the course of this discussion, MJ explains the value of knowing what your spend will be before you spend it — especially in the average organization where Finance is in a constant budgetary battle with Engineering.
If this is something your company struggles with, we highly recommend listening to him divulge some of his hard-earned wisdom:
4. Drift was able to buy better, build better, and sell better using in-depth cost data
For Drift, every engineering decision certainly was a buying decision, and those buying decisions became more cost-efficient and productive as the data rolled in. However, smarter infrastructure improvements were just the tip of the iceberg.
Listen to Erik and MJ explain just how much impact good cost visibility can have:
Ready To Start Shifting Your Company’s Culture And Widening Your Margins?
As MJ and Erik explained above, good cost management comes down to the availability of data. If your engineers don’t have the data to make good buying decisions, you’ll never be able to fully control cloud costs.
In practical terms, this means your first step should be providing your engineers with a way to see the effects of their purchasing decisions and to anticipate cost spikes before they occur.
Wondering how to make that happen quickly and efficiently?
to see how you can start getting valuable data insights within the first 24 hours of using our platform.
The Cloud Cost Playbook
The step-by-step guide to cost maturity
