Scaling Efficiently

Scaling is more than just engineering. What companies often neglect in regards to “scale” is the human aspect.

By Alex Exarhos, December 9, 2021

The term “scale” is tossed around liberally in the tech world. The general idea is simple: if you’ve built the infrastructure and tools to make a $1 margin on your product, you should be able to use the same infrastructure and tools to sell to a million people (or more) and make a million dollars. In practice, it’s never simple, even with an entirely web-based product like Ellacard.

Scaling is more than just engineering. Yes, of course your technical infrastructure should be set up such that you can simply add more servers to handle increased traffic. There is a ton of technology to support this on the engineering side: containers, scalable databases, and all sorts of other cloud services. We definitely use those.

But what companies often neglect in regards to “scale” is the human aspect. You can build the world’s most intuitive and accessible interface, invest countless hours in testing and maximizing device compatibility, but your servers will still occasionally run into problems, and you will still get direct inquiries from customers with questions you haven’t thought of. These problems are compounded at start-ups in general, since the product is constantly changing with new features left and right. There’s a popular mantra in the start-up world, “move fast and break things,” which might be good for helping you refine your product, but puts an incredible burden on your engineering and customer support teams, and definitely contributes to increased rates of burnout and stress.

Ellacard is entirely bootstrapped and does not have the luxury of having any kind of large engineering or customer service teams. Furthermore, our product is consumer-facing and by it’s nature needs to deliver a great customer experience (eg. we can’t afford to disappoint our customers by not sending their card on time). But we’ve still managed to sustainably scale our product to support hundreds of thousands of monthly users by making sure we can scale our human support resources, not just our technical ones.

Engineering On-call

It’s standard practice in the software world to have rotating engineers dedicated to “on-call” duties — any time there’s a server error, those engineers get notified to investigate the problem, mitigate any customer impacts, and push out a fix. We do have extremely sophisticated logging / alarming tools that make it as quick and easy as possible to discover and investigate any server errors.

But our perpetual goal is to get the number of customer-impacting weekly issues down to zero. If there’s a server issue that impacts just one customer, not only do they have a bad experience (possibly losing a sale), but there’s also a decent chance they’ll send us a support email as well. In the best case, the technical issue might be easy to fix, maybe 15 minutes of work, and then 10–15 minutes spent corresponding with the customer. In an average case, we could be looking at an hour or two. Per customer.

Any policy of allowing even a minimal amount of recurring customer-impacting issues breaks down at scale. Even if the issue only impacts one out of a thousand customers, we have hundreds of thousands of customers per month, so that’s hundreds of hours spent resolving issues that could have been better spent working on other aspects of the business. We’ve done the math: spending a few more hours up front to make sure we don’t have issues takes way less time than trying to operate with some minimum allowable tolerance of customer-impacting code issues.

Customer support

Even if there are no server issues / bugs, it’s inevitable that some customers will still email our support line. We have to be able to address those inquiries quickly and correctly. That’s why we’ve invested time and money in creating tutorial videos and a comprehensive FAQ to which we can point those customers, and it’s also why we prioritized building a custom set of administration tools, designed specifically to help resolve customer questions.

We’ve found that most companies tend to skimp on internal administration tools, but these tools are the backbone of being able to scale efficiently. For example, our internal user management dashboard lets us see a user’s activity when they email in with a question, so we can quickly figure out exactly what the problem is. We have a dedicated internal dashboard to help manage our gift cards. These dashboards are even mobile-friendly, and they make it possible to resolve most customer issues within a few minutes, from anywhere.

We’ve found that it is absolutely worth the time to invest in these aspects of scaling that minimize human involvement in day-to-day operation of the business. In addition to allowing us to scale incredibly efficiently, having good internal administration tools and a zero-tolerance policy for ongoing customer-impacting bugs also reduces the stress and headaches of support. We’re absolutely willing to be on-call 24 / 7 when we know our code is reliable and we have the tools to resolve customer inquiries quickly and pleasantly. So send us an email if you need anything: support@ellacard.com