I was at breakfast with a fellow founder this week. They’re doing well – good growth, interesting product, maybe even something that you can call product-market fit. And yet, they were unhappy with how things were going.
They went from happily building an incredible product to spending up to 4 hours per day in emails solving customer queries. Pressure from the user base was mounting, and in return this collapsed any innovation that could be done on the product.
It’s not a story I hear for the first time. Just two weeks ago, another founder shared the same story over coffee. Two months ago I heard the exact same thing meeting a founder in Japan over lunch. Their companies have so little in common, basically only the fact that they have an SMB user base – more customers paying less per account than in enterprise – and they were all fairly early-stage, growing quickly.
It made me think how come I don’t relate to these issues? And it made me recall the journey we’ve taken with customer support at MentorCruise, bringing it to a place where our support org handles thousands of inquiries every month with ease.
How I spent 50% of my revenue on my first employee
It was summer of 2020, and my company had just experienced its very first growth spurt. I had been building and refining for the better part of the past two years, and due to unforeseen global circumstances, revenue had suddenly scaled 400% in just a couple of months.
It’s not that customer support started to grow beyond what I could handle, but holding down a full-time job and wanting to develop the product further, I definitely had other priorities stacked higher. So I took a leap, and hired my first employee – customer support.
At the time, this was painful. What founders often don’t consider is how much of the history, context, and knowledge of a product really just lives within the containment of your own brain. And when a second person gets added to the team, all of this needs to be shared.
So our workflow for the first few months was as follows:
-
A customer would send in an email
-
My customer supporter would consult docs and a test version of our product trying to answer it
-
If they could, they’d write it down
-
If they couldn’t, they’d ask me
-
I would then tell them how, and document our approach, write a policy, or similar
In the first week I still replied to 90% of the emails. Then a few less, then a few less, until I was only consulted on edge cases and company-defining cases.
The three-level system
When I was first studying software engineering, our school put a mandatory customer support course in our curriculum: “three-level support”. It was essentially free credits, and a course I thought I’d be able to forget the second I walked out of it.
And yet, as our team grew, I found myself being drawn back to it.
The system is pretty simple. A support request can be escalated up to three levels. The first level is general support. It’s all those inquiries that can be answered by anyone reading the help articles or docs, but just choose not to. The first level can often be fully replaced by AI these days – but I’m going to talk about this later. New supporters can be onboarded at this level in a day or two.
Second-level support is technical support. This requires access to your systems and billing. For most companies, this is going to be the same person as in first-level support, but in some cases this might be tickets getting through the AI system, or a more senior supporter taking over a ticket from a new hire. This takes a bit of training and trust with access to your systems, as they’ll solve any issues that require system actions – blocked users, refunds, data inconsistencies, etc.
The third, and last level of support, is the expert support. At the early stages, this is likely you – as the founder. Later on this might be a marketer, developer, or admin. At this point, something is very wrong and requires your action. Maybe it’s as simple as a small bug or hole in your refund policy. Sometimes it’s something that needs a couple of days to be ironed out. That’s why the expert needs to take over.
The approach to consult me whenever something was unclear or needed further action didn’t scale, and so we adopted the three-level support system with the first two levels covered by the customer supporter.
Expectations are growing, and you’re doing worse
Even though it feels like a fairly static topic, customer support has evolved a lot since my first dive into it in 2020. Recent examples like fully configured AI agents come to mind, but the changes have been happening for the better part of the past decade.
First, my experiences with customer support only cover the email route. In the last couple of years, however, more customers have come to expect a live chat to be present, and answers to be almost instantaneous.
AI can take some of that heat off you, but all founders I’ve talked with who have explored such a system have had to backtrack immediately. If you imagine it’s hard for another human to have context and give accurate responses, try an AI agent!
Having gone through many of those developments and experiments myself, here are my three learnings:
- Use a customer support tool that gives you an in-app help widget, but sends all communications as long-form email as opposed to live chat. It makes customers give you longer messages, more context, and shifts response-time expectations from “a few minutes” to “up to a day”.
- Only use AI in first-level support and when absolutely necessary to get through the sheer volume. Using AI early in your customer support journey will make your company have weaker structures, worse documentation, and more load on second-level support, which will now receive all the “I want to talk to a human” messages.
- Consider tagging your customers to make sure that your most valuable customers – enterprise plans, big teams, special contracts – receive preferential treatment. Too many times have I seen consumers and prosumers fill up support queues, hurting the big accounts in the process.
That does come with downsides. The lack of real-time support will end up costing you customers, but they’re generally not the ones you want to have more of anyway.
A word on hiring
Having had north of 30 people join our company over the years, I will write a longer article on hiring one of these days. But I like to live by three principles:
- Take a lot of effort to hire the right person
- Fire fast if things don’t work out
- Do everything to retain a good employee
Customer support is an under-appreciated role in most companies. They’re the easy jobs that’ll get outsourced at a moment’s notice. I disagree strongly, and believe that good customer support is one of the big retention and revenue levers you have.
Additionally, customer supporters age like fine wine. A customer supporter that has been in your team for years is a bit like a digital clone of yourself. A historian you absolutely need to keep in your team.
For those folks who’d rather hire cheap or give that job to AI, I’m never confused why their churn usually increases just days after as well. So even if you don’t usually hire like I do, consider doing so for this one role.