An $8M company ran its entire customer service operation on a stack you could set up this afternoon.
There are more e-commerce and operations tools on the market than any team can reasonably evaluate. At Smoke Cartel, we stuck with simple wherever simple existed, and most of that stack still runs the business today, years after I left.
The whole stack
- Shopify for commerce
- HelpScout for email support
- Klaviyo for marketing email (a different job from support, even though both live in an inbox)
- Slack for everything internal, plus text-based customer contact
- Nextiva for phones, with a simple PIN menu so calls landed with the right department instead of a receptionist playing traffic cop
- Signifyd for fraud
- ShipSurance for damage and loss claims
Fraud tooling isn't free, by the way. One real Signifyd invoice came to $1,121.51 for a batch of 3,973 guaranteed transactions. It's worth knowing that before you assume fraud protection is a rounding error. It earns its cost on the one order it catches that you never would have known to check.
Buy the standard, build the specific
We never wrote it down as a rule, but this is how we decided: buy the pieces every commerce business needs, and build the piece that's specific to how you watch your own business.
No off-the-shelf rewards platform did what we wanted, so our loyalty points ran on a homemade internal API that let a support agent adjust a customer's balance from the same tool they were already using to talk to them. Instead of buying a business intelligence platform, we put one computer in the main office with the screen never dark, cycling through the numbers that mattered: that day's revenue, shipments coming in, shipments going out, and the customer backlog. Nobody had to open a report to know how the day was going. They just looked up.
More software is not more capacity
After Smoke Cartel, I advised a company running a sales team of about ten and a customer service team of five on more than $250,000 a year in software. The centerpiece was a popular, expensive CRM, customized with more fields, workflows, and options than anyone could hold in their head. So the team worked around it instead of in it, while the company kept paying for it.
The fix wasn't more training or more customization. It was downgrading to something smaller that the people using it every day could actually run.
A tool only counts as capacity once your team can use it without a workaround. The most customizable option on the market is often the one most likely to outrun a small team.
Sometimes the answer is a floor fan
Savannah warehouses in the summer are hot, no matter how many fans and water stations you add. So when I saw an ad for a bracelet that cooled your wrist pulse points and promised to make you feel up to three degrees cooler, I ordered two to test and emailed the company about a case study for our whole warehouse team.
They got used maybe twice, and I'm fairly sure the second time was out of pity for me. I'd never asked whether a wearable would get in the way of people whose hands and wrists move all day. I went back to Lowe's and bought two more industrial floor fans. They never needed charging, testing, or a single email. 😅
The rhythm matters more than the tools
A tool stack is only as good as the routine you run it on. Ours was simple enough to explain in a minute:
- Monday reset: clear the weekend inbox, close out last week, send partner emails, and set your week's direction before it sets you.
- Standups in Slack: everyone posts blockers at the start of the day where the whole team can see them and jump in, instead of saving them for a meeting.
- Monthly KPIs as "3 slides or 3 minutes": every department, same lightweight format, posted in a shared channel. No meeting unless the numbers call for one. If a number can't be explained in three slides, it usually isn't the number that matters.
The handful of numbers worth watching
- First response time: email within 2 hours (4 max), chat within 5 minutes (30 max). When these slip, it's usually coverage thinning out, not agents getting slower.
- Chargeback rate: under 1% of monthly volume. Track it weekly, not quarterly.
- Review and question backlog: we once found over 3,000 unanswered reviews and product questions on our site. That backlog costs you SEO and tells every future customer nobody's home. It's also the easiest number to fix.
- Customer happiness: aim for 90%+, but never judge an individual agent on it. A falling score almost always means something broke upstream, not in the support inbox.
A 30-minute tool audit
- List every tool you pay for and its monthly cost.
- Next to each one, write who actually uses it every day.
- Anything nobody uses daily gets downgraded or cut.
- Anything your team works around gets simplified before you pay for more training.
- Pick three to five numbers and put them somewhere nobody can ignore.
Want the book when it's ready?
This post is adapted from two chapters of The Team You Build, the book I'm writing about building a team that will outlast you. Drop your email below and I'll let you know when it's out, along with new posts on operations, e-commerce, and building teams.
Follow me on Instagram: @darbycox
Need help with your own operations? Book a call with me.