Tickets per week per agent
Cross this line and your single-agent CX is already underwater. Most brands wait until 280 to hire. That is the expensive mistake.
Almost every DTC brand we work with hires their second CX agent three months too late. It is not because the founders are slow or stingy. It is because the warning signs are too quiet to act on, and by the time they get loud, the damage is already in your refund rate, your repeat purchase rate, and your Trustpilot score.
There is a real number that tells you when to move. Once you know it, the decision stops being a judgment call and starts being arithmetic.
The number is 200
A single competent CX agent on Gorgias, Helpscout, Zendesk, or Intercom can handle around 200 tickets per week sustainably. Sustainably means they are still hitting your SLA, still writing personal responses, still routing escalations cleanly, and still taking lunch breaks.
At 250 tickets per week per agent, response time starts to slip. At 280, the agent is copy-pasting more, personalizing less, and missing the nuance on escalations. At 320, you are losing customers you do not see leaving, because they just stop opening your emails.
If you have one agent handling all your tickets and your weekly volume has been above 200 for three consecutive weeks, you are already late on the second hire. Not getting-ready-to-hire late. Already-losing-money late.
Why brands wait too long
Three reasons, in order of how often I see them.
Reason 1: the agent stops complaining. A new CX agent at 150 tickets per week tells you when they are stressed. A six-month CX agent at 280 tickets per week stops telling you. They have internalized that “this is just the job” and they keep moving. The silence is the problem. By the time the agent finally says something, they are already drafting their resignation.
Reason 2: the metrics look fine until they do not. Response time is the most-watched CX metric and it is also the slowest to break. A team can hold 4-hour response time on 250 tickets per week for a while, just by working harder. CSAT lags response time by 4 to 8 weeks. Refund rate lags CSAT by another 4 to 8 weeks. By the time refund rate moves, you have been understaffed for three months.
Reason 3: the math feels wrong. “We are at 220 tickets per week, surely we are not at 2x capacity yet.” You are not at 2x capacity. You are at 110% capacity, which is the worst place to be, because it is just enough room to fail slowly and just not enough room to actually breathe.
What it costs to wait three months
Three months understaffed at 280 tickets per week looks like this in a typical Shopify brand doing $200K to $400K MRR:
- Response time: 6h target, 14h actual
- CSAT: 4.6/5 baseline drops to 4.2/5
- Refund rate: 2.1% baseline rises to 3.4%
- Repeat purchase rate within 90 days: 24% drops to 19%
On $300K MRR, the lift in refund rate alone costs ~$3,900 a month. The drop in repeat purchase rate, over the next two months, is another $9K to $14K in deferred revenue. You delayed a $4K/month hire and it cost you $22K to $30K.
The cheap hire was always going to be the right hire. Waiting did not save money. It moved the same money from the payroll line to the refund line.
What to do when you hit 200
You have three real options, in order of how 7-figure brands actually solve this.
Option 1: hire a second CX agent. Adds capacity. Adds management load. You now have two W-2 employees you train, schedule, performance-review, and replace. Time-to-productive is 4 to 6 weeks.
Option 2: contract a freelancer through an agency. Faster. Variable quality. You give up control over who actually responds to your customers. Most agencies will not name the person responding to your tickets, which means the voice of your brand to your customers is whoever happened to be on shift.
Option 3: add a Pod. A Pod gives you the CX specialist plus the SOPs, the QA layer, the AI assistant pre-drafting responses, and the Pod Operations Lead (POL) running it. You do not hire, you do not train, you do not manage. The same POL also owns returns, marketplace ops, and content scheduling if you want, so the per-function cost drops fast as you add functions.
The right answer depends on where the rest of your operation is. If CX is the only thing you need help with, option 1 or 2 is fine. If CX is one of three or four functions hitting the wall at once, the Pod math wins.
How to know which one you are
A simple test. List the operational functions in your business that are running on either you, a single specialist, or a single agency. If the list is one item long, hire. If the list is three or more, you do not have a CX problem. You have an operations problem, and CX just happens to be the loudest symptom this month.