Most support agencies pick a pricing model in their first month and never revisit it. That decision quietly determines your margin, which clients you attract, and whether growth makes you more profitable or just busier.
There are four models in common use. None is universally right, but each one fails in a specific and predictable way, and knowing which failure you're signing up for is most of the decision.
The four models
1. Per ticket
You charge a fixed amount for each ticket resolved, commonly somewhere between $2 and $8 depending on complexity and region.
Where it works: high-volume, low-complexity work. Ecommerce order status, password resets, shipping questions. The client understands exactly what they're buying and can forecast cost against their order volume.
Where it fails: the incentives point the wrong way. You're paid per ticket closed, so you profit from volume and from speed, not from resolution quality. A client whose product improves sends fewer tickets and your revenue falls. You are financially rewarded when their product is worse. Sophisticated clients notice this eventually.
It also creates a definitional argument you will have repeatedly: is a customer who replies three times one ticket or four?
2. Per hour
You bill for time worked, typically at an agreed rate per agent hour.
Where it works: complex or unpredictable support, technical products, and early engagements where nobody knows the volume yet. It's the fairest model when the work genuinely varies.
Where it fails: it caps your income at hours available, so growth requires hiring in lockstep with revenue. It also punishes efficiency: every process improvement, macro and knowledge-base article you build reduces your own invoice. And it requires accurate time tracking per client, which agencies consistently underestimate. If you're not logging time against tickets, you're guessing at your own margin.
3. Per agent (dedicated staffing)
The client pays a monthly rate for a named agent or a fraction of one. Common rates run from roughly $1,200 to $3,500 per full-time agent per month depending on region, language and skill.
Where it works: clients who want continuity and product familiarity, someone who learns their catalogue, their tone, their edge cases. It's the easiest model to sell to mid-size clients because it maps onto how they'd think about hiring.
Where it fails: utilisation. If a dedicated agent is only busy 60% of the time, that idle 40% is your loss, not the client's. Agencies solve this by quietly sharing agents across clients, which works until a client asks why "their" agent takes four hours to respond. It also concentrates risk: losing one client means an agent with nothing to do and a payroll obligation that doesn't pause.
4. Flat monthly retainer
A fixed monthly fee for an agreed scope: a volume band, a channel set, defined hours of coverage.
Where it works: this is where most agencies end up, and for good reason. Revenue is predictable, which makes hiring plannable. Clients like knowing the number. Your incentives finally align with the client's: you profit by resolving efficiently and by reducing repeat contacts, which is exactly what they want.
Where it fails: scope creep. Without a defined volume band and a stated overage rate, a client whose volume doubles pays the same and your margin evaporates. The fix is a tier structure with explicit bands and a documented rate for exceeding them, reviewed quarterly rather than annually.
The common trajectory: start hourly with your first client or two because you have no idea what the work involves, then move to retainers once you can predict volume. Agencies that stay hourly past their fourth client usually do so because they never started measuring, not because hourly suits them.
Comparing the models
| Model | Revenue predictability | Incentive alignment | Main risk |
|---|---|---|---|
| Per ticket | Low | Poor | You profit from volume, not quality |
| Per hour | Medium | Poor | Efficiency reduces your own revenue |
| Per agent | High | Neutral | Idle capacity is your cost |
| Flat retainer | High | Good | Scope creep without volume bands |
Knowing your actual cost per ticket
Whichever model you use, you cannot price properly without knowing what a ticket costs you to handle. The calculation is simple and most agencies have never run it.
Take one month. Add up fully loaded agent cost for the client: salary, taxes, benefits, plus a share of management time. Add tooling. Add an allocation of overhead. Divide by tickets handled for that client.
The number is usually higher than people expect, and it varies enormously between clients. The same 400 tickets a month can cost you twice as much for a technical B2B client as for an ecommerce one, because handling time differs. If you're charging both the same, one of them is subsidising the other and you probably don't know which.
This is the argument for per-client reporting that includes time, not just volume. Without it you have an agency-wide margin and no idea which client is dragging it down.
The cost most agencies get wrong
Tooling is usually treated as fixed overhead. For support agencies it isn't: it scales with the thing that generates your revenue.
Most helpdesk platforms charge per agent seat. Zendesk Suite Growth, for reference, lists at $89 per agent per month billed annually as of mid-2026. That means every hire increases your tool bill before the new client's first invoice clears, and it means your software cost tracks headcount rather than client count.
Work through a realistic example. An eight-agent agency serving six clients on a per-seat helpdesk pays roughly $700 a month in tooling. Spread across six clients that's about $117 per client per month, straight off the top of each retainer. Win two more clients and hire three agents to cover them, and tooling rises to roughly $980. You've grown revenue, but a chunk of the increase went to your software vendor rather than your margin.
Two practical consequences. First, seat cost should appear in your per-client margin calculation, not sit in general overhead where it's invisible. Second, if you're paying per seat, every pricing decision you make carries a variable cost you don't control.
The alternative is tooling priced per agency rather than per agent, which makes software a genuinely fixed cost: hiring five more agents changes nothing. That's how Boridesk is priced, flat from $59 a month with unlimited agents on every plan, which we'll admit is not a neutral thing for us to point out. The general principle stands regardless of vendor: if your tool cost scales with headcount and your revenue scales with clients, those two curves will diverge.
Five rules that hold across models
- Charge a setup fee. Onboarding a client (learning the product, writing macros, configuring routing) is real work delivered before any recurring revenue arrives. A one-off fee covers it and filters out clients who aren't serious.
- Define scope in writing, including volume. "Customer support" is not a scope. Channels, hours of coverage, languages, expected monthly volume and the overage rate are.
- Price the second client higher than the first. Your first client is paying partly for the privilege of teaching you. Do not carry that rate forward for three years.
- Review pricing quarterly, not annually. Client volume changes faster than contracts do. A quarterly check-in against the agreed band makes the conversation routine rather than confrontational.
- Send a monthly report whether or not they ask. Support done well becomes invisible, and invisible services get cut at renewal. A monthly per-client report showing volume, response times and satisfaction is the cheapest retention tool available to you.
Where to start
If you're pricing your first client, go hourly for the first two or three months with an explicit note that you'll move to a retainer once volume is clear. Track everything. Then convert to a flat retainer with defined bands based on what you actually measured, not what you guessed.
If you already have clients and no idea which ones are profitable, start by calculating cost per ticket per client for last month. That single number tends to reorganise everything else, including, frequently, the discovery that your largest client is your least profitable one.
Tooling that doesn't scale with your headcount
Boridesk charges flat per agency, not per agent. Hire freely, add clients as you grow, and keep the pricing conversation about your service rather than your software bill.
Start free →