Help Desk Software Cost per Agent

The pricing page says $45 per agent per month, and for a five-person support team that reads like $225. Help desk software is rarely $225. Ticketing platforms price the seat, then price the channel, then price the assistant that helps the seat, and each of those lines multiplies by the same agent count. The result is a bill that moves in steps rather than a curve: one more hire is a whole seat, one missing feature is a whole tier, and one switched-on channel is a whole new per-agent charge across everyone.

This verdict prices help desk and ticketing software the way a support lead actually has to price it, from the seat model outward to the lines that only appear after go-live. It takes apart per-agent pricing and the step behaviour it produces, separates full agents from light agents and named seats from concurrent ones, maps which features tend to sit behind which tier gate, and puts illustrative numbers on chat, voice minutes, AI assist, extra brands, knowledge base, implementation, and migrating ticket history. It sits alongside our verdict on the true cost of business software, which prices per-seat models generally, and our payroll software cost verdict, which prices the other category where headcount drives the invoice directly. Run your own agent count through the true-cost calculator and the companion on this page before you compare a single quote.

What help desk software actually costs

Ask what a help desk costs and the useful answer has four parts rather than one. There is the agent seat, the recurring per-person charge that the pricing page advertises. There is the seat mix, because most desks contain people who work tickets full time and people who only ever add a note, and those two are priced very differently. There is the channel layer, where email is included and everything else is bought. And there is the one-time layer of implementation, configuration, and moving your ticket history, which lands once and is almost never quoted until you ask for it.

Only the first is predictable from a pricing page. The second depends on how you classify your own people, which nobody decides until they are inside the tool. The third depends on how customers reach you, which is a business fact rather than a software preference. The fourth depends on how much history you have and how much of it you insist on carrying forward.

Hold that four-part frame and the rest of this verdict takes each part apart with illustrative planning figures you can substitute your own into. Every number here is a planning band built to show how the pricing model behaves, not a quote from any vendor, and support software pricing moves often enough that a figure typical today may read differently in a year.

The headline: illustrative per-agent pricing by tier

Start with the numbers people search for, framed as bands rather than quotes, because help desk pricing varies by vendor, region, contract length, and feature mix.

Light agent or collaborator seats commonly sit around $10 per person per month, and some vendors include a handful free on upper plans. These accounts read tickets and add internal notes but cannot own or answer a customer conversation.

Basic ticketing commonly lands around $20 per agent per month. Email arrives as a ticket, tickets get assigned, and a simple view shows what is open. Automation is thin, reporting is a volume count, and there is usually no customer-facing portal.

Growth or standard plans commonly land around $45 per agent per month, which is the tier most small support teams actually buy. Expect routing rules, basic automation, a knowledge base, canned replies, and reporting that answers more than “how many”.

Professional plans commonly land around $75 per agent per month and are where the features teams keep asking for tend to live: real SLA management with breach alerts, a branded customer portal, custom reporting, multiple brands or mailboxes, and deeper permissioning.

Enterprise tiers commonly start around $115 per agent per month and climb, adding sandbox environments, advanced routing, audit logging, and contractual support commitments.

The chart is worth sitting with, because the ratio between the cheapest and the most expensive full seat is larger in this category than in most. A desk that outgrows basic ticketing does not pay a little more, it pays several times more, for everyone.

Why help desk software is priced per agent

Support tools are priced per person answering, not per person contacted, and the logic is structural rather than commercial. A ticketing platform gives each agent a workspace, a queue, a set of permissions, a signature, an activity trail, and a share of the reporting model. The unit of work is the agent, so the unit of pricing is the agent.

That has one pleasant consequence and one unpleasant one. The pleasant one is that your bill does not grow when your customers write more, which makes support software easier to forecast than most usage-priced tools. The unpleasant one is that the bill grows the moment you staff up, including seasonally, and seats bought for a busy quarter tend to survive into the quiet one.

There is also a quiet asymmetry worth naming. Ticket volume drives your staffing, staffing drives your seats, and seats drive your bill, so volume does reach your invoice, just with a lag and in steps. A desk that doubles its ticket load without hiring pays nothing extra and simply degrades. A desk that hires two people to absorb the same load pays two full seats plus every add-on those seats carry. As our verdict on the true cost of business software argues about per-unit pricing in general, the billing unit sets the shape of your cost curve, and in support the unit is a person you were going to hire anyway.

Why the bill moves in steps, not a curve

Most buyers picture software cost as a line that rises with usage. Per-agent support pricing produces a staircase instead, and knowing where the risers sit is most of the budgeting problem.

The first riser is the seat. There is no half licence, so a part-time weekend agent costs the same as a full-time one. Hiring a sixth agent onto an illustrative $45 plan with a $15 chat add-on adds $60 a month, or $720 a year, before that person has answered anything.

The second riser is the tier gate. When the feature you need sits one plan up, you do not buy the feature, you rebuy the team. On an illustrative $30 gap between a growth plan and a professional plan, a twelve-agent desk pays $360 a month, or $4,320 a year, to unlock a single capability. That is the calculation to run before anyone says “we just need SLA alerts”.

The third riser is the channel. Switching on voice does not add a phone line, it adds a per-agent charge across every seat plus usage on top. Three risers, all multiplied by headcount, and none of them visible on a page that shows one number.

Light agents, collaborators, and the cheap seat

The single largest avoidable overspend on a help desk is buying full seats for people who do not work tickets. Most platforms offer a reduced account, variously called a light agent, collaborator, or contributor, priced illustratively around $10 per month against $45 or more for a full seat, and some vendors include a limited number at no charge on upper plans.

The restriction is genuine and worth understanding before you assign anyone. A light agent typically cannot reply to a customer, own a ticket, change a status that matters, or appear in the assignment queue. They read, they comment internally, and they get notified. That covers engineers answering a technical question, an account manager adding context, or finance confirming a refund, and it does not cover anyone who occasionally clears the queue on a busy Friday.

The arithmetic is straightforward and often large. Ten light seats at $10 instead of ten full seats at $45 saves $350 a month, or $4,200 a year. Audit your user list against a single question: has this person replied to a customer in the last quarter. Everyone who has not is a candidate.

The trap runs the other way too. Misassigning a working agent to a light seat produces workaround behaviour, where a real reply gets pasted into a ticket by someone else, and the audit trail stops meaning anything. Save money on the seats that genuinely do not work tickets, not on the ones you wish did not.

Named seats versus concurrent seats

A smaller number of vendors sell concurrent seats rather than named ones, and the distinction changes the maths completely for a desk that runs shifts. A named seat belongs to one person permanently. A concurrent seat is a slot that any one agent can occupy at a time, so a team of twelve working three shifts might need only four concurrent licences.

Concurrent pricing is always higher per seat, illustratively around double a named seat, which is what makes the comparison interesting rather than obvious. On an illustrative $75 named seat against $150 concurrent, a twelve-agent desk pays $900 a month named. If those twelve people never exceed five simultaneous logins, five concurrent seats cost $750. If they routinely reach eight, concurrent costs $1,200 and loses badly.

The break-even is simple: concurrent wins when your peak simultaneous agents divided by your total agents is below the price ratio between the two models. On a double-price ratio, that means concurrent pays when fewer than half your agents are logged in at once.

Two practical cautions. Peak concurrency is higher than teams estimate, because handover overlaps, escalation huddles, and month-end all put more people in the tool than the rota suggests. And running out of concurrent seats does not degrade gracefully, it locks someone out, which is a support incident of its own. If you price a concurrent model, size it against observed peaks with headroom, not against the rota.

The tier gates: what each rung unlocks

Feature placement varies by vendor, so treat this as the shape of the ladder rather than a specification. What is consistent across the category is which capabilities tend to sit above the entry rung, and it is reliably the ones a growing desk discovers it needs in month three.

Usually at basic ticketing: email to ticket, assignment, statuses, canned replies, simple tags, a volume report.

Usually at growth or standard: automation rules, business hours, basic satisfaction surveys, a knowledge base, integrations with the tools you already run, and reporting that groups by agent, tag, and time.

Usually at professional: SLA policies with timers and breach alerts, a branded customer portal, custom fields and forms at scale, custom reporting and dashboards, multiple brands or mailboxes, skills-based or advanced routing, and finer permissions.

Usually at enterprise: sandbox environments, audit logs, data residency options, advanced approval workflows, and contractual support terms.

The buying implication is not “buy the top tier”. It is that the tier decision should be made against a written list of the capabilities you will genuinely operate, not the ones that sound useful in a demo. The step from growth to professional is the expensive one for most teams, and it is frequently triggered by exactly one requirement. When that happens, ask the vendor whether the capability can be added to your current plan as an option, because sometimes it can, and the answer is never volunteered. Our verdict on negotiating SaaS pricing covers how to ask.

Automation rules and why they sit one tier up

Automation is the feature most likely to justify a tier jump and the one most likely to be oversold in a demo. Trigger rules, macros with conditions, round-robin assignment, escalation timers, and scheduled actions all reduce manual handling, and vendors price them accordingly.

The honest way to price automation is against the hours it removes, using your own numbers rather than a vendor’s claimed averages. Estimate how many tickets a week are currently triaged by hand, how long each takes, and what a loaded support hour costs you. If a rule set removes two hours a week and a loaded hour is worth $40, that is roughly $4,160 a year of recovered capacity, which comfortably covers the illustrative $4,320 tier step for a twelve-agent desk. If it removes fifteen minutes, it does not.

There is a second cost nobody quotes: building and maintaining the rules. Automation that is set up once and never reviewed becomes a source of misrouted tickets within a year, as products change and queues get renamed. Budget somebody’s recurring attention, not just the licence.

Test the automation on real tickets during the trial rather than on the vendor’s sample data. Our verdict on running a software trial sets out how to structure that so the answer means something.

SLA management and the escalation tier

SLA management is the classic upper-tier feature, and it is worth being precise about what it means, because most teams already have service targets and think they therefore need the module. The module is not the promise, it is the machinery: clocks that start and pause on defined conditions, business-hours calendars, breach warnings before the deadline rather than after, escalation paths that fire automatically, and reporting that shows attainment by queue and by agent.

That machinery matters when someone outside your team is counting. If you have contractual response commitments, a parent company asking for attainment numbers, or a customer segment with priority handling, the timers earn their place. If your service target is an internal aspiration nobody audits, a tagging convention and a saved view often deliver most of the value at no extra cost.

The pricing consequence is the same step described earlier: SLA policies commonly sit at the professional rung, so a desk that needs them pays the full per-agent gap for everyone. On the illustrative $45 to $75 ladder, a ten-agent desk adds $300 a month, or $3,600 a year.

Before committing, ask two questions. Can SLA policies be scoped to the queues that need them, or do they apply account-wide. And does the reporting show attainment in a form the person asking for it will accept, because a dashboard that cannot be exported to the format your customer wants has not solved the problem you bought it for.

The customer portal and self-service

A customer portal is a signed-in space where customers see their own tickets, submit new ones through a form, and search your published articles. It is a genuine feature with a genuine cost, and it is also the one most often bought for the wrong reason.

The right reasons are specific. Customers who raise many tickets want to see the status of all of them without emailing to ask. Businesses selling to other businesses often need several people at a customer to see the same ticket. Structured forms capture the details your agents currently chase in a second reply, which shortens every conversation that follows.

The wrong reason is a general hope that customers will serve themselves and volume will fall. Vendors commonly claim substantial deflection from portals and knowledge bases, and those claims are marketing material rather than measured outcomes for your desk. Deflection depends on how good your articles are, how searchable they are, and whether your customers are the kind who look before they write. None of that is supplied by the licence.

Price the portal as a tier step, then decide whether you have the content to make it work. A portal in front of six thin articles is an expensive front door to an empty building. If you do not have the articles yet, the cheaper sequence is to write them first, publish them somewhere public, measure whether anyone reads them, and buy the portal afterwards.

Knowledge base pricing and where it lives

Knowledge base functionality sits in three places depending on the vendor, and which one applies changes your bill materially. Some include a basic public help centre from the entry tier. Some place it at the growth rung as a headline feature. Some sell it as a separate product with its own per-agent charge, illustratively around $10 per agent per month on top of the desk.

The functional differences that matter for cost are narrower than the feature lists suggest. Ask whether the knowledge base is public, internal, or both, because internal-only documentation for agents is a different requirement from a searchable customer help centre. Ask whether articles can be suggested to an agent inside the ticket, which is the feature that actually changes handling time. Ask whether the search is any good, which you can only judge by loading your own content during a trial.

There is a cost the licence never shows, and it is the larger one. A knowledge base is a writing commitment. Producing thirty solid articles and keeping them current as the product changes is a recurring editorial job, and desks that treat it as a one-time setup task end up with a help centre that answers last year’s questions. Budget the hours honestly, because an unmaintained knowledge base still costs the subscription while returning nothing.

If you are choosing between a slightly cheaper desk without a knowledge base and a slightly dearer one with a good one, price the standalone alternative before deciding, because a separate documentation tool is frequently cheaper than the tier step that includes one.

Multiple brands, mailboxes, and shared instances

The moment a company supports two products, two brands, or two customer segments under different names, the pricing conversation changes. Vendors handle this in one of three ways, and each carries a different cost shape.

Some allow multiple mailboxes or support addresses on any plan, which is the cheapest outcome and worth confirming explicitly rather than assuming. Some charge a flat fee per additional brand, illustratively around $50 per brand per month, covering its own portal, sender address, branding, and article set. Some reserve multi-brand support for the professional or enterprise rung, which reprices every agent rather than adding a line.

The third is by far the most expensive and the most common. A desk with ten agents that adds a second brand on an illustrative $45 to $75 ladder pays $300 a month for the privilege, or $3,600 a year, whether the second brand generates ten tickets a month or a thousand.

That asymmetry is worth planning around before a product launch or an acquisition rather than after. Ask the question at shortlist stage even if you have one brand today, because the answer determines whether your support platform can follow the company or has to be renegotiated when the company changes. It is the same discipline our verdict on choosing a CRM applies to record structure: the constraint that bites is rarely the one that was relevant on the day you bought.

Reporting and analytics behind the upper plans

Reporting is where the gap between tiers is widest and where trial evaluations are weakest, because a demo dataset makes every dashboard look competent.

Entry tiers typically report volume: tickets created, tickets closed, maybe a first-response average across everything. Growth tiers add grouping, so you can see volume by tag, by agent, by channel, and over time. Professional tiers add custom reporting, where you define the metric, the filter, and the breakdown yourself, and usually add scheduled delivery and export.

The distinction that decides whether you need the upper tier is whether anyone is going to ask you a question the built-in reports cannot answer. “How many tickets did we get” is answered everywhere. “What proportion of tickets from customers on our top plan breached target in the last quarter, split by product area” is a custom-reporting question, and if someone in your business asks it monthly, you will pay for the capability one way or another.

There is a cheaper path worth pricing first. If the platform exports raw ticket data on a schedule, a spreadsheet or an existing reporting tool can answer most bespoke questions without the tier step. That is more work and less polished, and for a ten-agent desk saving an illustrative $3,600 a year it is frequently the right trade. Ask specifically what export is available on the plan you are pricing, because export is sometimes gated too.

Live chat and messaging as a channel add-on

Email arrives with the licence. Everything else is bought, and live chat is the most common first purchase. Vendors commonly price it as a per-agent add-on, illustratively around $15 per agent per month, or fold it into a higher tier, which returns you to the tier-step arithmetic.

The per-agent structure is the part that catches people out. Chat is usually staffed by a subset of the team, but the add-on is often billed across every agent seat rather than the ones who take chats. On an illustrative $15 rate and twenty agents, that is $300 a month, or $3,600 a year, even if four people handle chat. Ask whether the channel can be licensed to a subset, because sometimes it can and sometimes it genuinely cannot.

Messaging channels behave similarly and add a wrinkle: some carry per-conversation charges levied by the underlying messaging network rather than by your vendor, which is a usage line your subscription cannot cap. If you plan to support customers through social or business messaging, get the pass-through charges in writing alongside the per-agent rate.

The operational cost is easy to miss and larger than the licence for many teams. Chat is a synchronous channel with a response expectation measured in seconds, which changes staffing patterns, breaks, and coverage. A desk that adds chat without adding people usually discovers that email response times slipped, and no line on the invoice explains why.

Phone, voice, and the per-minute layer

Voice is the channel where help desk pricing stops being a subscription and becomes a utility bill. Typical structure is a per-agent charge for the voice seat, illustratively around $25 per agent per month, plus usage: per-minute rates for inbound and outbound calls, monthly rental for each phone number, and often separate rates for recording storage and transcription.

The per-minute rates are individually tiny, illustratively a few cents a minute, and collectively significant. A desk handling 4,000 support minutes a month at an illustrative $0.03 a minute spends $120 a month, or $1,440 a year, purely on talk time. Double the volume and it doubles, which is a different cost behaviour from every other line in this verdict.

Two further charges deserve attention because they are commonly omitted from quotes. International and toll-free numbers usually carry different, higher rates than local ones, so a support line advertised as free to the caller is not free to you. And call recording, which many teams treat as standard, often carries a storage charge that grows every month it is retained, since recordings accumulate rather than turn over.

Voice also carries the least visible people cost of any channel. Calls cannot be batched, deferred, or handled two at a time, so a phone queue sets your minimum staffing rather than the other way around. Price the seat, price the minutes, then price the coverage the phone line commits you to, because that last one is usually the largest of the three.

AI assist priced as a separate per-agent add-on

AI features arrived in support tooling as a premium module and mostly stayed there. Two pricing shapes dominate, and they suit opposite kinds of desk.

The per-agent add-on charges illustratively around $30 per agent per month for reply drafting, summarisation, tone adjustment, and article suggestions. It is predictable, it scales with headcount rather than volume, and it is poor value for a large desk where only some agents use it. On twenty agents that is $600 a month, or $7,200 a year, whether the team leans on it daily or forgets it exists.

The usage-based shape charges per automated resolution or per conversation handled without a human, illustratively somewhere around $1 to $2 each. That aligns cost with outcome, which is attractive, and it introduces a variable line into a budget that was previously fixed. It also makes the definition of a “resolution” commercially important, so read that definition carefully before signing.

Vendors routinely present large deflection percentages and handling-time reductions for these features. Those figures are marketing claims rather than measured results for your desk, and the honest position is that nobody can tell you what your deflection rate will be until your own content and your own tickets are in the system. Treat any promised saving as a hypothesis to test during a trial, with your real ticket mix, before you buy a tier or a module to capture it.

Per-ticket and usage-based pricing models

Outside the mainstream per-agent model sit a few genuinely different shapes, and they are worth knowing because they occasionally fit a desk far better.

Per-ticket or per-conversation pricing charges for volume handled rather than people employed. It suits an operation with many occasional handlers and modest volume, for example a product team that fields a trickle of support alongside other work, where per-agent pricing would charge full rates for people who touch three tickets a week. It suits a high-volume desk badly, because cost then tracks the thing you cannot control.

Flat-rate plans with an agent cap appear at the small end, illustratively a fixed monthly fee covering up to a set number of agents. These are clean to budget and punishing at the boundary, since the agent above the cap moves you to the next bracket in one step.

Open-source and self-hosted options replace the subscription with infrastructure, maintenance, and somebody’s time. They are genuinely cheaper on licence and rarely cheaper all-in for a small team, because the hours have to come from somewhere. Price them honestly by putting a loaded hourly rate on the maintenance you will actually perform.

The general rule holds across all three: pick the model whose billing unit matches the thing that grows in your business. If headcount grows and volume is flat, per-agent hurts. If volume grows and headcount is flat, per-ticket hurts.

Implementation and migrating ticket history

Small desks often pay nothing to start, because importers handle a mailbox and a simple ticket export without help. Larger desks, older desks, and anything with custom fields commonly face a one-time implementation charge covering configuration, workflow rebuilds, integration wiring, and training.

As an illustrative planning figure, budget a few hundred dollars for a desk under about ten agents and roughly one to two times the monthly subscription for anything larger. The worked example later in this verdict uses $500 for the small case and 1.5 times the monthly subscription for the larger one, which is the same assumption the companion on this page applies.

Ticket history is the variable that moves the number, and the decision to make early is how much of it you actually need. Full history, with attachments, threading, custom fields, and original timestamps intact, is the expensive answer and often the one people ask for reflexively. A read-only archive of the old system for a defined period is cheaper and covers most real needs. Migrating only open tickets and starting fresh is cheapest and is entirely reasonable for a desk whose tickets close within weeks.

Ask what specifically survives the import, field by field, and ask for it in writing. Attachments, internal notes, satisfaction ratings, and the identity of the original agent are the four that most often arrive degraded. The sequencing discipline in our verdict on migrating to new software applies directly, and support migrations have an unusual advantage: you can run both systems in parallel for a fortnight far more easily than you can with payroll or accounting.

Integrations, apps, and the marketplace layer

Every help desk advertises an integration catalogue, and the catalogue is not the same thing as the integrations being included. Three cost patterns show up.

Some integrations are free and native, built by the vendor, included on the plan you are pricing. Some are free to install but require a paid plan on the other side, so connecting your desk to a chat tool or a status page assumes you are paying for that tool at a tier which permits the connection. And some are third-party apps with their own per-agent subscription, which is the pattern most likely to surprise a budget, because they appear inside the vendor’s own marketplace and read as part of the product.

The integration that most often carries a hidden cost is the one connecting your desk to your CRM or your billing system, because bidirectional sync is usually a paid capability on at least one side. Confirm what “integrates with” means concretely: does it open a link, does it show a read-only panel, or does it write data back.

There is also an API dimension worth pricing. If you plan to build anything custom, ask about rate limits and whether higher limits require a higher plan, because a limit that is fine for a dashboard is not fine for a nightly sync. Ask what happens when you exceed it, since throttling a support integration is an outage in slow motion.

Help desk software versus a shared inbox

For a genuinely small team, the honest comparison is not between two help desks. It is between a help desk and a shared inbox, and the shared inbox wins more often than support software marketing admits.

Shared inbox tools commonly price around $18 per user per month against an illustrative $45 for a growth-tier help desk. For three people that is $54 a month against $135, a difference of $972 a year. For that saving you keep assignment, internal notes, canned replies, and a record of who answered what, which covers a large share of what a small team needs.

What you give up is the machinery that only starts paying above a certain volume: routing rules that scale past a handful, SLA timers, a customer portal, a searchable knowledge base tied to the ticket, reporting on anything beyond volume, and the ability to run several brands. Each of those is dead weight at ten tickets a day and indispensable at two hundred.

The signal to upgrade is not headcount and not revenue. It is the first month you cannot answer, from records rather than memory, which conversations were missed and why. When the inbox stops being auditable, buy the desk. Until then, our free versus paid CRM verdict walks the same logic in a neighbouring category, and the honest answer there is the honest answer here.

Annual billing, renewals, and the price you keep

Support platforms follow the wider software pattern: the advertised per-agent rate is usually the annual-commitment rate, and month-to-month costs more, commonly somewhere in the ten to twenty percent range. On an illustrative $3,170 monthly subscription for a twenty-agent desk, that gap is roughly $3,800 to $7,600 a year.

Help desks are stickier than most tools, because the ticket history, the macros, the rules, and the customer portal all accumulate inside the platform. That lowers the risk of committing annually to something you will abandon in a quarter, and it raises the cost of being wrong about the vendor. The sequence that protects you is the same one that protects a payroll migration: start monthly, run one complete busy period including whatever your peak looks like, confirm the reporting answers the questions you actually get asked, then take the annual discount.

Watch the seat count in the contract as carefully as the rate. Annual agreements frequently fix a minimum number of seats for the term, which means seasonal hiring is easy to add and impossible to remove. Ask whether seats can be reduced at renewal, whether unused seats can be paused, and what happens to your rate if headcount falls.

Renewal is the other place money leaks. Auto-renewal at an increased rate is common, and the leverage to correct it exists only before the notice window closes. Diary the date and approach it with the method in our negotiating SaaS pricing verdict, and if the answer is no, our verdict on cancelling a SaaS subscription covers the exit properly.

A worked example: a 5-agent desk and a 20-agent desk

Numbers make the model concrete, so here are two desks priced end to end on the same illustrative ladder. Both use $45 per agent for a growth plan, $75 for professional, $10 per light agent seat, $15 per agent for chat, $25 per agent for voice, $30 per agent for AI assist, $0.03 per voice minute, and $50 a month for a second brand.

The 5-agent desk. Growth plan, three light agents, live chat, no voice, no AI. Seats are 5 times $45, which is $225. Light agents add 3 times $10, which is $30. Chat adds 5 times $15, which is $75. The monthly subscription is $330, or $3,960 a year. Implementation on a self-serve setup is an illustrative $500 one-time. First-year all-in is $4,460, which is $74.33 per agent per month against a $45 sticker. At 700 tickets a month, that is about $0.53 per ticket.

The 20-agent desk. Professional plan, ten light agents, chat, voice, and AI assist, 4,000 voice minutes a month, and a second brand. Seats are 20 times $75, which is $1,500. Light agents add 10 times $10, which is $100. Channels at $70 per agent add $1,400. Voice minutes add $120. The second brand adds $50. Monthly is $3,170, or $38,040 a year. Implementation at 1.5 times the monthly subscription is $4,755. First-year all-in is $42,795, which is $178.31 per agent per month. At 4,500 tickets a month, that is about $0.79 per ticket.

The two desks show the same ladder producing different shapes. The small desk’s cost is dominated by seats and one channel, so its lever is the tier. The larger desk’s cost is nearly half add-ons, so its lever is which channels it genuinely staffs. Load your own agent count, light seats, tier, and channel mix into the companion on this page to run this waterfall on your numbers.

Total first-year cost by team size

Scaling the same illustrative ladder gives a planning table you can sanity-check a quote against. All figures assume a growth plan at $45 per agent, light seats at a third of the agent count, live chat at $15 per agent, and implementation at $500 below ten agents or 1.5 times monthly above.

A 3-agent desk: $135 in seats, $10 in light seats, $45 in chat, so $190 a month, $2,280 a year, plus $500 setup, giving $2,780 first-year, or $77.22 per agent per month.

A 5-agent desk: $330 a month as calculated above, $3,960 a year, plus $500, giving $4,460, or $74.33 per agent per month.

A 10-agent desk: $450 in seats, $30 in light seats, $150 in chat, so $630 a month, $7,560 a year, plus $500, giving $8,060, or $67.17 per agent per month.

A 25-agent desk: $1,125 in seats, $80 in light seats, $375 in chat, so $1,580 a month, $18,960 a year, plus $2,370 setup, giving $21,330, or $71.10 per agent per month.

The effective per-agent figure barely moves, which is the defining property of per-agent pricing and the reason it differs so sharply from base-fee models like the one our payroll software cost verdict prices. There is no scale relief built into the list price. Whatever discount you get at size, you get by asking for it.

Signs your help desk bill has drifted

Support bills grow quietly, because the inputs change without anyone reviewing the plan. Several patterns reliably indicate a bill that has drifted past what the desk needs.

Full seats for people who never reply. The most common and the most expensive. Run the last-reply audit quarterly and convert everyone who fails it to a light seat.

Seats for people who left. Deactivating an agent and removing the licence are usually two separate actions, and only one of them stops the billing.

A tier bought for a feature nobody configured. SLA policies that were never written, a portal that was never launched, custom reports nobody built. This is a full per-agent step being paid for an unopened box.

Channel add-ons billed across the whole team. If four people take chats and twenty seats carry the chat charge, that is a conversation to have at renewal.

Voice recording storage that only grows. Retention that was set to forever on day one becomes a line item nobody recognises three years later.

Peak-season seats that never came off. Seasonal hiring is the classic cause, especially where the contract fixed a seat minimum for the term.

An AI module with no measured effect. If you bought it on a claimed deflection number and never measured your own, you do not know whether it is working, and the default assumption should not be that it is.

How to size a help desk plan to your team

Sizing a support platform is a classification exercise before it is a purchasing one, and doing it in the right order removes most of the cost surprises.

Count the people who will genuinely own and answer tickets. That is your full-seat number, and it is usually smaller than the list of people who want access. Then count everyone who needs to read and comment, which is your light-seat number. Getting this split right is worth more than any discount you will negotiate.

Then list your channels honestly, meaning the ones you will staff rather than the ones you would like to offer. An unstaffed chat widget is worse than no chat widget, and it is billed either way.

Then write down the three capabilities that would make you move up a tier, and check whether all three sit on the same rung. If they do, the step is easy to justify. If only one does, ask whether it can be bought as an option, or whether a workaround covers it for a year.

Only then price vendors, with your seat split, your channels, and your tier fixed, so the comparison is like for like. The true-cost calculator will run the seat arithmetic against your numbers, and the companion on this page adds the channel and implementation layers.

What to ask a help desk vendor before you sign

The quote answers the questions the vendor chose. These are the ones that change the number, and every one should be answered in writing.

What is a full agent seat and what is a light seat, precisely? Get the permission difference listed, not described.

Are light agents included, capped, or charged? And does the cap sit on the plan you are actually buying.

Which channels are included and which are per-agent add-ons? Ask specifically whether a channel can be licensed to a subset of agents.

What are the voice per-minute, number rental, and recording storage rates? Including international and toll-free, which usually differ.

Which features sit above my tier, and can any be added as an option? The answer is sometimes yes and is never volunteered.

How is AI priced, per agent or per resolution, and how is a resolution defined? The definition is the commercial term that matters.

What does implementation cost, and what exactly survives a ticket-history import? Field by field, including attachments and internal notes.

Does a second brand or mailbox require a higher tier? This one decides whether the platform can follow the company.

What is the renewal rate, the notice period, and the seat minimum for the term? And can seats be reduced at renewal.

Give me the full fee schedule, not the plan comparison. A vendor who produces it readily is one you can budget for.

The bottom line

Help desk software is priced on a model that reads as simple and behaves as a staircase. The per-agent rate is real, and it is the smallest part of what a working desk pays. Around it sit the seats you should not have bought at full price, the channel add-ons that multiply across everyone whether or not everyone uses them, the usage layer that voice introduces, the AI module priced like a second product, and the tier gate that reprices your entire team the day you need one feature that lives upstairs.

None of that makes the category bad value. A desk answering two hundred conversations a day with routing, timers, and a searchable knowledge base is worth considerably more than it charges, and the alternative is measured in missed tickets rather than saved subscription. The finding is that the seat rate is a poor basis for comparison, and that most of the money is decided before you look at a vendor at all: how many of your people genuinely need a full seat, which channels you will actually staff, which single feature is pulling you up a tier, and how much ticket history you insist on carrying across. Settle those four, price two vendors with the same assumptions, ask for the fee schedule, and treat every deflection claim as something to test rather than something to buy. Do that, and support tooling becomes a line you can forecast rather than one you reconcile in surprise.

VetLoft is paid for by readers rather than the vendors it reviews, and this verdict reflects that: it is educational material only and not procurement, contractual, or financial advice for your organisation. Every tier rate, light-seat price, channel add-on, per-minute figure, implementation estimate, and first-year total on this page is an illustrative planning number chosen to show how per-agent pricing behaves, not a quotation from any vendor, and support software pricing changes often enough that figures typical when this was written may read differently by the time you shop. Feature placement across tiers varies by vendor and moves between releases, so treat the ladder described here as a shape to test rather than a specification. Claims about ticket deflection, resolution rates, and productivity gains from automation or AI features are vendor marketing unless you have measured them on your own ticket mix. Confirm seat definitions, channel inclusions, usage rates, migration scope, renewal pricing, and seat minimums directly with each vendor in writing before you rely on any figure here.