The Hidden Costs of Commercial Help Desk Software

Monday, 09:15, at Stonebridge Insurance. Tomas runs a desk of six and has just opened the renewal notice for the cloud help desk the company bought three years ago. It was bought for seven agents. The notice lists twelve. The fee per seat is higher than it was at the last renewal, which was higher than the one before. There is a line for a reporting add-on that Tomas vaguely remembers agreeing to. At the time, the quarterly report needed a chart the base product would not draw. Nadia from procurement has forwarded it with one sentence: "Can you confirm these numbers are right before I approve?" Tomas checks. Every number is right. Six people work tickets at Stonebridge, twelve seats are being paid for, and nobody has done anything wrong. That is the most expensive kind of invoice there is.

A licence is the right to use a piece of software. A subscription is a licence that lasts as long as you keep paying. A seat is one person's right to use it. Most commercial help desks count seats for the people who work tickets, called agents, rather than the people who raise them. The visible cost of a commercial tool is the number on the quote. The hidden costs are the ways that number changes after you sign, and the costs that never appear on it at all. This article is part of our Open Source vs Commercial Helpdesk Tools series. It is the commercial half of the ledger.

It covers why the invoice is the visible part, then the licensing models in plain words, with a table. Next come per-agent fees that grow with the team and add-ons, the features behind a second door. We cover lock-in and what it really is, the renewal conversation, and data export at the exit. Next come counting the five years and when the hidden costs are worth paying. We finish with a short war story and the renewal checklist to copy, in that order.

The Invoice Is the Visible Part

A commercial tool makes its cost visible, and the visibility is a genuine advantage over the free tool whose hours nobody counts. A quote arrives, someone approves it, an invoice follows, and the organisation knows what it spent. The trouble is that the quote describes the first year, for the people on the desk today, with the features in the base product. It assumes you will never leave. Four things move after that: the seat count, the feature list, the fee per seat, and your ability to walk away. Each is a hidden cost in the precise sense that it was real on the day you signed and visible only later.

None of these is a trick, and this article is not an exposé. Vendors publish their models, the renewal terms are in the contract, and the add-ons are on the order form. The costs hide because the people reading the quote are busy and the quote is for this year. The person who signs it is rarely the person who will be running the desk when the fourth renewal arrives. Commercial costs are not concealed. They are deferred, and deferred costs are paid by somebody who did not choose them.

Licensing Models in Plain Words

Every commercial help desk is sold under one of a handful of models. Knowing which one you are looking at tells you where the hidden cost will sit before you have read a single clause. The table below is the whole menu, in plain words, with no products on it. The first column is what you will see on the quote; the last is where to look for what you will not.

Most tools mix two of these. A subscription per agent with feature bundles is the commonest pairing in cloud help desks. A licence per installation with an optional update plan is common among self-hosted commercial tools. One disclosure belongs here rather than anywhere else in this article. The publisher of this library sells a self-hosted help desk, QuerySolv, licensed once per installation, which is the third row of the table. It is fair to say what the third row costs as well as what it saves. The server underneath it is yours to run. A single larger sum is, in many finance departments, a harder signature to obtain than a smaller recurring one. The models themselves, and which suits which kind of desk, are compared at more length in Hosting and Licensing Models. This article is about what happens after you have picked one.

Per-Agent Fees That Grow With the Team

A seat is paid for whether or not anyone sits in it, and seats accumulate. The desk hires a part-timer for the afternoons: a seat. The facilities manager handles two tickets a month about the car park gate and needs to be able to reply: a seat. The new IT manager wants to see the queue: a seat. An auditor needs read access for a fortnight in the spring: a seat, often for the year. A technician leaves and nobody releases their login because the leaver process does not mention the help desk: a seat, indefinitely. None of these people is an agent in the sense the quote assumed, and every one of them is an agent in the sense the invoice counts. Seats are like meeting invitations. They are far easier to add than to withdraw.

The fee per seat is the second mover. Most subscriptions are repriced at renewal, and the direction is rarely down. A desk of six on a per-agent subscription is therefore paying a number that rises on two axes at once, headcount and rate. Because each movement is small and arrives a year apart, no single renewal looks alarming. The fourth one, compared with the first, often does. Some tools soften this with cheaper roles for people who only need to view or occasionally reply. Find out whether yours does, and whether the facilities manager qualifies. Do that before you give him a full seat because it was the only option on the screen that day.

The alternative models move the question rather than remove it. A perpetual licence per agent is still per agent. A licence per installation is how some self-hosted tools including QuerySolv are sold. It means the question of whether the facilities manager counts as an agent never reaches the invoice. The question reaches the server instead, which is yours to run and keep patched. That is the hours column of the trade-off table. There is no model under which the occasional user is free of consequence. There are models under which the consequence is a login you administer rather than a line you pay for. For a desk with many occasional users that difference is the whole decision.

Add-ons: The Features Behind the Second Door

An add-on is a feature sold separately from the base product. And the hidden cost is that the demo showed you the product with every add-on switched on. In some tools, reporting beyond the basic counts, automation rules, a knowledge base, extra mailboxes, and more storage for attachments are each behind a second door with its own fee. The same can apply to single sign-on (logging in with the same account you use for everything else, instead of a separate help desk password). It can also apply to access to the API (the application programming interface, the way other software talks to the tool without a person clicking). The base product does the demo's first ten minutes. The add-ons do the last twenty.

The cost arrives in the second year, not the first, because that is when the desk discovers what it needs. Nobody asked for the reporting add-on at Stonebridge on day one. The quarterly report asked for it eighteen months later, when the finance director wanted a chart and the base product drew a table. By then the desk was built on the tool, the alternative was a spreadsheet every quarter, and the add-on was approved in an afternoon. That is how add-ons are designed to be bought, and it is not sinister. It is simply a cost the original quote did not carry and the original signer never saw.

The defence is a sentence, said during the trial: "Quote me for everything you just showed me." Then compare the quote against your requirements list. It should exist before any demo for the reasons in Designing a Fair Trial. Mark every requirement that lives behind a second door. A requirement met only by an add-on is a requirement with a cost. And the cost belongs in the five-year total, not in a surprise in year two.

Lock-in: What It Really Is

Lock-in is the cost of leaving, and it is best measured in hours rather than feelings. It has three parts. Data lock-in is the difficulty of getting your tickets, notes, history, attachments and knowledge articles out in a form the next tool can read. Process lock-in is everything the desk has built in the tool's particular way. That includes the workflows, the forms, the automation rules, the report layouts, and the habits of six people who know where every button is. Integration lock-in is the connections, the scripts that read the API, the monitoring that opens tickets, and the directory sync. Each was written for this tool and will be rewritten for the next. The data is usually the smallest of the three. The habits are usually the largest.

Lock-in grows every month you use the tool, which is why it is hidden: on the day you sign, it is zero. A year in, the desk has a hundred automation rules and nobody remembers why half of them exist. Three years in, the knowledge base has four hundred articles with the tool's formatting baked into them. The exit, which the first article of this series argues you should plan before you sign, costs more every quarter you postpone thinking about it. A vendor does not have to do anything to lock you in. Time does it for them.

The honest view is that lock-in is a cost of every tool in every category, open-source included. The only question is whether you know its size. Estimate it once a year in hours. Count how long to export and verify the data, how long to rebuild the processes in a generic tool, and how long to rewrite the integrations. Write the number on the renewal checklist. A desk that knows its exit would take two hundred hours negotiates differently from one that has never asked. And a desk that knows the number is six hundred has learned something about its automation rules.

The Renewal Conversation

The renewal is the one day a year the fee can go down, and it only goes down if you arrive with a count. Most subscription contracts renew automatically unless you give notice by a date some weeks before the term ends. The notice period is that gap, and missing it means another year at whatever the new rate is. Find the date in the contract now and put it in the calendar with a reminder a quarter ahead. And treat the quarter as the project it is. The vendor's account manager will have a renewal conversation whether or not you prepare for it. The only variable is which side has done the arithmetic.

Prepare three numbers. For the seat count, find who logged in during the last quarter and how many of them worked tickets. Find which seats belong to leavers, auditors or people who could be moved to a cheaper role. For the feature count, find which add-ons were used, by whom, and for what. Find which could be dropped or replaced by something the base product does well enough. And the alternative is what a comparable tool in another category would cost over the next term, converted to the same unit. You only know that if you have looked. None of these is a threat. They are the facts of your desk, and a renewal conversation held without them is a conversation about the vendor's facts instead.

Decide what you want before the call, and ask for it in writing. Fewer seats, the add-on dropped, the uplift capped, a shorter term, or a longer term in exchange for a fixed rate: each is a reasonable request. Each has a cost of its own, since a three-year fixed rate is also three years you cannot leave. A desk of six that reduces twelve seats to seven and drops an add-on nobody has opened since the spring has had a good renewal. A desk that signs the notice as it arrived because the numbers were correct has had a renewal, and will have another one in a year, slightly larger.

Data Export at the Exit

The exit is where every hidden cost in this article is finally paid, and the export is the part that decides how much. Most tools can export tickets to a spreadsheet file. And that sentence is both true and insufficient, because a ticket is more than its row. Check, during the trial and again each year, what comes out. Look for the public replies and the internal notes, told apart, and the event history, who changed what and when. Look for the attachments, which are frequently a separate export or a separate request, and the knowledge base, with its images. Look for the list of requesters and agents, so the next tool knows who everyone was. An export that contains the subject lines and the dates and nothing else is a list of what you have lost.

Test the export before you depend on the tool, not after. Run it during the trial on the sample data you loaded. Open the result and look for the internal note you wrote and the screenshot you attached. If they are missing, ask how they would be retrieved and get the answer in writing. If the answer is "contact support at the time", understand that the time will be the week you have given notice. That is when your account manager's enthusiasm for your problems may be at its yearly low. A cloud tool's export is whatever the vendor built. A self-hosted tool of either kind at least leaves the database on a server you own, though a readable export is still work. Either way, what you do with the history afterwards is the subject of Keeping History Without the Old System.

Build the exit into the ordinary year. Take a full export annually, even if you have no plan to leave, store it with the backups, and open it to check it is readable. It costs an hour. It turns the exit from a negotiation into a file you already have. And a vendor who knows you hold a readable copy of your own history negotiates with that knowledge too.

Counting the Five Years

Put the four movers on one page for the term you expect, and count in units that will still make sense when the fee changes. Stonebridge's desk, as a worked example, paid for seven seats in its first year. It paid for nine in its second after a part-timer and the facilities manager were added. It paid for twelve in its third after two technicians left and nobody released their logins. By then, a manager had also been given a seat to watch the queue, and an auditor's spring fortnight had become a permanent entry. Six people worked tickets throughout. In seat-months, the desk paid for 336 over three years and used about 216. So roughly a third of the subscription bought empty chairs, before the add-on and before the uplift at each renewal.

Set that against what the subscription genuinely bought, because it bought real things. Stonebridge ran no server for the help desk, patched nothing, backed up nothing, and renewed no certificate. It had a support line that answered within the contracted hours on the three occasions in three years when the mail connection failed. Those are hours the desk did not spend, and in the free-tool worksheet they would have run to several working weeks a year. The honest five-year sum for a commercial tool has both columns. One is the seats, add-ons and uplifts you paid for and did not need. The other is the hours you did not spend because somebody else did. A desk that counts only the first column has decided to be angry. A desk that counts only the second has decided to be billed.

When the Hidden Costs Are Worth Paying

For many small desks, every cost in this article is a fair exchange, and the fair version of the ledger says so. A desk with no server, no administrator and no appetite for either is buying exactly what a subscription sells. That is somebody else's hosting, patching, backups and support, for a predictable monthly sum that finance can approve without a capital request. A desk whose six people are fully occupied with the queue is right to pay for the vendor's hours rather than spend its own. A desk that counts its seats each year, drops the add-ons it does not open, and holds its export in hand has turned the hidden costs into visible ones. At that point they are simply costs, and costs can be decided about.

The desks that get hurt are the ones where the quote was signed once and the renewals were approved by whoever received them. There, seats were added by the only person with admin rights and released by nobody. The exit was never tested because leaving was never imagined. That is not a flaw in commercial software. It is a flaw in treating a subscription as a utility bill rather than a decision renewed yearly. The vendor renews its attention every year. The desk should too.

A Short War Story: The Renewal Notice at Stonebridge Insurance

Tomas took the renewal notice back to his desk and spent an afternoon on the seat list. Of the twelve seats Stonebridge was paying for, he found two belonged to technicians who had left the company more than a year earlier. One belonged to an auditor whose fortnight had ended the previous April, and one to a manager who had logged in twice. One belonged to the facilities manager, who handled about two tickets a month. He could have been given the cheaper reply-only role the tool had offered all along. Tomas counted the empty seat-months since the first renewal and arrived at roughly seventy. At the per-seat fee that was more than the desk's whole training budget for the year. The reporting add-on, he discovered, had last been opened by anyone the previous summer. Nadia took the counts into the renewal call. Stonebridge dropped to seven seats, moved the facilities manager to the cheaper role, cancelled the add-on, and had the uplift capped in writing for two years. The new invoice was lower than the one from three years earlier. The lesson Tomas wrote in the desk's runbook was one line: a renewal is the only day the fee goes down, and only if you arrive with a count.

The Renewal Checklist

Start this a quarter before the notice date and work down it over three or four weeks. And keep the finished copy with the contract so that next year's version starts from this year's answers. Every line is a count or a date, because the renewal conversation is won by whoever has more of those.

THE RENEWAL CHECKLIST — ONE COMMERCIAL TOOL, ONE PAGE
Tool: ______________   Model: [ ] per agent  [ ] per installation  [ ] bundle  [ ] usage
Term ends: ________   Notice deadline: ________   Auto-renews: [ ] yes  [ ] no
Owner of this checklist: ______________   Started on: ________

SEATS
  Seats paid for: ____    Logged in last quarter: ____    Worked a ticket: ____
  Leavers still holding a seat: ____   names: ______________________________
  Occasional users who could take a cheaper role: ____   names: ______________
  Seats to request at renewal: ____

ADD-ONS
  Add-on ______________   used by ______   last used ________   keep [ ] drop [ ]
  Add-on ______________   used by ______   last used ________   keep [ ] drop [ ]
  Add-on ______________   used by ______   last used ________   keep [ ] drop [ ]

FEE
  Fee per seat this term: ______   last term: ______   change: ______
  Uplift cap requested: [ ]   in writing: [ ]
  Term length requested: ______   in exchange for: ______________

EXIT
  Full export taken on: ________   opened and checked: [ ]
    tickets [ ]  internal notes [ ]  history [ ]  attachments [ ]  knowledge base [ ]
  Hours to leave, estimated: ______   (data ____  process ____  integrations ____)
  Alternative costed in the same unit: ______________   total for next term: ______

DECISION
  Renew as is [ ]   Renew with changes [ ]   Give notice [ ]
  Agreed changes confirmed in writing on: ________   by: ______________
  Next checklist starts on: ________

Tomas's first checklist took an afternoon and saved more than it cost by a margin he did not need a spreadsheet to see. His second, a year later, took forty minutes, because the counts were already on the page and only the dates had moved. The notice still arrived. It just arrived to someone who was expecting it.

Frequently Asked Questions