A small business rarely decides to buy a help desk. It decides that the current arrangement has stopped working. The arrangement is usually one shared address, two or three people answering it between other jobs, and a spreadsheet that somebody started and nobody has updated for a fortnight. The trigger is a customer saying they never got a reply, and being right.
The market's answer to that moment is a category of software built for companies with a support department, sold on features a three person team will never open. The useful question is narrower: what is the smallest setup that still holds when volume doubles, and what does it cost to get there.
What a help desk is actually buying
Strip the feature lists back and there are five capabilities underneath. Every product in the category has all five. The differences are in depth, price, and how much configuration sits between the team and a working queue.
One queue. Every request in one place, regardless of who it was addressed to, so nothing exists only in a personal mailbox.
An owner per request. A named person on each item, visible to everybody, so the question of who is dealing with it has an answer that does not require asking.
A state. At minimum open and closed, and in practice a few stages in between, most importantly some way to say that the request is waiting on the customer rather than on the team.
A history per person. Everything this customer has asked before, in one view, so a new handler does not start from nothing.
Reusable replies. Saved wording for the questions that arrive weekly, which is where most of the time savings in the category actually comes from.
Anything beyond those five is a refinement: satisfaction surveys, a public knowledge base, chat widgets, telephony, automated triage. Refinements are worth buying when the volume justifies them, which for a small business is usually later than the sales process implies.
The smallest setup that still holds
Adding the five capabilities in order is what keeps the setup honest, because each step is only worth taking once the previous one is in daily use.
Start with the queue and the owner. This can be done without spending anything if the business already has Google Workspace, since a group with collaborative inbox features enabled gives a shared conversation list where items can be claimed, assigned to a named member, and marked complete. It is manual and it has no reporting, and it removes the two failures that hurt most: the message answered twice and the message answered never.
Add states next, and resist the urge to design a dozen. Three stages cover almost everything a small team needs: new, working on it, and waiting on the customer, with closed as the fourth. The waiting on the customer stage is the one that pays for itself, because it stops a queue of thirty items from looking like thirty things the team is behind on when twenty of them are sitting with somebody else.
Add saved replies third. Ten templates covering the ten most common questions will cut more minutes per week than any automation feature sold at a higher tier.
Add reporting last, and only when somebody is going to act on it. Volume by week, time to first reply, and volume per handler are the three numbers that change decisions. Everything else is interesting.
What the entry tiers cost today
Published rates for the products a small business is most likely to shortlist, at the time of writing.
| Product | Entry plan | Next tier | Notable terms |
|---|---|---|---|
| Freshdesk | 19 dollars per agent per month billed annually | 55 dollars per agent per month | No free plan, 14 day trial on the top tier |
| Help Scout | Free plan, then 25 dollars per user per month | 45 dollars per user per month | Free plan covers 5 users, 1 inbox, 1 documentation site |
| Hiver | 25 dollars per user per month billed annually, 35 dollars monthly | 55 dollars per user per month | All plans start at 2 seats |
| Front | 25 dollars per seat per month billed annually | 65 dollars per seat per month | Entry plan is a single channel type, up to 10 seats |
| Zoho Desk | Free edition | Paid tiers per agent | Free edition covers three user licences |
What the price is counted on matters more than the price. All five count people, so a busy month costs the same as a quiet one, which is good news for seasonal businesses and bad news for a team that wants everyone to have a login for occasional use.
Some capability sits outside the seat price, and a quote built from plan prices alone will be low. Help Scout charges 75 cents per resolution for its AI answers feature, and 10 dollars a month annually for each additional inbox, rising to 12 dollars monthly. Front lists Copilot at 20 dollars per seat per month as an add-on and prices its Autopilot feature from 5 cents per conversation. Hiver sells voice as a separate seat at 30 dollars per month including a thousand minutes, with further minutes at 3 cents each. None of those are hidden, and all of them are easy to leave out of a comparison spreadsheet.
The free floor, and where it runs out
Three genuinely free options exist, and for a business with two or three people answering a modest number of requests, one of them may be the whole answer for a year.
Zoho Desk publishes a free edition with three user licences. Help Scout publishes a free plan for five users with one inbox and one documentation site. And any business already paying for Google Workspace has the collaborative inbox features included, with no additional licence.
The limits that bite are rarely the ones people expect. The user count is usually fine, since a small team is small. What runs out first is the number of inboxes, because a business that started with one address discovers it wants a second for applications or invoices, and a second inbox is often a paid line. After that comes automation, because free tiers generally have no rules, so every routing decision stays manual. Reporting is third, and the absence of it is tolerable right up to the week somebody asks for last quarter's numbers.
Where a help desk is the wrong shape
This is the part the category does not discuss, and it accounts for a good share of abandoned subscriptions.
A help desk is built around a conversation. Something arrives as a message, it goes back and forth, it ends. That model fits questions, complaints, and faults very well. It fits a whole category of work badly, and small businesses are full of that other category: applications, bookings, quote requests, registrations, onboarding, repair intake, membership renewals. Those are not conversations. They are records with a fixed set of fields that must be known before work can start, and a defined path from arrival to done.
Forced into a ticket, that kind of work loses the thing that makes it manageable. The date, the model number, the file, and the choice from a list all end up as prose inside a message, so nothing can filter on them, sort on them, or count them. The team compensates with a spreadsheet beside the help desk, and the two drift apart within weeks.
The alternative is to keep the submission as a record. A form tool where each response carries its own owner, stage, and history covers the same five capabilities listed at the top, and does so without flattening the answers, because the structured fields and the handling state are the same object. For work with a fixed shape that is the closer fit, and it also removes the round trips that come from asking for missing information after the fact. The use cases page sets out which request shapes suit that arrangement.
A sensible test: if the last twenty items in the queue could all be described by the same ten questions, the work is form shaped. If each one needed a different conversation, it is ticket shaped.
How to run the trial so it tells you something
Two weeks of trial can be spent either confirming that the software works, which it does, or answering the three questions that actually decide the purchase.
The first question is whether the team will use it when busy. That is tested by putting real requests through it during a genuinely busy period, not by a quiet afternoon of clicking. If people keep replying from their own mailbox because it is faster, the tool has failed the test regardless of how good the feature list is, and the reason is usually one extra step in the reply path.
The second question is what the tier being bought can actually do. Trials generally run on the top plan, so a feature that made the fortnight pleasant may be an add-on or a higher tier. Confirming which of the routing rules, reports, and templates belong to the plan that will be paid for avoids a renewal conversation later.
The third question is what happens to the records when somebody leaves. Reassignment of an open queue, and whether history stays visible after a user is removed, is dull to check and expensive to discover. A small business loses a seat for ordinary reasons, and the queue has to survive it without a rebuild.
One more thing is worth measuring during the trial: how many of the requests needed a reply that only asked for missing information. That count is the clearest signal in the whole exercise, because it is the part no help desk can reduce and a structured intake form can remove outright.
The costs that do not appear in the quote
Four of them, in the order they usually surprise people.
Seat creep. The manager wants visibility, the bookkeeper wants access to invoice queries, the intern needs a login for two months. Per seat pricing turns each of those into a recurring line, which is why view only access at no charge, where a product offers it, is worth checking for.
Minimums. Some plans start at two seats, and some top tiers require ten. A sole trader on a plan with a two seat minimum is paying for a colleague who does not exist.
Migration. Getting existing history out of the current arrangement and into the new one is a day of work at least, and it is the step most likely to be skipped, which leaves the old mailbox as a permanent second place to look.
Unused depth. Paying a second tier price for a feature nobody configured is the most common form of waste in the category. The way to avoid it is to run the trial on the tier that will actually be bought, rather than on the top tier the vendor enables by default.
What to change first
Before comparing products, spend a week putting every request into one list with a name and a stage against it, even if the list is a shared group inbox. The pattern in that week decides everything: conversational volume points to a help desk, and a repeating set of the same ten questions points to a form with response management instead. Halict shows what the second path looks like from submission through to reply.
Q1. How much should a small business expect to pay for a help desk?
Published entry rates currently sit between roughly 19 and 25 dollars per agent per month on annual billing, with second tiers between 45 and 65 dollars. Free options exist from Zoho Desk for three users and Help Scout for five users with one inbox, which covers a genuinely small team for a while.
Q2. Is a free help desk enough for two or three people?
Often yes, for a year or more. The constraint that appears first is usually the number of inboxes rather than the number of people, followed by the absence of automation rules and reporting. A business with one address, a modest volume, and no reporting requirement has little to gain from paying.
Q3. Do you need a help desk if you already pay for Google Workspace?
Not immediately. A group with collaborative inbox features gives a shared queue where conversations can be claimed, assigned, and marked complete at no extra cost. What it does not give is stages beyond complete, saved replies, per person volume, or reporting, and those are the reasons teams eventually move.
Q4. What is the difference between a help desk and a shared inbox?
A shared inbox is a mailbox several people work, organised around threads. A help desk adds the apparatus around the thread: statuses, a customer history, saved replies, satisfaction surveys, and reporting. For a small team the practical difference is how much structure comes ready made versus how much has to be maintained by hand.
Q5. Should application forms and bookings go through the help desk?
Only if the answers stay as fields. Work with a fixed set of required information is better kept as a record with its own stage, because a ticket turns those answers into prose in a message body and the team ends up maintaining a spreadsheet alongside to recover the structure.