Zendesk Alternatives in 2026: Compare the AI Billing Rule, Not the Price Per Resolution

Last updated: August 18, 2026

Zendesk alternatives do not meter AI support the same way. Fin bills an outcome. HubSpot bills credits against a resolution decided three days later. Freshworks bills a time window. Help Scout bills a chat session. Gorgias bills an automated interaction and a ticket. Front bills a conversation. Zoho bills tokens. Run one month of your own traffic through each rule before you compare a rate.

Verdict in one sentence. These products can be compared, but only against the same month of your own conversations, because the rate printed next to the word resolution is attached to a different event at each vendor.

What each vendor actually charges for

Six of these seven pricing pages print a number with a dollar sign and a slash; the seventh prints a starting figure. The numbers are real. What sits to the right of the slash is a different object at each one.

Published rates, separately, because a rate and a definition should not sit in the same column.

The second column of the first table is the one that changes a forecast. A product that meters a time window will consume a unit on conversations it failed to resolve. A product that meters tokens produces a cost that varies with model tier and with the length of both the question and the answer. A product that meters two things at once will charge twice for the interaction its own marketing describes as the win.

Sources: Intercom Help, accessed 18 August 2026 · Fin, accessed 18 August 2026 · HubSpot Knowledge Base, 10 August 2026 · HubSpot Legal, accessed 18 August 2026 · Freshdesk Support, 4 August 2026 · Help Scout Docs, 25 July 2026 · Gorgias Help Center, accessed 18 August 2026 · Front, accessed 18 August 2026 · Zoho, accessed 18 August 2026

One email thread, seven different rules

Take a single sequence and hold it constant.

A customer emails a question. The AI agent replies with an answer drawn from the knowledge base. A workflow rule assigns the conversation to a support rep, who does not reply. The customer does not reply either. Seventy-four hours later, the customer replies to the same thread.

Applying each vendor’s published rule to that sequence produces different documented results. What follows is a reading of the documents, not a test. Where a document does not cover the case, that is recorded rather than filled in.

HubSpot. The agent posted a reply that shared a content source, so the resolution condition is met unless a qualifying handoff occurs. HubSpot states that only visitor-initiated handoffs qualify, and that manual assignment, including assignment through a workflow, does not. At 72 hours after the last visitor response the resolution is recorded and credits are consumed. The reply at 74 hours falls outside that window, and HubSpot states that a reply after the window reopens the conversation, restarts it, and can consume credits again. Two charges are documented as possible on this sequence.

Fin. The customer left after receiving an answer without asking for a person, which Intercom describes as an assumed resolution and bills. Intercom states that a teammate joining does not change that where the customer had already received an answer and had not asked for a person. When the customer returns to the same conversation seeking further assistance, Intercom states the resolution is deducted and not charged, and that this applies across billing periods. One charge, then a documented reversal.

Gorgias. A message was sent from the helpdesk, which Gorgias states makes the ticket billable. Whether an automation fee also applies turns on whether the customer required a human agent within 72 hours. The documentation does not state whether an assignment without a reply counts as human involvement, so this specific case is not settled by the page read. On the email channel Gorgias states that a reply on a ticket closed more than ten days earlier opens a new ticket; at 74 hours the reply falls inside that period.

Freshworks. The email session window is 72 hours from the customer’s first email, and all AI responses inside it count as one session. The reply at 74 hours falls outside. Whether it opens a new session is not stated on the page read.

Help Scout. AI Answers is defined for chat sessions on the page read, and the email case is outside that definition. The billing condition is stated in terms of whether the customer asks for more help after the last AI response, which this customer eventually does.

Front. No rule for this sequence is published.

Zoho. Tokens are consumed by the model calls that produced the answer. Nothing in the sequence after that point reverses them, and nothing on the page read addresses reversal.

Four of the seven pages settle the case. One settles it partially. Two do not address it. That distribution is the finding, and it is the reason a shortlist built from unit prices is unstable.

Sources: HubSpot Knowledge Base, 10 August 2026 · Intercom Help, accessed 18 August 2026 · Gorgias Help Center, accessed 18 August 2026 · Freshdesk Support, 4 August 2026 · Help Scout Docs, 25 July 2026 · Front, accessed 18 August 2026 · Zoho, accessed 18 August 2026

The customer comes back: reversal or second charge

The single behavior that separates these products most cleanly is a customer returning to a thread the AI already answered.

Intercom’s documentation states that if a conversation is considered resolved and the customer later returns to that same conversation seeking further assistance, even across billing periods, the resolution is deducted and not charged.

HubSpot’s documentation states that if a customer replies after more than 72 hours in an email thread, the conversation is reopened and the evaluation window restarts. If a further resolution is delivered, additional credits are consumed.

Both are sold with outcome language. The same customer action produces a reversal in one and a second charge in the other. The effect on a forecast is direct: a support operation with a high repeat-contact rate will see its effective cost move in opposite directions on these two products.

Sources: Intercom Help, accessed 18 August 2026 · HubSpot Knowledge Base, 10 August 2026

Human handoff does not always stop the charge

The common assumption is that a person taking over ends the AI charge. The documentation splits on this.

HubSpot’s position carries a second consequence worth stating plainly. It also states that after the status is set, subsequent live-agent messages, transfer requests, and negative feedback do not change it.

Sources: HubSpot Knowledge Base, 10 August 2026 · Intercom Help, accessed 18 August 2026 · Freshdesk Support, 4 August 2026 · Help Scout Docs, 25 July 2026 · Gorgias Help Center, accessed 18 August 2026

Gorgias: two meters on one conversation

Gorgias is the one product here whose documentation describes two charges landing on the same conversation, and it states the condition directly:

“An automation fee and a ticket fee may apply to the same ticket only when the ticket is entirely resolved by AI Agent, without human intervention.”

Two further points from the same page widen the exposure. A helpdesk ticket becomes billable when at least one message is sent from Gorgias, including an automatic message from a Rule. And an automated interaction is not restricted to the AI agent: Flows, Order Management, and Article Recommendations are each listed, with one of the examples being a customer reading a suggested help article and clicking that it was helpful.

The pricing page, captured on annual billing with the AI Agent option enabled, lists Basic at $77 a month made up of $50 of Helpdesk and $27 of AI Agent, with 300 tickets and 30 automated interactions; Pro at $471 made up of $300 and $171, with 2,000 tickets and 190 interactions; and Advanced at $1,227 made up of $750 and $477, with 5,000 tickets and 530 interactions.

FSR arithmetic. Dividing the AI Agent component by the included interactions gives $0.90 at every tier on annual billing, and $1.00 on monthly billing. This is an implied allocation inside a bundle, not a marginal rate published by Gorgias, and it should not be used as a contracted unit price. The published overage is $1.50 per automated interaction and does not fall on higher tiers, while the ticket overage falls from $0.40 to $0.36.

FSR arithmetic, stress test. Included interactions run at roughly a tenth of included tickets: 30 against 300, 190 against 2,000, and 530 against 5,000. Using the Pro plan’s own included ticket count as the base and its own published overage, 30 percent automation gives 600 interactions, 410 of them above the allowance, at $1,215 for 50 percent and $615 for 30 percent. The stress test excludes ticket overage and assumes ticket volume equals the plan allowance. It is a sensitivity check, not a forecast, and it is not a statement about any named customer’s bill.

One more thing belongs in a procurement note rather than a rate table. Accounts created before 28 May 2025 that have not subscribed to AI Agent and upgraded to use Shopping Assistant skills are on a legacy model where a ticket counts toward one meter rather than both. Gorgias states that choosing to upgrade means agreeing to the updated pricing model.

Sources: Gorgias, accessed 18 August 2026 · Gorgias Help Center, accessed 18 August 2026 · Gorgias Help Center, accessed 18 August 2026

What happens when the allowance ends

The control rule can matter more than the rate, because it decides whether spend stops, continues, or raises your committed floor.

The Freshworks page carries an internal inconsistency worth resolving before signing. It states that a session pack contains 100 sessions and costs $49, and separately that the default auto-recharge volume is one pack of 50 sessions. Both sentences appear on the same support article.

Sources: HubSpot Legal, accessed 18 August 2026 · HubSpot Knowledge Base, 10 August 2026 · Help Scout Docs, 25 July 2026 · Freshdesk Support, 4 August 2026 · Gorgias, accessed 18 August 2026 · Fin, accessed 18 August 2026 · Zoho, accessed 18 August 2026

The fixed layer: seats, caps, and required fees

The AI meter sits on top of a platform cost, and three parts of that cost are easy to miss.

Seat caps force plan changes independently of features. Front’s Starter is capped at 10 seats and Professional at 50. Help Scout’s Standard is capped at 25 users and Plus at 50; Pro is unlimited with a minimum of 10, which sets a floor of $750 a month before any AI charge. Gorgias prices by ticket volume rather than seats and lists 3 user seats on Starter and 500 on the tiers above it.

Buying the assistive AI on a middle tier can cost more than the tier above it. On Front’s pricing page, Professional is $65 a seat, Copilot is $20, and the Smart QA plus Smart CSAT bundle is $25. FSR arithmetic: $110 a seat on annual billing. Enterprise is $105 a seat with Copilot, Smart QA, and Smart CSAT included. Autopilot stays an add-on on all three tiers.

Required one-time fees are not on the rate card. HubSpot’s Japanese Service Hub page states a required one-time Professional onboarding fee of ¥180,000 and a required Enterprise onboarding fee of ¥420,000. HubSpot’s own product and services catalog states an onboarding requirement of $3,500 for Service Hub Enterprise and $1,500 for Sales Hub Professional, and states no onboarding requirement in its Service Hub Professional entry. Front states that an onboarding package is required for contracts over $25,000.

Monthly billing is not always a short commitment. On HubSpot’s Service Hub page, the Starter plan’s Pay Monthly option is labeled “Commit monthly.” The Professional plan’s Pay Monthly option is labeled “Commit annually.”

Regional price books are set separately. HubSpot’s Japanese page prices credits at ¥1,080 per 1,000 on annual billing, while the FAQ on the same page states $0.010 per credit; its Professional seat is ¥10,800 on that page, while the product and services catalog states Service Hub Professional starts at $100 per month per seat. Zoho’s Japanese Desk page prices Express at ¥990 per user per month on annual billing and notes that local taxes are charged. These are separately published list prices rather than conversions of one another, and neither page states a conversion method. Intercom states that it bills in United States Dollars only. Confirm which figure appears on your own order form. Zoho also publishes a table of which models are available in which data center for Zia Agents, and the entries differ by region; read it for your own region before assuming a model is available.

Published seat prices, annual billing, as they appear on each vendor’s own page. The last two rows are from Japanese price books and are not converted.

Sources: Freshworks, accessed 18 August 2026 · Freshworks, accessed 18 August 2026 · Help Scout, accessed 18 August 2026 · Intercom Help, accessed 18 August 2026 · Front, accessed 18 August 2026 · HubSpot, accessed 18 August 2026 · HubSpot Legal, accessed 18 August 2026 · Zoho, accessed 18 August 2026 · Gorgias, accessed 18 August 2026

Published resolution rates are vendor claims

Four of these vendors publish a percentage. None of the pages read set out a denominator or a measurement method.

These figures can be used to build a sensitivity range. They cannot be used as a shared denominator, for a reason that follows from section 01: for the vendors that bill a resolution, the thing being counted in the numerator is also the thing being billed, and each defines it differently. A stricter definition produces a lower published rate and fewer billable events on the same traffic. Freshworks and Zoho sit outside that relationship entirely, because neither bills a resolution.

Fin is the one vendor in this set whose public page attaches a financial remedy to a stated threshold. Its guarantee page describes a satisfaction guarantee returning up to $1M of Fin spend to new customers not satisfied within 90 days, subject to a minimum of 250 conversations, and a success program paying $1M to qualifying customers above 250,000 monthly conversations who do not reach a 65 percent resolution rate. Eligibility, geography, and remedy terms belong in the contract rather than in an article.

One movement is worth recording as version drift rather than as an inconsistency. HubSpot’s April 2026 announcement of outcome pricing stated 65 percent. Its Service Hub pricing page in August 2026 states 70 percent or more. Neither statement is accompanied by a method.

Sources: Fin, accessed 18 August 2026 · HubSpot, accessed 18 August 2026 · HubSpot, 13 April 2026 · Help Scout, accessed 18 August 2026 · Gorgias, accessed 18 August 2026

Build the model: how to compare these properly

Different units do not make comparison impossible. They make a headline-rate comparison meaningless and a workload comparison necessary. The work is mechanical.

Use one historical month of your own conversations. For each conversation, record seven fields:

1. Channel: chat, email, social, voice
2. Whether the AI delivered a substantive answer, as opposed to a greeting or a clarifying question
3. Whether the customer explicitly asked for a person
4. Whether a teammate was assigned or joined without the customer asking
5. Whether the customer came back inside 24 hours
6. Whether the customer came back inside 72 hours, and separately after 72 hours on the same thread
7. Whether an automation other than the AI agent closed it, such as a flow, an order action, or a help article marked useful

For a token-metered product, add input and output token counts from a representative sample of your real answers, measured on the model tier actually available in your data center.

Then apply each vendor’s documented rule to the same rows. Section 01 gives the rule and section 02 shows the method on one sequence. The output is a count of billable events per vendor, not a count of resolutions.

Build cost in three layers.

Report effective cost per 1,000 inbound conversations, not price per resolution. Then run three cases by moving four assumptions: repeat-contact rate, human-assignment rate, automation share, and answer length. A vendor can be cheapest in the expected case and most expensive when one of those moves. That result is the useful output of the exercise, and it is the one a ranked list cannot produce.

Ask for the evidence before signature, not after the first invoice. The definition of the billable event, the rate at your volume, the overage rule, and the behavior at the allowance limit all belong in the controlling commercial document. For a self-serve purchase there is no order form, so capture the pricing page and the checkout state on the day you buy, set whatever usage limit the product offers, and reconcile the first invoice against the underlying conversation list. FSR did not establish, from the pages read in this pass, whether each of these vendors exposes a per-event list of the interactions it billed. That question is worth putting in writing to each vendor before signature.

Questions buyers ask

Sources: Intercom Help, accessed 18 August 2026 · HubSpot Knowledge Base, 10 August 2026 · HubSpot Legal, accessed 18 August 2026 · Help Scout Docs, 25 July 2026 · Freshdesk Support, 4 August 2026 · Gorgias Help Center, accessed 18 August 2026 · Zoho, accessed 18 August 2026

Methodology

This is a Tier C briefing: built from primary documents, with no account, no purchase, no invoice, no dashboard access, and no product testing on any of the seven products. Tier describes evidence depth, not product quality.

What was read, and how. Every figure traces to a page published by the vendor it describes. FSR opened and read in full: the Gorgias billing and legacy-pricing help articles and the Gorgias Master Subscription Agreement; the HubSpot Customer Agent knowledge base article, the HubSpot product and services catalog, and the April 2026 pricing announcement; the Intercom Fin outcomes and plans help articles; the Freshdesk Freddy add-ons support article; the Help Scout AI resolutions pricing article; the Zoho Zia Agents platform details page; and the pricing pages of Front, Freshworks, Help Scout, Zoho, Gorgias, and HubSpot. One source was read only through search results rather than by opening the page: the Fin guarantee page. Its terms are reported as a vendor claim and should be read in full before being relied on.

Dynamic pages. The Gorgias pricing page renders differently depending on two toggles. The figures here were captured with annual billing selected and the AI Agent option enabled, on 18 August 2026. The HubSpot and Zoho figures marked JP were captured from the Japanese price books on the same date. Regional and toggle state changes these numbers; check the state of the page you are reading.

Arithmetic. Four calculations are FSR arithmetic rather than vendor-published figures, and are labeled in the text: the implied Gorgias allocation of $0.90 and $1.00; the ratio of included interactions to included tickets; the two Gorgias overage stress tests; and the Front seat comparison of $110 against $105. The Gorgias implied allocation is a division inside a bundled price and is not a marginal rate published by the vendor.

What is deliberately absent. No cross-vendor cost ranking, because the billing units are not equivalent without a workload model. No Zendesk figures, because FSR did not open Zendesk’s pages in this pass; that product is covered separately. No monthly rates for Help Scout or Front, because the monthly view of those pricing pages was not captured. No per-seat figures for Fin taken from the Intercom pricing page, which renders them as an animated counter; the seat prices used here come from Intercom’s own help documentation. No answer-quality assessment, because nothing was tested. No comparison of data-training policies, which is a separate question and belongs in a separate audit.

Unresolved. Whether AI Answers is available on Help Scout’s Free plan, where the pricing table and the billing documentation differ. Whether the Freshworks session pack contains 100 or 50 sessions in the auto-recharge default, where the same support article states both. Whether HubSpot’s free Customer Agent access is 14 days or 28 days, where the pricing page and knowledge base state 14 and the April 2026 announcement states 28; the current transactional pages should lead. Whether each vendor exposes a per-event list of billed interactions. Whether the Gorgias pricing page figures and the wider range published in that vendor’s other materials describe the same entitlement.

One observation from the pass, recorded but not developed. Two of the vendors checked publish a file addressed to AI assistants rather than to readers: Freshworks at /llms.txt and Gorgias at /llm-info. The Freshworks file is an index of links with no prices. The Gorgias file contains a pricing section. Whether any assistant retrieves either file, and what it does with them, was not tested, and the remaining vendors were not checked.

Replacement notice. This briefing replaces an earlier article published at this address on 16 April 2026. That article compared per-resolution rates across vendors as if the units were equivalent, and its Gorgias plan prices did not match the figures on the vendor’s published pricing page. It was withdrawn rather than amended, because the error was in the premise rather than in a figure.

Future Stack Reviews has no affiliate, referral, or reseller relationship with any product named in this briefing, and received no compensation, product access, or advance material.

Verdict

These seven products use similar outcome language for materially different invoice events. That does not make them incomparable. It makes the comparison you were handed the wrong one.

The replacement is not clever. Take one month of your own conversations, record the seven fields in section 09, run the rows through each vendor’s documented rule, and report effective cost per 1,000 inbound conversations. Then move four assumptions and see whether the ranking holds. In most support operations it will not, and the assumption that flips it is the one worth negotiating.

Two findings from this pass survive contact with any workload. A conversation the AI closed can carry two charges on one product, and a customer returning to a thread produces a reversal on one product and a second charge on another. Both are stated by the vendors themselves. Neither appears in a rate comparison.

Whatever you conclude, put the definition of the billable event into the controlling document alongside the rate, and confirm before signature that you can retrieve a list of the interactions you were charged for. The rate is the part of the deal that is easiest to negotiate and the least likely to determine what you pay.