At around 500 chats a month, the cheapest option is usually the one that bills only for chats the bot solves, without seat fees you do not need. As of November 2026, Gorgias charges $0.90 to $1.00 per automated interaction, Intercom Fin $0.99 per outcome, and Tidio sells its Lyro AI in blocks of conversations.
The difficulty is that these three prices are not measuring the same thing. One counts a "resolution" only if nobody on your team steps in afterwards. One can bill a hand-off to a human as an outcome. One counts every conversation the bot touches. Compare the unit, not just the number, and a quoted $0.99 can cost less or more than a quoted $1.00 depending on definitions.
The three billing definitions in plain words
Before any sums, set out what each vendor counts. The figures below come from each vendor's pricing page in November 2026 and are USD list prices; they change, so check before relying on them.
| Platform | What you pay for AI | What counts | Catch |
|---|---|---|---|
| Gorgias AI Agent | $0.90 per automated interaction on an annual contract, $1.00 on monthly; above the plan's included allowance the page lists $1.50 | A customer request fully resolved without a human agent | If a human replies within 72 hours it is billed as a helpdesk ticket instead |
| Intercom Fin | $0.99 per outcome | Customer confirms the answer helped or leaves without asking for more; also a completed procedure, including one you set to end in a hand-off | Leaving counts as an assumed resolution; with another helpdesk there is a 50-outcome monthly minimum |
| Tidio Lyro | Conversation blocks (first 50 free for life, then from $32.50 a month on annual billing) | Any conversation with at least one Lyro reply | Per-tier prices not published on the page; pay-per-resolution billing is listed only on the Premium plan |
The Gorgias allowance wording needs care. Its pricing page lists 30 automated interactions with the Basic plan and 190 with Pro, and it carries a second set of plan prices ($77 for Basic, $471 for Pro) that equals the helpdesk fee plus those included interactions at $0.90 each. Above the allowance you have bought, the page lists $1.50 per interaction. Read the page's footnotes against your plan before assuming $0.90 applies to every resolution.
The worked example: a small store with 500 chats a month
This is an illustrative store selling home-textile goods, with 500 customer chats a month. Assume, for the sums, that the AI resolves 60 percent of them without a person (300 chats) and the remaining 200 are escalated to the owner. Treat the 60 percent as an assumption, not a promise. Your own rate may be 30 or 80, and the next section shows how to test it.
Gorgias
The Basic helpdesk plan lists 300 tickets a month at $50 billed annually or $60 billed monthly. In this example the 200 escalated chats are ordinary tickets, so Basic covers them with room to spare. The AI side: 300 automated interactions at $0.90 is $270 on an annual contract, or $300 at $1.00 monthly, so an allowance sized to your volume gives about $320 a month in all ($50 plus $270). Basic on its own includes only 30 automated interactions ($77 a month on annual billing with them included); if the other 270 were billed at the $1.50 overage rate, that would add $405, about $482 in all. The realistic range is therefore about $320 to $480 a month, and the gap is why you ask Gorgias in writing which rate applies beyond your allowance before you sign.
The Complete Excel Course with AI Integration
From formulas and pivot tables to Power Query, DAX, VBA and AI with Copilot, ChatGPT and Claude — hands-on from the first lesson.
By Sagnik Bhattacharya and Virginia Thorn, Primetime Emmy Award winner
Buy nowIntercom with Fin
Fin at $0.99 per outcome across 300 resolutions is $297. Hand-offs bill only when Fin completes a procedure you configured to end in a hand-off; escalations from Fin's default behaviour or workspace rules are not billed. If all 200 escalations ran through such a procedure, that would add $198, which is why the definition matters. Then add helpdesk seats. Intercom's pricing page lists Essential at $29 per seat a month, currently shown at $19 with a 35 percent new-customer discount, Advanced at $85 and Expert at $132 (the annual-billing prices; monthly billing costs more). One seat on Essential brings the month to roughly $316 to $326 with no hand-off billing, and up to about $525 if hand-offs bill. If you use Fin with an existing helpdesk instead, the Fin pricing page lists a minimum of 50 outcomes a month.
Tidio
The free tier gives 50 AI conversations for life, the Starter plan is $24.17 a month, and the standalone Lyro AI Agent starts from $32.50 a month with quotas from 50 to 1,000 conversations (both prices match annual billing, which the page sells as two months free; above 1,000 it asks you to contact sales). The page does not publish the price per tier. Because Tidio counts any conversation with at least one Lyro reply, 500 chats could consume up to 500 conversations, not 300. Use the vendor's calculator at the 500 tier, and add the human-handled chats under the helpdesk plan.
The summary for this example: Fin and Gorgias land within a few dollars of each other if hand-offs do not bill and the Gorgias allowance rate is $0.90. Tidio is potentially cheaper at this volume, but you have to price its 500-conversation tier yourself, because the per-tier price is not on the page I could read.
Why the resolution rate matters more than the price
If your real resolution rate is 40 percent rather than 60, both per-resolution vendors cost you less per month (200 resolutions at $0.99 is $198) but the 300 remaining chats come back to your inbox. The cheapest platform on paper is not the cheapest if your inbox fills with chats you hoped the bot would close. Add your time to the sum: if the owner earns $30 an hour and each escalated chat takes six minutes, 100 extra escalations is 10 hours and $300 of time.
Work the break-even rule: a bot is worth keeping when the cost per resolved chat is lower than the cost of a person resolving the same chat. At six minutes and $30 an hour that is $3 a chat, so a $0.90 to $1 resolution is a bargain, provided it is a good resolution. The risk lies in the bad ones.
Testing your real rate in a fortnight
- Switch on the trial or free tier and leave it on for 14 days, with a human-handoff option visible.
- Export every AI conversation for the period.
- Sample 50 at random. Score each: solved, partly solved or wrong.
- Compare with the vendor's dashboard. An illustrative result: the dashboard claims 62 percent resolved, your sample finds 41 solved and 9 partial out of 50 (82 and 18 percent), but 6 of the "solved" ones gave a refund timeline that is wrong. After correction, your honest rate is nearer 70 percent of chats handled with 12 percent needing a fix to the bot's knowledge.
- Fix the knowledge base, then run another week and re-score.
The knowledge base is the real lever. The FAQ a chatbot needs before launch covers building it, and order tracking and returns explains which questions a bot should never answer without a human check.
Silent resolutions and other billing traps
Intercom counts an assumed resolution when a customer leaves the conversation after Fin's answer without asking for more help; its help article names no fixed time window, only an inactivity timeout. (If Fin asks a clarifying question and the customer never answers, no resolution is recorded and nothing is billed.) Leaving is not the same as satisfaction: a customer who gave up and bought elsewhere counts as a paid outcome. Gorgias avoids part of that with its 72-hour rule (a human reply within 72 hours turns the AI charge into an ordinary ticket), but a customer who comes back after the window has passed can still leave you paying for an answer that did not work.
Another trap is the minimum commitment. A per-resolution price looks like a pure usage bill, but Fin used alongside another helpdesk carries a monthly minimum, and Gorgias or Tidio plans can have ticket or conversation allowances that you pay for whether you use them or not. Before you sign, write down: the minimum monthly spend, the allowance, the overage rate, the contract term and the cancellation notice.
When a free tool is enough
For stores under about 100 chats a month, start free. Tidio's free tier has 50 lifetime AI conversations, which is a trial, not a plan. Shopify Inbox, a free app, can run an AI agent that answers shoppers; Shopify's help centre says it needs the Basic plan or above and new customer accounts (it does not work with legacy accounts). If you already run Shopify, try it before you pay a per-resolution vendor. The comparison in free AI helpdesks and their limits shows where those allowances stop.
Two more volumes, worked through
The 500-chat case is the centre of the range. Two other sizes show where the answer flips. Again these are illustrative stores and assumed resolution rates, not vendor claims.
A 120-chat store that sells handmade candles
At 120 chats a month with a 50 percent resolution rate, the bot resolves 60 chats. On per-resolution pricing at about $0.99, that is roughly $59 plus a seat or helpdesk fee. If the owner can handle the other 60 chats from a shared inbox for free, the real choice is between roughly $59 a month and a flat Tidio tier. The owner's time saved is 60 chats at five minutes, which is five hours. At $25 an hour that is $125 of time for a $59 bill: a small gain, worth taking only if the answers are right. If a single wrong answer about allergens or a fragrance ingredient could cause harm, keep the bot away from product-safety questions entirely.
A 1,800-chat store in a seasonal peak
An illustrative outdoor-gear shop sees 1,800 chats in its busiest month, mostly sizing, delivery cut-off dates and returns. Assume 65 percent resolution: 1,170 resolutions. At $0.99 that is $1,158; at $0.90 it is $1,053. The monthly total looks big, but a human team handling those 1,170 chats at six minutes each would take 117 hours. At $20 an hour, that is $2,340, so the bot is cheaper by about $1,180 in the peak month. The sum changes when quiet months follow: if the other eleven months run at 400 chats, an annual contract with a committed volume would waste money, and a monthly contract at the higher rate ($1.00 instead of $0.90 on Gorgias) is worth paying for the flexibility. Ask yourself whether your volume is steady or seasonal before you pick the annual option.
A scoring prompt for your fortnight sample
You can use an assistant to speed up the sample review, with one caution: remove customer names, emails, addresses and order numbers from the export first. Then paste batches of ten conversations with this prompt and check its verdicts yourself on at least a handful.
You are auditing a store's chatbot. For each conversation below, answer:
- Was the customer's question answered correctly using only our policy text? (yes / partly / no)
- Did the bot state any fact not in the policy text? (quote it)
- Did the customer need a human anyway? (yes / no)
Policy text:
[paste returns, delivery and sizing policy]
Conversations:
[paste ten, names removed]
An illustrative output row: "Conversation 4: partly. The bot said refunds take 3 to 5 days; the policy says 5 to 10 working days. Customer asked again after the answer." That single row is worth more than a dashboard figure. It shows a wrong promise that Intercom would have counted as a resolution if the customer simply stopped replying.
Four questions the bot should hand over, whatever the price
- Anything about a damaged or unsafe product. A person needs to see the photo and decide.
- Refund exceptions. Rules can be automated; exceptions cannot.
- Angry customers. Set a keyword or sentiment hand-off and test it with a few rude messages of your own.
- Anything involving personal data requests. Deletion or access requests have legal timelines and should land in a person's queue.
Hand-offs cost something on some plans, as noted above, but a wrongly resolved dispute costs more. Treat the hand-off count as a quality signal rather than a failure: a bot that hands off 20 percent of chats at the right moments is doing its job.
Reading your first monthly invoice
When the first bill arrives, reconcile it in ten minutes. Pull the vendor's list of billed events, count them, and multiply by the unit price to see whether the total matches. Then take five billed conversations at random and read them. Ask: would I have called that resolved? Illustrative findings from a first invoice: 14 of 300 billed outcomes were a customer typing "thanks" after the bot said it could not help, and 9 were greetings that never reached a question. If those counted as billable, 23 outcomes at about $1 is $23 of noise, 8 percent of the bill. That is a reasonable thing to raise with the vendor and to fix in your bot's opening message.
Keep a one-page record each month: chats, billed events, cost per billed event, cost per truly resolved chat (billed cost divided by the number of resolutions you verified in your sample) and the escalation rate. After three months you have a trend you can show to anyone who asks whether the bot earns its place, and a clear reason to switch if another vendor's pricing model suits your pattern better.
A decision table by volume
| Monthly chats | Sensible first move | Why |
|---|---|---|
| Under 100 | Free tier or built-in storefront agent | Per-resolution fees are not worth the set-up |
| 100 to 300 | Price Tidio's tier and Fin side by side | Flat blocks can beat per-outcome fees; per-outcome wins if the bot resolves little |
| 300 to 1,000 | Gorgias or Intercom, with a rate test first | Helpdesk and AI in one bill; your resolution rate drives the total |
| Over 1,000 | Ask each vendor for a quote and negotiate | Published list prices stop being the real price |
For sibling comparisons see Gorgias vs Intercom for a small online store, which prices the helpdesks on their own, and Gorgias alternatives if neither feels right.
A mistake that doubles the bill
A delivery-delay example makes the point. An illustrative store sets its bot to answer "where is my order?" questions. A carrier delay hits, order-status chats triple for a week, and the bot resolves most of them, each billed at about $1. The store pays several hundred dollars for a week of "resolutions" that were really the same one question, which a status banner on the order-tracking page could have answered for nothing. The fix is to look at your top five question types before launch and ask which of them is better solved by a page, an email or an automatic message than by a billed chat.
Put a spending cap or an alert in place on day one. Most platforms let you set a usage limit or watch a dashboard; if yours does not, schedule a weekly look at the count so a surprise comes to you as a text message rather than an invoice.
Negotiating and switching without losing your data
List prices are a starting point at any volume above a few hundred chats. Two moves are worth making. First, ask whether an annual contract with a lower per-resolution rate can include a cap, so that a spike in chats cannot multiply the bill. Second, ask what you can export if you leave: your conversation history, your knowledge-base articles and your macros (saved replies). A helpdesk that holds your history hostage is more expensive than its price suggests.
If you do switch, run the old and new bots in parallel for a week on different traffic slices, and compare your sampled resolution rate on each. Do not rely on the vendor's own figure for either. A switch that saves $40 a month and loses five percentage points of real resolution costs you in escalations what it saves on the invoice, and the owner is the one answering them. Write the decision into the same one-page record as above, so the next review starts from numbers rather than memory.
For a store with a single owner answering the escalations, the most valuable setting is often the least glamorous: office hours. A bot that tells customers "a person will reply by 10am tomorrow" and then does so keeps trust, even if the bot itself resolves a modest share. Cost per resolved chat matters, but cost per annoyed customer matters more.
A short checklist before you sign
- Which exact event is billed (resolution, outcome, conversation, ticket)?
- Are hand-offs and "assumed" resolutions billable?
- What is the minimum monthly spend, the allowance and the overage rate?
- Is the price annual or monthly, and what is the cancellation notice?
- Can you see every billed conversation, so you can dispute a charge?
- What customer data does the bot see, and where is it stored?
Keep the answers in one note. If you would rather work through this with someone using your own chat export, that is what my AI implementation consultation is for.
Billing questions to settle before you sign
What counts as a resolved chat?
It depends on the vendor. Intercom bills an outcome when the customer confirms Fin's answer helped or leaves without asking for more, or when Fin completes a procedure you configured, including one that ends in a hand-off. Gorgias bills an automated interaction only if no human replies within 72 hours. Tidio counts any conversation where Lyro replied at least once. Read each definition before comparing prices.
Is a free chatbot good enough for a small store?
For a few dozen chats a month, often yes. Tidio's free tier has a lifetime allowance of 50 AI conversations, and the free Shopify Inbox app can run an AI agent on eligible plans. Once chats reach a few hundred a month, free allowances run out and the per-resolution maths matters.
Do I pay for chats the bot fails to resolve?
On pure per-resolution pricing, no, but you still pay for the helpdesk or seats that handle the escalated chats. At Intercom a hand-off is billed only when it ends a procedure you configured, so check the equivalent rule at any vendor. Also check whether a human reply shortly after an AI answer turns the charge into an ordinary ticket, as it does at Gorgias within 72 hours.
How do I test the real resolution rate before paying?
Run the bot in a trial for two weeks and sample 50 conversations by hand. Mark each as truly solved, partly solved or wrongly answered, then compare your count with the vendor's dashboard. The gap between the two is the number to use in your cost forecast.
Further reads
- Your First 30 Days of AI Customer Support for an Online Shop — A month-by-month plan for launching AI support.
- AI Ticket Triage: Tag, Route, and Prioritise Support Requests — Tagging and routing the chats the bot hands over.
- How Much Does It Cost to Add AI to a Shopify Store? — The wider cost of adding AI to a Shopify store.
- How to Automate Returns and Refunds With Clear AI Rules — Rules for returns the bot may and may not decide.
- How to Use Shopify Sidekick to Run Your Store Faster — Shopify's built-in assistant for store tasks.
- AI Online Shopping Solutions for a Small Store: Chat, Quiz or Search? — Compare AI chat, recommendation quizzes, and smart search for small stores, with costs, decision rules, and a 20-question test.
- Best AI Tools for a Small Ecommerce Store: 7 Worth Paying For — Seven paid tools by job, with plan prices, what Shopify already includes free, what to skip and three monthly budgets.
- Best AI for Ecommerce Operations: Orders, Stock, Returns, Suppliers — Back-office AI for an online store: Sidekick for orders, Gorgias or Tidio Lyro for returns, and an assistant for supplier emails and reorder checks.
- Ecommerce Customer Service Outsourcing vs AI: Costs for a Small Brand — Cost per ticket at 300 and 2,000 tickets a month for outsourced support, an AI helpdesk and a hybrid, with contract points to check.
- Ecommerce Product Recommendation AI: Quiz, Widget or Chat Assistant? — A quiz, an on-page widget and a chat assistant costed for a small store, with a rule by catalogue size and a method to measure add-to-cart lift.
- AI Tools for Ecommerce Websites: What a 50-Orders-a-Month Store Needs — A need-versus-skip checklist for a store shipping about 50 orders a month, with free tiers, break-even sums and the volume at which each skipped tool pays.
- AI Tools and AI Development: The Complete 2026 Guide — the AI hub, including every tutorial in the AI-for-business series.
Sources: Gorgias, Tidio, Intercom and Fin pricing pages, Intercom's help article on Fin resolutions and Shopify's help centre article on the Inbox agent, all checked November 2026.