BotSupport vs Intercom Fin: which AI support agent fits your online shop?

What is Intercom Fin, and what does it actually do?

It helps to be precise about the category, because Fin is often described loosely as a chat widget. It is not. Fin is the AI layer of a full customer service platform that also includes a shared inbox, ticketing, a knowledge base product, outbound messaging and an agent-side assistant Intercom calls Copilot. You can buy Fin without Intercom's helpdesk and point it at another helpdesk, but the product was designed as part of a suite and most of its depth lives there. That matters for a small shop, because you are not evaluating a bolt-on. You are evaluating whether to adopt a platform.

In May 2026 Intercom announced Fin for Ecommerce, described in its own launch post as a role purpose-built for Shopify merchants. It combines shopping assistance, meaning product discovery, recommendations, cart help and checkout guidance, with post-purchase support such as order tracking, returns, refunds, exchanges and order changes. Intercom's documentation describes Fin choosing which mode a customer needs and switching between them inside a single conversation.

Fin's ecommerce solutions page publishes performance figures that should be read as vendor self-reported rather than independently audited. As checked in August 2026 it states a 67% average resolution rate across customers, with many ecommerce brands reaching 85% or higher, and support for 45 or more languages. On the scale of the business, coverage of the Salesforce acquisition in June 2026 described Fin as bringing a customer base of roughly 30,000 companies. Public review scores point the same way: Fin sat at around 4.5 out of 5 across several thousand G2 reviews when we last checked in August 2026, and its Shopify App Store listing rates similarly across a much smaller number of reviews. Review counts and scores drift weekly, so check both listings for the current position rather than relying on ours.

Two things follow from that. First, Fin is a genuinely capable and widely deployed product, and any comparison that pretends otherwise is not worth reading. Second, it is a platform purchase. You are adopting a system, configuring it, and taking on the work of keeping it accurate. That is where the real difference with BotSupport sits, and it is a difference of operating model rather than of raw capability.

Before we compare: what BotSupport is today

It would be easy to write this page entirely in the habitual present tense and let it read as a mature service. We would rather say it plainly. BotSupport is pre-launch. There is no installed base, no case study library and no years of uptime to cite. When this page says a person builds your assistant on your live catalogue, that is a description of the offer and of the process we run, not a claim about hundreds of previous builds.

This is also why the page contains no adoption or popularity claims. We are not going to tell you that hundreds of shops trust BotSupport, because they do not yet. What we will tell you is what we commit to: we build and hand over a working assistant within one working day, grounded in your real products, prices and policies, and there is a free trial period so you can hold us to that rather than take it on faith.

The honest way to use this comparison, then, is to treat every Intercom claim as sourced and checkable, which it is, and every BotSupport claim as a commitment you can test cheaply before you commit budget. If the trial output does not convince you, the rest of the comparison does not matter. We would rather you found that out in a week than in a year.

Who builds and maintains the assistant, you or the vendor?

Fin's answer quality is a direct function of the content it can read. Intercom is fairly open about this and publishes guidance on knowledge management for AI-first support. This is not a quirk of Fin. It is the honest shape of any retrieval-based agent sold as software: the answers are only as good as the documentation behind them, and somebody has to own that documentation.

A retrieval-based agent also does not go quiet when its sources go stale. It keeps answering, confidently, from whatever it can still read. On a usage-based model those conversations may still be billable, so it is worth asking Intercom directly how incorrect or out-of-date answers are treated for billing before you sign. To be clear, we are not alleging that Intercom charges customers for wrong answers. We are saying that the interaction between stale content and per-outcome billing is a question any buyer should get answered in writing.

In most support teams the person who ends up owning the content is a support lead who already has a queue to clear. That is the quiet cost of the self-serve model, and it does not appear on any invoice.

BotSupport removes that role from your side of the line deliberately. There is no self-serve tier to sign up for. A person reads your product range, your delivery and returns policies and your real customer questions, builds the assistant, tests it and hands it over working. Changes are requested rather than configured. The assistant is grounded in the live catalogue, so product names, prices, variants and availability are read from the store rather than transcribed into help articles that then have to be kept in step with it.

The fair counterweight: today you request changes rather than making them yourself, so a same-minute edit at eleven at night is not something we offer. If your team has strong support operations people who want direct control of prompts, workflows, routing and deflection thresholds, that will frustrate them, and Fin is the better fit. If you do not have those people and were never going to hire them, the managed model is the difference between an assistant that works and one that quietly rots.

Does Intercom Fin work with WooCommerce, or is it Shopify-first?

Be careful with the shorthand that Fin is Shopify-only, because it is both more and less true than it sounds. Fin the support agent is platform-agnostic and omnichannel. Fin for Ecommerce, the role that does product discovery, cart help and automated order actions, is the part that leans on Shopify's APIs. Intercom's own help documentation for Fin for Ecommerce describes Shopify as the supported ecommerce provider for that role, checked August 2026, while Intercom also documents no-code data connectors for systems with an API and custom carts configured that way. If you are on WooCommerce, Magento or BigCommerce, treat that as a sales conversation with one specific written question: which order actions can Fin actually take on my platform, and which are Shopify-only?

There is a second limit worth knowing before you buy, and it is documented rather than inferred. Intercom's documentation as of August 2026 describes Fin for Ecommerce as best deployed on the web Messenger, where the full shopping capabilities including product recommendations, cart updates and checkout are supported. On iOS and Android it is documented as switching to support-only responses, and email, social, voice and Slack are listed as not currently supported for the ecommerce role. Channel coverage on a product moving this fast changes, so confirm the current position with Intercom rather than with this page. The general point stands either way: the omnichannel breadth Fin markets applies to the support agent, not to the shopping experience.

BotSupport takes the opposite approach to platform coverage. Because a person builds each assistant against the specific store, WooCommerce gets the same treatment as Shopify. Grounding is on the live catalogue, so a price change or a product going out of stock is reflected without anyone editing an article. We work with WooCommerce, Shopify and most custom storefronts. Tell us your stack and we will confirm it before you pay, rather than after.

The honest limitation on the BotSupport side is scope rather than platform. It is an answering and advising assistant grounded in your real data, not an order-mutation engine. Intercom documents Fin as able to complete order actions such as returns, refunds, exchanges and cancellations on Shopify through Procedures, subject to how you configure them. If your priority is a bot that issues the refund and books the exchange inside the chat with no human touching it, that is a real capability gap and Fin is the product that closes it.

How do Intercom Fin and BotSupport pricing models compare?

Take the competitor's numbers from the competitor. As published on Intercom's pricing page in August 2026, Fin is charged at $0.99 per outcome, and helpdesk seat plans run from roughly $29 per seat per month at the entry tier to roughly $132 per seat per month at the top tier, with the billing term affecting what you actually pay. Intercom also defines more than one billable outcome type and charges some, such as lead qualification, at a higher rate than a standard resolution. We check these figures periodically, but pricing pages change and repricing after a change of ownership is a live possibility, so confirm the current rates and definitions with Intercom before you budget.

The 50-outcome question is worth stating carefully, because it is easy to get wrong in a comparison's own favour. Fin deployed on a helpdesk other than Intercom's is documented with a base fee of $49 per month that includes 50 resolutions, with additional outcomes at the published per-outcome rate. That is a floor price with volume bundled into it, not a bare minimum spend stacked on top of usage charges. If your monthly volume is genuinely tiny, the $49 floor is the number to compare against a flat subscription, not $0.99 multiplied by a handful of chats. Check Intercom's current pricing page before you budget, because this is exactly the sort of term that moves.

The characteristic of usage-based pricing is that cost rises with volume, which usually means it rises exactly when you are busiest. At the published per-outcome rate, a month with around a thousand billable outcomes runs into three figures of Fin charges before any seat cost. Model your own volume against Intercom's current published rate rather than relying on our example. This is not a criticism, it is simply the model, and for a larger team replacing headcount it can be excellent value. It does make budgeting harder, and cost predictability comes up repeatedly in public reviews of usage-based AI support pricing, Fin included. Read the recent G2 reviews yourself rather than taking our summary of them.

BotSupport's flat monthly fee at Early Access pricing is the mirror image. It is predictable and it does not penalise a good month, but it is the same price whether the assistant handles thirty conversations or three thousand. If you already pay for Intercom seats and field only a handful of chats a month, Fin's entry cost can genuinely work out cheaper than a flat subscription, and we would rather say so than pretend otherwise. Volume is what flips it: the higher your chat volume, the better a flat fee looks against a per-outcome rate.

What counts as a billable outcome, in Intercom's own words?

This is the single most important paragraph on the page for anyone budgeting, so we are quoting rather than paraphrasing. Intercom's help documentation states: 'A Resolution is an outcome where the customer confirms Fin resolved the issue or does not request more help after Fin answers.' It also states that 'if a customer disengages from the conversation for 24 hours after Fin's last answer, it is considered an assumed resolution.' Intercom's own term for that case is assumed resolution, which is a candid piece of naming and to Intercom's credit is published rather than buried.

What that means in practice is that the billable event is not only an explicit thumbs-up. A conversation where the customer got what they needed and left, and a conversation where the customer gave up and closed the tab, can look similar to the billing logic once the 24 hour window passes. Some reviewers have argued this flatters both apparent resolution rates and invoices. Whether it matters to you depends entirely on your traffic mix and how much of it is casual browsing. Intercom's documentation also notes that escalations triggered by default behaviour or workspace-level rules are not charged, and that a conversation is charged once regardless of how many questions or Procedures it contains, both of which cut the other way.

The practical advice is the same whichever way you lean. Ask Intercom to confirm the billing trigger in writing for your account, and ask specifically how abandoned conversations and incorrect answers are treated. This is a normal commercial question and any competent vendor will answer it. It is far cheaper to ask before the contract than to reconcile it afterwards.

BotSupport has nothing equivalent to explain, which is the main practical advantage of a flat fee. There is no billable event to define, no assumed anything, and no invoice that moves with a busy week. That simplicity is worth much less to a large support organisation than it is to a small shop. For a small shop it is worth quite a lot, because the alternative is building a forecast on a definition you did not write.

What is the staff-time cost that buyers forget to price in?

It is worth being concrete about what the work actually is with a self-serve agent. Someone audits the existing help content and rewrites the contradictory bits. Someone decides which Procedures to build and tests them. Someone reads a sample of conversations each week and identifies where the agent answered thinly. Someone updates the content when delivery charges change or a supplier stops stocking a line. None of that appears on an invoice, which is precisely why it gets left out of the comparison.

How long the initial pass takes depends entirely on the state of your documentation, and we are not going to invent a number for your business. If your policies are already written down accurately in one place, it may be short. If they live half in a help centre, half in a spreadsheet and half in the head of the person who runs dispatch, it will not be. The reliable part is that the work is recurring rather than one-off: every product change and every policy change reopens it.

Estimate it the way you would estimate any other recurring task. Count the number of content-affecting changes you make in a typical month, add a weekly review of a sample of conversations, and price the result at whatever an hour of your support lead's time is worth to you. Whatever the figure comes to, price it, because nothing is going to bill you for it. For a large support organisation it is a rounding error against the savings. For a five-person ecommerce business it is often the reason a promising AI pilot is quietly switched off six months later.

The managed model is not magic, it just moves the work. BotSupport's answer to the same problem is that a person does the auditing, the testing and the ongoing updates, and catalogue grounding removes a large slice of the content maintenance entirely, because product data is read live rather than documented by hand. What you still owe is a clear statement of your policies and a person who will answer questions during the build. That is a real obligation, it is simply a much smaller one.

How quickly can each one be live and answering real questions?

Fin's technical install is fast and Intercom is not overstating that. Connect Shopify, enable the app embed in your theme, install the Messenger and preview. Intercom documents catalogue and order data syncing on connection. On the raw mechanics this is one of the smoothest setups in the category and it deserves credit for that.

The gap between technically live and actually good is where the timelines diverge, and it is not a fault in the software. Fin answers from what it can read. Where the underlying content is thin, scattered or contradictory, the answers will be confidently vague until that is fixed, and fixing it is a content project rather than a configuration one. How long that takes is a fact about your documentation, not about Intercom.

BotSupport compresses that by making it someone else's job. The build includes reading the store, the policies and the real question patterns, testing the answers and correcting them before handover. We commit to a working assistant within one working day, and the free trial period exists so you can hold us to that. We are pre-launch, so that commitment is the first thing you should test and the easiest thing to falsify.

If speed to a first answer is your metric, both are fast. If speed to an assistant you would be happy to leave unattended on a Friday night is your metric, the managed build is the shorter path for a small team without a support operations function. For a team that has one, the calculation is different and Fin's head start on install is a genuine advantage.

Who should honestly choose Intercom Fin instead of BotSupport?

These are not polite concessions. They are the situations where Fin is simply the better product, and a comparison page that cannot say so is marketing rather than analysis.

If Intercom is already your inbox, the integration argument is close to decisive. Fin sits inside the tooling your agents use, escalation is native, and Copilot assists the humans on the same conversations. Adding a separate assistant beside it creates two sources of truth and two places to fix a wrong answer.

If you are on Shopify and the bulk of your volume is post-purchase order actions rather than pre-purchase product questions, Fin's Procedures and native Shopify integration let it complete work that BotSupport would hand to a person, subject to how you configure it. That is a real capability gap and it matters most in high-volume, high-returns categories such as fashion and footwear.

If your customers reach you across many channels and languages, Fin's breadth is hard to match. Intercom describes the support agent as spanning live chat, email, WhatsApp, SMS, phone and Slack, and its ecommerce solutions page reports support for 45 or more languages. A single-channel web assistant is not a substitute for that, and we would not pretend it is.

If you have a support operations function that wants to own prompts, routing rules, deflection thresholds and reporting, a managed service will feel like a constraint rather than a relief. Buy the software and use the people you already have. The managed model earns its keep by replacing a role you do not want to fill, and if you have already filled it the argument evaporates.

And if you are procuring at enterprise scale, with security questionnaires, SSO, data processing agreements and contractual uptime, a company that has signed a definitive agreement with Salesforce is a far more straightforward answer to your risk committee than a UK product at Early Access stage with no track record. We would give the same advice to a friend.

What does the Salesforce acquisition mean for buyers?

For most ecommerce buyers this is neutral to positive in the short term. A larger owner usually means more engineering resource, a stronger compliance posture and less risk of the product disappearing. Coverage of the deal described Fin as bringing a customer base of roughly 30,000 companies to Salesforce, which is not the profile of a product about to be quietly retired.

Note the timing carefully, because it is often reported loosely. The announced expectation is completion before the end of Salesforce's fiscal fourth quarter of 2027, which is a fiscal quarter and not the same thing as the opening months of the 2027 calendar year. If you are deciding how long a contract to sign, that distinction is the one that matters, so read the press release rather than the secondary coverage.

The uncertainty is directional rather than financial. When an acquired agent platform is absorbed into an acquirer's own agent platform, product priorities tend to follow the acquirer's core market. Salesforce's core market is not a fifty-product WooCommerce shop in the Midlands. If you are a small ecommerce business, it is reasonable to ask where your use case sits on that roadmap two years out, and reasonable to prefer a shorter initial term while you find out.

None of this should stop you buying Fin if it fits. It should inform the term you are willing to commit to. This page is not updated in real time, and repricing after a change of ownership is the most likely near-term change of all, so check Intercom's pricing page for the current position rather than relying on the figures quoted here. Last reviewed August 2026.

How do you trial both fairly, or move from one to the other?

A fair trial is mostly about the test set. Pull the last two hundred conversations and pick the questions that actually recur: does this come in a size 12, is it in stock, when will it arrive in Ireland, can I return it after 30 days, what is the difference between these two models. Include three questions you know are awkward, such as a discontinued line or a policy exception. Vague prompts like tell me about your returns policy will make every assistant look competent and tell you nothing useful.

Score two things separately. Accuracy against your live data, meaning did it quote the right price, the right stock position and the right delivery window. And escalation behaviour, meaning did it say it did not know and pass to a human cleanly, or did it invent something plausible. The second matters more than the first, because a confident wrong answer costs you a customer, and on a per-outcome model an abandoned conversation may also be a billable one.

Practical trial mechanics. As of August 2026 Intercom advertised a 14-day Fin trial with no credit card required. Check the current terms, including any usage caps, on Intercom's own site before you plan around them, because trial terms change more often than almost anything else on a software website. If you want Fin on a helpdesk other than Intercom's, ask Intercom directly whether a self-serve trial is available on that path and what contract term applies, because the buying route differs from the standard self-serve flow. Ask before you invest the time in a test set.

If you are switching in either direction, the mechanics are lighter than people fear, because neither product owns your store data. Removing the Intercom Messenger means disabling the app embed in your Shopify theme or removing the snippet from your site. Before you switch, export your conversation transcripts, because those recurring questions are the most valuable input to whatever you build next. Confirm your notice period and whether any annual commitment applies, in writing, on both sides. That is not a comment on either vendor. It is what you should do with any support tool you are about to make load-bearing.

Sources

1. Intercom pricing page — Fin at $0.99 per outcome; helpdesk seat plans published at $29, $85 and $132 per seat per month on annual billing; 14-day free trial with no credit card required. Accessed August 2026.
2. Fin pricing page (fin.ai) — Per-outcome pricing and plan structure for Fin, including deployment on non-Intercom helpdesks. Accessed August 2026.
3. Fin help centre: Fin AI Agent outcomes and resolutions — Verbatim definitions of Resolution and Procedure handoff; the 24-hour assumed resolution rule; charged once per conversation; escalations by default behaviour or workspace rules not charged; $49 per month base fee including 50 resolutions with overages at $0.99. Accessed August 2026.
4. Intercom Help: Fin for Ecommerce explained — Shopify described as the supported ecommerce provider for the role; full shopping capabilities on the web Messenger; support-only responses on iOS and Android; email, social, voice and Slack listed as not currently supported for the ecommerce role. Accessed August 2026.
5. Intercom Help: set up Fin for Ecommerce on a Shopify store — Onboarding flow: connect Shopify for a live catalogue connection, install the Messenger, preview and deploy. Accessed August 2026.
6. Intercom blog: Announcing Fin for Ecommerce — Launch post, May 2026, describing the role as purpose-built for Shopify merchants and combining shopping assistance with post-purchase support.
7. Fin ecommerce solutions page — Vendor self-reported figures: 67% average resolution rate, many ecommerce brands at 85% or higher, 45 or more languages. Shopify and Stripe named among direct commerce integrations, with no-code connectors for other systems with an API. Not independently audited. Accessed August 2026.
8. Intercom on the Shopify App Store — Public app listing with rating and review count; check for current figures, which move weekly. Accessed August 2026.
9. G2: Fin by Intercom reviews — Around 4.5 out of 5 across several thousand reviews when checked in August 2026. Read recent reviews directly for current sentiment on cost predictability rather than relying on our summary.
10. Macha: Intercom Fin AI Agent complete guide and pricing (2026) — Independent third-party assessment of knowledge base dependency, stacked seat plus usage costs, and the buying route for non-Intercom deployments. Third-party commentary, not vendor documentation.
11. Salesforce newsroom: Salesforce signs definitive agreement to acquire Fin — Press release dated 15 June 2026: approximately $3.6 billion, completion anticipated before the end of Salesforce's fiscal fourth quarter of 2027, subject to regulatory approval and customary conditions.
12. Salesforce investor relations: definitive agreement to acquire Fin — Investor-facing version of the 15 June 2026 announcement, used to corroborate the deal value and the fiscal fourth quarter 2027 closing expectation.
13. Salesforce Ben: Salesforce acquires Fin, adding 30K AI customers — Third-party coverage describing Fin's customer base at acquisition as roughly 30,000 companies and positioning Fin alongside Agentforce. Trade press, not a company statement.

Details on Intercom Fin reflect its public documentation and independent reviews as of August 2026 and change frequently. This comparison is published by BotSupport, so read it as a vendor comparison and check the alternative for yourself. If something here is out of date, tell us and we will correct it.