BotSupport vs HubSpot Chatbot: which suits an online shop?

What is HubSpot Chatbot, and what does it actually do?

HubSpot's chatbot builder page describes a no-code, template-driven tool for qualifying leads, booking meetings and triaging service requests using if/then branches. It is free to start, and it is CRM-native: what a visitor tells the bot lands on the contact record, and what the CRM already knows about that contact can personalise the reply. For a business that already lives in HubSpot, that two-way sync is the whole point, and it is a real advantage that standalone chat tools generally have to bolt on through an integration rather than get for nothing.

The AI side is a separate product. Breeze Customer Agent draws on content sources you assign to it, typically HubSpot knowledge base articles, website pages and uploaded files, and it can take actions through connected systems, including order status lookups where an integration supplies the underlying data. HubSpot's knowledge base article on the customer agent listed it, when we checked in August 2026, as available with Professional or Enterprise tiers of Marketing Hub, Sales Hub, Service Hub, Data Hub, Content Hub, Smart CRM and Revenue Hub, with HubSpot Credits consumed on top. Hub naming, packaging and eligibility all move, so treat that list as a snapshot and confirm the current requirements page before you plan a purchase around a specific hub.

Channel coverage is a genuine strength. As of August 2026, HubSpot's knowledge base describes the customer agent being assigned to channels connected to the conversations inbox or help desk workspace, including live chat, email, Facebook and WhatsApp, with further channels such as calling described as in beta at the time of writing. That is broad coverage from a single configuration, and it is a real strength of the platform rather than a marketing line. Beta status in particular changes quickly, so treat any individual channel as something to confirm rather than assume.

Two limitations are worth stating plainly, with a date attached to each. As of August 2026, the free chatflows builder is rule-based rather than AI, so a visitor who types a question outside the flow you built does not get an intelligent answer, and the widget carries HubSpot branding on free tiers. Chatimize's independent review of the HubSpot chatbot makes the same criticism, singling out the absence of intent recognition on the free builder and the thin default analytics. HubSpot has been steadily pushing AI features down into lower tiers, so check HubSpot's current plan comparison rather than relying on this paragraph six months from now.

The second limitation is architectural rather than commercial, and it follows directly from how the product is documented to work. Breeze Customer Agent answers from the content sources you assign it. If your help content is thin, outdated, or lives somewhere the agent cannot read, then the agent has nothing accurate to answer from, and the predictable results are wrong answers and escalations to a human. That is not a criticism of HubSpot's model quality, it is arithmetic about the inputs, and it applies just as much to every other retrieval-grounded assistant on the market, ours included. The difference between vendors is not whether grounding matters. It is who is responsible for keeping the grounding good.

Who does the building and the upkeep work?

Every chatbot comparison eventually collapses to one question: after the demo is over, who sits down and does the work? HubSpot's answer is you, or a partner you pay to do it for you. Creating the agent itself takes minutes, and HubSpot deserves credit for that. Getting it to a standard you would put in front of paying customers takes considerably longer, because the agent is only as good as the content it is pointed at. In practice that means auditing the help content you already have, writing the articles you do not, restructuring the ones that were written for humans skimming rather than for retrieval, and then revisiting all of it every time a policy or a product line changes.

BotSupport's answer is that we do it. The build is done for you against your live store, and we test it against the real questions from your own inbox before handover rather than against a script we wrote ourselves. Upkeep is part of the subscription rather than a recurring task on somebody's list. We want to be careful about how strongly that is stated: with essentially no customers yet, we have no delivery record to average and no benchmark to quote. It is a delivery model claim, not a technical superiority claim, and it is why we think a small team without spare hands gets to a working assistant sooner when the build is somebody else's job.

It is worth being concrete about what upkeep actually consists of, because it is the part that gets waved away in both directions. A shop assistant goes stale when the delivery cut-off changes, when a courier is dropped, when the returns window moves from 28 days to 14, when a product line is discontinued but the FAQ still mentions it, and when a seasonal promotion starts and ends. None of those are hard changes on their own. The difficulty is that they arrive irregularly, they arrive from different people in the business, and nobody owns noticing them. A self-serve tool hands that ownership to you by default. A managed service takes it on by contract.

The trade-off runs the other way too, and it genuinely matters. Self-serve means you can change a flow at eleven at night without asking anyone or waiting for a reply. If you have a person in-house who enjoys that control and has the time for it, HubSpot's model gives you more of it, and you should weight that heavily. Managed means you gain the time back and give up some of the immediacy. Neither is objectively better. They suit different teams, and the honest question is which describes yours.

How does each one know your products, prices and stock?

For a support desk, knowledge base grounding is the right architecture, and it is the architecture almost every helpdesk AI has converged on for good reasons. For an online shop it is only half the job, because a large share of pre-purchase questions are catalogue questions rather than policy questions: is this in stock in a 12, what is the price now the sale has started, does this come in navy, will it reach the Highlands before Friday and what does that cost. Those answers do not live in a help article. They live in the product data, and they change daily.

This is where HubSpot's ecommerce ceiling shows, and it is worth stating carefully rather than triumphantly. HubSpot's Shopify Data Sync app does sync products, orders and carts into the CRM, which is genuinely useful for marketing, segmentation and reporting. The documentation for that app also lists limitations, and when we checked it in August 2026 those included Shopify product variants not syncing to HubSpot and Shopify meta fields not being supported for sync via the Data Sync app, alongside company sync being tied to Shopify Plus and order sync running one way only. Variants are precisely where size, colour, per-variant price and per-variant stock live.

Read that as a documented limitation of one specific app on one specific date, not as a permanent incapability. Sync limitations are exactly the kind of thing a vendor ships a fix for, and a determined store can move data through HubSpot's API, a partner app or middleware instead. Check the current limitations list before you rely on it either way. Taken as it stands, though, the practical consequence is clear enough: the agent can be excellent on returns policy and order status while being weaker on the exact question that decides whether somebody adds to basket.

BotSupport is designed the other way round. The assistant is built against the live catalogue, so it is working from current products, prices and policies rather than a summary of them somebody wrote months ago and forgot. It is worth being precise about what that claim is and is not. It describes how the product is built. It is not a benchmarked accuracy figure, there is no published evaluation behind it, and BotSupport is at Early Access stage with no meaningful customer base to point at. The architecture is the argument, and you should test it rather than take it on trust.

Does it work on WooCommerce as well as Shopify?

The chat widget itself is platform-agnostic on both sides, so nobody is locked out of anything. The difference is what sits behind the widget. On Shopify, HubSpot's own Data Sync app maps contacts to customers, companies to companies and products to products, with orders and abandoned carts syncing one way into HubSpot. That is what makes order-status style answers and actions practical without custom work. Even there, HubSpot's own documentation lists real constraints, and at the time of writing those included company syncing being available only for Shopify Plus subscriptions, custom field mappings requiring at least a Data Hub Starter subscription, and product variants and metafields being out of scope for the app.

On WooCommerce, we could not find a HubSpot-built connector of that kind when we looked. Stores generally rely on third-party plugins to move customers, orders and product data across, or they build against HubSpot's API themselves. The best known of the plugins is MakeWebBetter's HubSpot for WooCommerce, which lists over six thousand active installations on the WordPress plugin directory, so this is a well-trodden route rather than an exotic one. It does add a vendor, a licence and a maintenance surface, and usually a field-mapping exercise at the start. It works. It is not the same product experience as a first-party sync, and the gap is worth pricing in before you commit.

If you are on WooCommerce and your priority is answering shop questions rather than feeding a CRM, that asymmetry is the single most practical reason to look at ecommerce-specific tools. It is not that HubSpot cannot be made to work on WooCommerce. It is that the effort of making it work lands on you, and the payoff is CRM data rather than better answers in chat.

BotSupport treats WooCommerce and Shopify the same way, because the build is done by a person against your store rather than assembled from whichever connector happens to exist for your platform. That is a deliberate consequence of the managed model rather than a clever piece of engineering, and it is the reason platform choice is a smaller question on our side of the comparison than on HubSpot's.

How do the pricing models compare, including staff time?

HubSpot's outcome-based move is, in fairness to them, a genuine improvement for buyers and deserves to be described as one. HubSpot's April 2026 announcement, corroborated by independent trade coverage at MarTech, described the change from charging per conversation to charging per resolved conversation, with the customer agent metered in HubSpot Credits. HubSpot's documentation defines a resolution narrowly: broadly, the agent must have replied using a content source or performed an action, and no human takes the conversation over within a set window, 72 hours at the time of writing. Credits are consumed only when a resolution occurs. You are not paying for conversations the bot fumbled, which is more than most usage-priced AI tools can say.

We are deliberately not quoting a single per-conversation figure as though it applies to everyone. The credit cost per resolution and the cash value of a credit both depend on your plan and on any credit bundle you have bought, and included credit allowances differ between tiers. HubSpot's pricing and credits documentation is the only place to get a number that is true for your account. Anyone quoting you a flat cents-per-answer figure for HubSpot without asking what plan you are on is guessing.

The cost that catches people out is the entry ticket rather than the meter. The AI agent needs a Professional or Enterprise tier somewhere in your HubSpot stack, and those are seat-priced with a one-time onboarding fee attached. If you are not already on a qualifying plan, the seat subscription plus onboarding is likely to dominate the cost comparison well before any per-resolution charges enter it. Price it on HubSpot's current pricing page for your actual seat count, and remember that list price is a starting point: figures vary by region and currency, and onboarding fees and seat prices are routinely discounted on annual and multi-year contracts. If you already pay for a qualifying plan, the marginal cost of the agent is small and the maths is very good indeed.

Then there is the line nobody budgets: staff time. Writing and maintaining the knowledge content the agent answers from is real work, done by someone whose day it comes out of, and it recurs every time the catalogue, the delivery terms or the returns window changes. It never appears on an invoice, which is exactly why it gets underestimated in a spreadsheet comparison. BotSupport's flat fee exists to absorb that work rather than hand it back to you. That is the trade being offered, stated plainly.

Two honest caveats on BotSupport's own pricing. It is early-access pricing at Early Access stage, so it is not a committed long-term rate and you should not plan three years around it. And a flat fee is the better deal as volume rises, which means the reverse is also true: at genuinely low volumes, an existing HubSpot customer whose included plan credits already cover their handful of monthly resolutions may pay nothing extra at all. If that is you, the honest recommendation is to use what you are already paying for.

How long until you have a working assistant?

Both vendors can quote a short setup time and both are telling the truth about the step they are describing. Creating a HubSpot agent, assigning content sources and switching on a channel really is quick, and anyone who tells you otherwise has not tried it. The point that matters is that setting up the agent and deploying something you are comfortable showing customers are not the same milestone. The distance between them is the state of your help content, and that distance is different for every business.

That is why time-to-live is the metric worth asking about rather than time-to-configure. Our one working day is a target for Early Access builds, measured to a tested, handed-over assistant answering questions about your actual products, and it is a target rather than an observed average because we have not delivered enough builds to have an average. We would rather say that than imply a track record we do not have. The reason we think the target is realistic is structural: the intervening work is somebody else's job, and that somebody does it every day.

If your documentation is already excellent, HubSpot will close that gap fast and the managed argument weakens considerably. If your documentation is a folder of PDFs, a Word document from 2023 and three things only your longest-serving colleague knows, the honest answer is weeks rather than days, and most of those weeks are writing rather than configuring. Be sceptical of any vendor, including this one, who quotes a setup time without asking what state your content is in.

HubSpot does reduce the risk of trialling it, which is worth knowing. HubSpot's April 2026 announcement referenced a free trial of the Breeze agents, and its knowledge base described an initial free-access window during which the customer agent does not consume credits, listed as 14 days when we checked. Trial lengths and eligibility are among the most volatile things any vendor publishes, so confirm the current terms on HubSpot's site rather than on ours. Either way, a free window is the sensible way for an existing HubSpot customer to find out where their own content actually stands before committing credits to it.

Who should honestly choose HubSpot Chatbot instead?

There are several situations where HubSpot is straightforwardly the better buy, and pretending otherwise would not survive contact with a real evaluation. A comparison page written by a vendor is only worth reading if it is willing to say where the other side wins, so here is that list without hedging in our own favour.

The strongest of those cases is the first one. If a qualifying hub is already in your budget, the marginal cost of switching the AI agent on is small, your included credits may absorb a good share of your resolutions, and everything the agent does lands in the CRM you already run. That is a genuinely good position to be in, and no amount of catalogue grounding on our side changes the arithmetic for you.

BotSupport is also not the right answer for a company that needs one vendor, one contract and a formal onboarding programme. We are Early Access, UK-based and small, with essentially no customer base yet. For enterprise procurement, security review, vendor due diligence and multi-hub consolidation, HubSpot wins that comparison on every axis those buyers care about, and it is not a close contest.

• You already run Marketing, Sales, Service, Content or Data Hub at Professional or Enterprise. The agent's marginal cost is small and your included plan credits may cover much of your volume.
• Chat is a sales and CRM instrument, not just support. Lead qualification, meeting booking and writing straight to contact and deal records are things HubSpot does natively and does well.
• You need multiple channels under one agent. Live chat, email, Facebook and WhatsApp from one configuration, with further channels in beta as of August 2026, is broad coverage. BotSupport is website chat only.
• Your questions are mostly non-catalogue. Account access, billing, policy and order status answer well from a knowledge base, and order status can generally be answered where the Shopify sync or another integration supplies the data.
• You want hands-on control and hourly iteration in-house rather than a managed service, and you have the person and the time to do it.
• You are on Shopify Plus and want company-level syncing, which HubSpot's documentation tied to that tier at the time of writing.
• You need enterprise procurement, formal onboarding, a partner ecosystem and an established vendor with a long track record.

How should you trial both, and what does switching involve?

A fair test is easy to design and most people skip it in favour of a demo. Pull twenty genuine questions from your last month of emails and chats, and make sure the mix reflects reality rather than the demo script: several variant-level product questions, a couple on stock, one on a price that changed when the sale started, delivery to an awkward postcode, a returns edge case, an order status query, and two deliberately ambiguous ones where the right behaviour is to ask a clarifying question or hand over. Score each answer correct, wrong or escalated. The number that matters is how many were answered correctly without a human touching them, and the second number that matters is how many wrong answers were delivered confidently.

Run the same twenty through both tools on the same afternoon. Comparing a tool you tested in January against one you tested in June tells you mostly about how your own catalogue changed in between. Involve whoever normally answers those emails, because they will spot a subtly wrong answer that a founder skimming a transcript will wave through.

There is very little switching cost in either direction, which is worth knowing before you agonise over the decision. Both are embedded as a widget, so removing either one is a script change rather than a migration. Keep exports of your transcripts whichever way you go, because they are the best source material for improving whatever assistant you end up keeping. If you are already a HubSpot customer, note that your CRM, your tickets and your knowledge base are entirely unaffected by trialling something else on the website widget. You are testing a chat layer, not replacing a system of record.

Two practical warnings on the HubSpot side, both taken from its own documentation as it stood at the time of writing. Resolution status is evaluated after a set window following the last visitor message, 72 hours when we checked, so your first week of analytics will lag reality and a trial judged on day three will read as worse than it is. Confirm the current definition in HubSpot's documentation, because it determines both what you are charged for and how quickly your reporting catches up. And HubSpot's knowledge base states that when credits run out the agent temporarily stops being assigned to new conversations across connected channels until they reset or are topped up, with conversations routing to human agents instead. That can look like a quality collapse mid-trial when it is a billing event.

Where is BotSupport the weaker choice as of August 2026?

Being specific about the gaps is more useful than a summary that admits nothing in particular. BotSupport has no track record to point at yet, no published customer numbers, no awards and no case studies, because it is pre-launch. Any vendor at that stage carries risk, and you should treat every forward-looking sentence on this page accordingly. The free trial period exists precisely so that you can test the product rather than take a claim on faith, and that is the only sensible basis on which to buy from a new company.

It is also deliberately narrow. There is no CRM, no ticketing system, no marketing automation, no deal pipeline and no multi-channel inbox. If you need any of those, you need a platform, and HubSpot is one of the best in the market at being one. BotSupport answers shop questions on your website, built by a person against your live catalogue, and hands over to your team when it should not answer. That is the whole scope, and we would rather be clear about the boundary than let you discover it after signing.

The fully managed model has a cost as well as a benefit, and it is the same fact viewed from two sides. There is no DIY tier, so if you want to log in and rewrite a response yourself at midnight, that is not the product we have built. Some teams find that liberating and some find it constraining. It is better to know which you are before you trial anything, because it is the one difference here that no feature comparison will resolve for you.

Finally, a note on how to read this page at all. It is written by one of the two vendors being compared, and the incentive is obvious. That is why the HubSpot claims are attributed to HubSpot's own documentation with dates on them, why the figures we could not confirm have been softened rather than sharpened, and why our own delivery time is described as a target rather than an average. Check the sources. If any of them no longer say what we say they said, believe them and not us.

Sources

1. HubSpot: Free Chatbot Builder (product page) — Vendor page, checked August 2026. Confirms the builder is rule-based (if/then branches), free to start and CRM-integrated, with premium features in paid hubs.
2. HubSpot: Customer Agent and Prospecting Agent, now you pay when the task is complete — Vendor announcement, April 2026. Source for the move from per-conversation to per-resolved-conversation pricing in HubSpot Credits, and for the referenced free trial of the Breeze agents. HubSpot's self-reported performance figures in that announcement are not relied on here.
3. HubSpot Knowledge Base: Understand the customer agent — Checked August 2026. Source for the Professional or Enterprise hub requirement, the definition of a resolution, the 72-hour evaluation window, the free access window during which credits are not consumed, and the behaviour when credits run out.
4. HubSpot Knowledge Base: Connect HubSpot and Shopify via data sync — Checked August 2026. Source for which objects sync, one-way order sync, product variants not syncing and Shopify meta fields not being supported by the app, company sync limited to Shopify Plus, and Data Hub Starter being required for custom field mappings. Limitations change, so check the current list.
5. HubSpot: Service Hub pricing — Checked August 2026. Source for per-seat pricing, included credit allowances and one-time onboarding fees on that date. List pricing only: figures vary by region, currency and contract, and are frequently negotiated.
6. MarTech: HubSpot moves to outcome-based pricing for some Breeze AI agents — Independent trade coverage corroborating the 2026 pricing change from per conversation to per resolution.
7. Chatimize: HubSpot chatbot review — Named independent review. Critical of the free rule-based builder for lacking intent recognition, limited channel coverage on the free tier, and thin default bot analytics.
8. WordPress.org: MWB HubSpot for WooCommerce (MakeWebBetter) — Checked August 2026. Third-party WooCommerce to HubSpot connector listing over 6,000 active installations. Supports the point that WooCommerce integration is community-built rather than HubSpot-built.
9. eesel AI: Breeze Customer Agent review — Third-party review written by a competing vendor, so read with that in mind. Useful on knowledge sources, setup steps and customisation limits. Its pricing figures predate the 2026 change and were not used here.

Details on HubSpot Chatbot 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.