Store support is three questions, not one
Read a week of a store's inbox and almost everything falls into three buckets. Pre-purchase questions about a product — will this fit, what is it made of, is it in stock. Order questions — where is it, can I change the address, can I cancel. And policy questions — returns, shipping times, warranty.
Only the third is answerable from a help centre. The first needs your catalogue, live. The second needs your orders. A support tool that indexes your FAQ page and stops there will deflect the smallest of the three buckets and hand you the rest, which is why generic chatbots disappoint on stores specifically.
So the question to ask of each tool below is not how good the model is. It is what the model can see.
The shortlist
Prices and plan contents below are what each vendor's own public pricing page stated on 20 September 2026. Support pricing moves fast — check theirs before you commit.
helpo
- Best for
- Stores that want catalogue-aware answers with citations, a human one click away, and the same setup working on WooCommerce.
- What you get
- Products are imported from the storefront's public API — titles, descriptions, prices, variants and stock all become answerable — and the AI can search that catalogue mid-conversation and render product cards in the chat rather than describing a product in prose. Order data is pushed to helpo by the app rather than pulled with a stored credential, so there is no admin token to leak.
- The catch
- It is not a full CX suite: no voice, no SMS, and no refund or cancellation actions. When a customer needs their order changed, the AI files a ticket and hands over — it does not reach into your store and do it.
Gorgias
- Best for
- Established stores at real volume who need the helpdesk to take actions in Shopify — refunds, cancellations, order edits — not just answer questions.
- What you get
- Their pricing page states the AI Agent is on every plan and billed only when it resolves a conversation, and that the helpdesk scales from 50 to 5,000 tickets a month and is never priced per agent. Voice and SMS are add-ons that scale with volume.
- The catch
- It is built for the volume it is priced for. A store doing forty tickets a month is buying a suite it will not use, and the per-resolution model means your best month is also your biggest invoice. This is the honest incumbent answer: at a few thousand tickets a month, Gorgias is hard to beat.
Shopify Inbox
- Best for
- Stores whose support is entirely on-Shopify and who want something working this afternoon for nothing.
- What you get
- Free, first-party, and rated 4.6 across more than five thousand reviews. It connects to your catalogue, inventory and policies automatically, lets you set the agent's tone and rules, personalises replies for buyers signed in with Shop using their purchase history, and improves as you rate chats.
- The catch
- It works with Shopify Admin and your online store, and that is the boundary. No email channel, no ticket lifecycle, no WooCommerce, and nothing for the part of your business that is not the storefront.
Tidio
- Best for
- Small stores that want visual flows for the predictable journeys and an AI agent for everything else.
- What you get
- Starter is $24.17/mo billed annually with 100 billable conversations and 50 one-off Lyro AI conversations. Lyro is also sold standalone from $32.50/mo for 50 AI conversations, and higher tiers offer pay-per-resolution billing.
- The catch
- The AI is an add-on with its own meter, so the headline plan price is not the price of an answering bot. The free tier is no longer shown on their plan table — check their pricing page.
Richpanel
- Best for
- Stores switching off a per-agent suite who want the same shape of product for less.
- What you get
- It positions itself directly as the Gorgias and Zendesk alternative. The pricing page lists an entry Incubator tier at $9/month for three agents and per-agent plans from around $59 per agent per month.
- The catch
- Their page states all subscriptions are annual contracts, and the same tier appears at different figures depending on which pricing view or calculator you land on. Get the number you were quoted in writing before signing a year.
Zendesk
- Best for
- Stores that are one channel of a larger business, where support already spans retail, wholesale and a contact centre.
- What you get
- Per-agent suite tiers with AI resolutions billed separately from seats.
- The catch
- The pricing page geo-redirects and quotes the entry plan in local currency, so read it in yours rather than trusting a figure from an article. Shopify is an app in a marketplace here, not a first-class model — expect integration work.
Intercom
- Best for
- Stores with a product as well as a catalogue, where the same tool has to serve logged-in users and shoppers.
- What you get
- Essential is $29 per seat per month, Advanced $85, Expert $132, each with the Fin AI Agent at $0.99 per outcome. Fin can also run on top of another helpdesk entirely.
- The catch
- Two meters — seats and outcomes — so scale shows up twice on the invoice. Commerce is an integration rather than something the model understands natively.
Choosing by volume
Store support tooling is priced on ticket volume far more than on features, so volume is the cleanest way to narrow the list.
- Under about 100 tickets a month: Shopify Inbox or a free plan. Paying for routing and SLAs at this volume is paying for shelf space.
- 100 to 1,000: pick on knowledge. This is where catalogue-aware retrieval with citations earns its price, because the pre-purchase bucket is now big enough to deflect.
- Over 1,000, with lots of refunds, cancellations and order edits: pick on actions. Gorgias exists for this and is honestly the right answer if writebacks into Shopify are most of your work.
- Multi-store or multi-platform: check the data model, not the integrations page. A tool that treats one Shopify store as the tenant boundary will fight you on the second one.
One test that separates them
Before you commit, ask each tool the same question a shopper would: is the blue one in stock in large, and how long does delivery take. A tool that knows your catalogue answers it in one turn. A tool that indexed your FAQ page answers the delivery half and guesses the rest, or tells the customer to check the product page — which is the page they were already on when they opened the chat.
Then ask something the tool should refuse: a question your documentation genuinely does not answer. The right behaviour is an admission and a hand-off to a person, not a confident paragraph. A model that invents a returns window costs more than the ticket it deflected.