E-commerce chatbots built around your shop
Help customers find the right product, understand delivery and get useful answers—with a clear route back to your team.
From Chester, we build e-commerce chatbots and automation around your catalogue, policies and customer journey. We assess Shopify, WooCommerce and other integrations before agreeing what the system can access or do.
Illustration. See the live retail project and documented walkthrough below.
Useful answers. Clear boundaries.
Product information, agreed permissions and a human handover route.
Start with the friction in your own shop
Choose whether clearer content, a simpler form or an assistant is the right response. A useful pilot solves a specific problem you can measure.
Questions before buying
Unclear delivery charges, product details or returns information can put shoppers off. Review the journey before assuming a chatbot is the answer.
Make the information easy to find, then test where an assistant adds value.
“Where is my order?”
Repeated order questions take staff time. A connected lookup can help if the status is reliable and the customer is authenticated before private information is shown.
Agree the source, access controls and fallback for uncertain tracking.
Returns and exceptions
A returns enquiry needs the right policy, product details and next step. Explain the rules and collect the request without promising an automatic refund.
Keep refunds, cancellations and unusual cases behind agreed approval rules.
Questions outside office hours
Shoppers can need help when your team is offline. An assistant can offer useful public information, but customers also need to know when a person will reply.
Set clear expectations and a working route back to the retail team.
How an e-commerce chatbot fits the journey
Illustrative workflows, not default permissions: we choose the useful stages for your shop and test them before launch.
Step 1
Find a product
Use the agreed catalogue to find relevant products and categories. Test the links, suitability information and what happens when an answer is uncertain.
Example: A shopper asks for guppies; the assistant offers the appropriate category.
Step 2
Answer a question
Bring delivery and policy information into the conversation. Use approved sources and acknowledge missing details.
Example: A delivery question leads to the current delivery page, not an invented promise.
Step 3
Connect a task
Use authenticated, limited access where private information is involved. Agree which actions require approval.
Example: An order enquiry requires an identity check before showing private order details.
Step 4
Bring in the team
Give unresolved enquiries a named owner and a realistic response expectation. Agree what context can be passed on.
Example: An unusual return becomes a support request for a person to review.
Decide what the assistant is allowed to do
Public answers, private customer information and actions with financial consequences need different controls.
Public information
Product, delivery and policy answers should use the agreed source. Test stale information, broken links and questions outside the catalogue.
Private information
An order lookup needs suitable authentication and limited permissions. Knowing an email address must not be enough to reveal another customer’s information.
Actions and exceptions
Refunds, cancellations and stock changes need explicit rules and approvals. Define the fallback when a connection fails or a customer needs a person.
Choose a useful starting point for your business
These are examples to discuss, not a promise that every feature is included or every marketplace can be connected.
Specialist online stores
Start with product, delivery and care questions that repeat in your inbox. Specialist advice needs reliable source material and clear limits.
- Product and category discovery
- Delivery and policy answers
- Uncertain advice routed to staff
Direct-to-consumer brands
Explain products in your brand voice while making it clear that shoppers are using an automated assistant.
- Product education
- Size or specification information
- Customer feedback and handover
Multi-channel sellers
Connect selected channels where their APIs and permissions allow it. Keep a clear owner for unresolved enquiries.
- Channel-specific support routes
- Agreed stock-data sources
- Consistent policy information
Supplier-led fulfilment
Help shoppers understand lead times and route delivery exceptions to the right person. Supplier data determines what can be answered reliably.
- Lead-time explanations
- Updates from connected systems
- Delivery exceptions for staff
Check the fit with your existing tools
We assess APIs, product editions, access permissions and costs before committing to an integration. These are areas to investigate, not a blanket compatibility claim.
Shopify
Assess the storefront, product catalogue, app access and any customer authentication. Our TFC case study shows a real Shopify-connected implementation.
WooCommerce
Check the WordPress setup, API access, plugins and order permissions. Compatibility and maintenance depend on the specific store.
Helpdesk
Agree ticket creation, the context passed to an agent and who owns the response. Confirm the helpdesk edition and available integration.
Email and messaging
Assess permitted channels, recipient permissions, opt-outs and usage costs. Marketing follow-ups are not enabled simply because someone used a chatbot.
Using a different platform? Bring the name of your storefront, helpdesk and the task you want to connect. Ask about your setup.
See a real assistant before discussing yours
Tropical Fish Co is our founder's own shop: an owner-led retail project, not an independent client. Its case study lets you inspect the work and a dated assistant walkthrough, not just a list of promised benefits.
Text chat · checked 21 September 2026
Delivery information and product discovery
In two text-chat checks, Aqua found delivery, contact and FAQ pages, then the guppy shopping category with related links. The destinations were checked and matched the shop.
This demonstrates information finding. It does not establish support savings, an automation percentage, voice quality or private-order handling.
See the walkthrough and visit the shopAgree a baseline before a pilot
Track whether answers are correct, which questions need a person, repeat contacts and staff time spent resolving the same tasks. Include usage and support costs when assessing value.
Search visibility is a separate measurement. Our retail case studies show dated Search Console figures and their limits; they are not proof of chatbot revenue or saved wages.
E-commerce chatbot questions, answered
What can an e-commerce chatbot do for my shop?
It can help shoppers find products and answers in your agreed catalogue, delivery information and policies. Connected workflows can also support order enquiries or create a support request. We agree the allowed tasks, information sources and route to a person before building it; an assistant does not automatically have access to your orders or permission to change them.
Can a chatbot help customers who leave before buying?
A useful answer about delivery, product suitability or returns can remove uncertainty, but a chatbot cannot fix every reason someone leaves. We review the shopping journey and test a specific improvement. Any effect on completed purchases needs reliable analytics and a comparable baseline, not a promised percentage. Marketing follow-ups also require appropriate channel permissions and opt-outs.
When is AI useful, and when is a simple rule better?
AI can help interpret a shopper’s question and find relevant information. Clear rules are often better for eligibility, permissions and actions such as changing an order. We combine these where useful, test uncertain questions and keep a human route available. The assistant should say when it cannot establish an answer rather than invent a policy.
Can the assistant manage returns or issue refunds?
A returns workflow can explain the policy, collect an authenticated request and prepare the information your team needs. Labels and status updates depend on the connected platform and agreed permissions. Refunds, cancellations and unusual cases stay behind explicit approval rules. The public TFC walkthrough does not demonstrate a completed refund or private-order action.
Can you work with Shopify or WooCommerce?
We assess your storefront, apps, API access and permissions before agreeing the integration. Tropical Fish Co is a live example of our Shopify-connected storefront and shopping-assistant work. That does not establish compatibility with every Shopify or WooCommerce setup. We confirm product-data access, any customer authentication and the fallback if a connection fails.
How much does an e-commerce chatbot cost?
Our AI chatbot plans are Website at £99/month (up to 1,000 conversations), Business at £149/month (website and WhatsApp, up to 3,000 conversations, hand-over to a person) and Pro at £249/month (more channels, CRM integration, up to 10,000 conversations), plus a one-off £149 setup. Month to month, cancel with 30 days’ notice. Custom shop integrations are quoted, and wider automation is from £399 per workflow plus £49/month for hosting and monitoring. No VAT to add. Hand On Web is not VAT-registered. Any third-party costs are confirmed in the proposal. We estimate value from your actual workload; we do not promise a two- or three-month payback.
Can it recommend products without making things up?
Recommendations should use the agreed catalogue and product information. We test whether suggested products match the question, whether links work and what happens if stock or suitability is uncertain. Sensitive product advice and missing information need clear limits. Purchase-history personalisation requires appropriate access and permissions; it is not part of every installation.
What evidence can I see before commissioning a project?
The Tropical Fish Co case study documents a real text-chat walkthrough that found delivery, contact, FAQ and guppy-category pages, with the destinations checked. Our separate retail case studies show dated Search Console results. Those search figures demonstrate search visibility, not chatbot sales uplift or support savings. For your pilot, we agree measures such as correct answers, useful handovers and time spent by staff.
Can customers use the assistant outside office hours?
An online assistant can be available beyond your staffed hours, subject to its hosting and connected services. We explain when a person will respond, how a failed connection is handled and what should happen when the assistant cannot help. Around-the-clock availability is a design goal, not a promise that every query is resolved or every service has perfect uptime.
How do you launch and keep improving the system?
We start with one useful task, agree the source information and access, then test common questions and failure cases before a pilot. The timetable depends on integration complexity and the quality of the source material. After launch, agreed reviews use conversation outcomes, staff feedback and consented analytics to refine answers and handover. Changes to prices, policies or products need an update process.
Explore the work behind your shop
Tropical Fish Co Case Study
Our founder’s own shop: 12,693 Google Search clicks, up 92.3% against the preceding 28 days. All countries and devices; 22 August–18 September 2026.
Learn about Tropical Fish Co Case StudyE-commerce SEO
Search research, useful categories and product information for online retailers.
Learn about E-commerce SEORetail Automation
Connect useful shop workflows, reporting and stock alerts.
Learn about Retail AutomationRAG Chatbots
Assistants grounded in agreed business information, with testing and handover.
Learn about RAG ChatbotsView Pricing
Explore advertised starting points and discuss the scope for your shop.
Learn about View PricingWhich customer question keeps coming back?
Bring your shop, the tools you use and one task that takes too much time. Our Chester team can discuss whether clearer content, an assistant or a connected workflow would help—and what we would need to test.
Discuss Your Store