If you are comparing ecommerce AI assistant pricing, the cheapest-looking plan is not always the safest plan to launch. A useful quote should explain what the assistant is expected to answer, how much product and policy knowledge it needs, how often that knowledge changes, and how much launch support is included.
Start with the job the assistant has to do
A simple website assistant that answers opening hours, delivery policy, service FAQs, and contact questions is a different workload from a product advisor that compares products, handles sizing or compatibility questions, and uses catalogue data. Pricing should reflect that difference, not just the presence of a chat box.
Monthly chat volume still matters
Chat volume is the easiest number to understand because it follows site traffic. A low-traffic store can usually start with a smaller allowance. A busy store, or a store placing the assistant on high-intent product pages, needs more monthly conversations and stronger review controls. The important point is that chat volume is only one part of the cost model.
Knowledge base size changes the work
An assistant can only give reliable answers when the right knowledge is connected and kept tidy. Pricing should account for the number of website pages, FAQs, policy documents, guides, manual answers, and approved files that need to be imported and reviewed. Thin knowledge produces thin answers; better source material takes more setup but creates a safer customer experience.
Product catalogue complexity is the big ecommerce variable
For ecommerce, the catalogue often matters more than the page count. A store with 50 simple products has different needs from a store with thousands of SKUs, variants, dimensions, materials, compatibility notes, product-feed rows, and changing availability. If customers ask detailed product questions, the plan should include enough product records and attributes to support those answers.
Freshness and refresh frequency affect reliability
Prices, stock, product descriptions, and policies can change. A plan should be clear about how often knowledge can be refreshed and what happens when a source is blocked, stale, or missing. Weekly refresh may be enough for some brochure-style websites. Product-heavy stores may need more frequent refreshes and a review loop for weak or unsupported answers.
Managed setup is not just installation
Managed setup should cover the work that makes the assistant safe to show to real visitors: product-feed mapping, source cleanup, assistant instructions, fallback wording, test questions, widget installation checks, and launch QA. That work is especially valuable for technical catalogues, high-SKU stores, or any business where a wrong product answer could create support issues.
Questions to ask before choosing a plan
Before choosing a package, ask:
- How many monthly visitor conversations do we expect?
- How many pages, FAQs, documents, or manual answers need to be approved?
- How many products and product attributes should the assistant use?
- How often do prices, stock, delivery details, or specifications change?
- Do we need product-advice behaviour, or just general website support?
- Who reviews weak answers and improves the knowledge base after launch?
A practical way to decide
Choose a website AI plan when the assistant mainly answers general site, service, policy, or lead-capture questions. Choose a Product Advisor plan when visitors need help comparing products, narrowing a catalogue, checking fit or compatibility, or understanding technical product information. If the product data is messy or commercially important, include managed setup instead of treating launch as a copy-and-paste widget job.
Next step
If you are not sure which route fits, start by mapping your top customer questions against the product and policy data needed to answer them. That gives a clearer plan choice than chat volume alone.