BigCommerce Companion: What the Built-In AI Assistant Does

BigCommerce Companion is an AI assistant built into the BigCommerce control panel. For a small store it answers how-to questions from BigCommerce documentation, draws charts of your store analytics, lets you browse the catalogue in a chat, takes you to the right settings page, and updates order statuses after you preview and approve. It does not talk to shoppers.

BigCommerce's page says Companion is included with your store, during free trials and on paid subscriptions, with no extra price listed. That page describes features; it cannot tell you whether Companion saves you an hour a week, which depends on what your store does on a typical Tuesday. The ten-task test further down answers that on your own store, with a scoring sheet. I have not run it on a live store, so it sets out what to check, not what happened.

Follow me on Instagram@sagnikteaches

What BigCommerce says Companion does

From BigCommerce's own page, as of November 2026, Companion has five jobs:

Connect on LinkedInSagnik Bhattacharya

• Answers how-to questions about running your store, drawing on BigCommerce documentation.
• Shows store analytics as on-demand charts, including comparisons of one period against another.
• Browses your product catalogue inside the chat.
• Navigates you around the control panel, so you do not hunt through menus.
• Updates order statuses, but only after you preview and approve the change.

Two other points on the page matter. It is a merchant-facing tool, meaning it helps you, the person running the store; it is not a customer chatbot, and for shopper chat the page points to apps in the BigCommerce App Marketplace. And it makes no changes without your approval. The page also says new capabilities ship regularly and that you can ask Companion what's new, so treat any list, including this one, as a snapshot. One example prompt on the page reads "Issue a store credit to this customer", shown beside a store-credit screen, yet store credit is not among the changes the page lists; test that on a test customer, and look for a preview, before relying on it. The page does not say it edits prices, writes product descriptions or sends emails, so do not expect those until you see them.

Subscribe on YouTube@sagnikteaches

One naming point. There is a separate app called Merchant AI in the BigCommerce App Marketplace, from an outside developer and with its own monthly fee on its listing. It is not Companion, and nothing here describes it. If a search result or a colleague says "the BigCommerce AI", check which of the two they mean before comparing features or prices.

Where it sits among your options

A small BigCommerce store has three kinds of AI help available, and mixing them up causes most disappointment.

The first is the assistant inside the platform, which is Companion. It sees your store, so it needs no exports and no pasted data. The second is a general chat assistant such as ChatGPT or Claude, which does not see your store unless you export or paste data, but can do far more with text: product copy, email drafts, policy wording. The third is shopper-facing AI, such as chat on the storefront, product recommendation quizzes and search, covered in chat, quiz or search for a small store.

The sensible pattern is to use Companion for questions about the store and its numbers, a general assistant for writing, and a customer-facing tool only after your policies are written. If you are still choosing a platform, the store builder comparison of Shopify, WooCommerce and BigCommerce goes through the trade-offs. The Shopify equivalent of Companion is Sidekick, described in how to use Shopify Sidekick, and the two make a fair side-by-side if you run both stores.

A ten-task test on your own store

Official pages describe features; they cannot tell you whether the answer to your question is right. The test below takes about 45 minutes. Use real tasks from your last month, run them in order, and score each result on three points: correct (matches what you know or can check), useful (saved time against doing it by hand) and safe (no change made without a clear preview).

Open the Companion panel and ask the following, adapted to your products. Each task has what to check.

1. A how-to you already know the answer to. Ask how to create a coupon code for 10 per cent off one category. You know the steps, so you can judge accuracy. Check that it points to the right menu and mentions the right options.
2. A how-to you do not know. Ask how to set a shipping rule for orders over a set amount. Follow its answer in a test, and see whether it matches what the control panel offers. If a step does not exist on your plan, note that.
3. A sales chart for a known period. Ask for orders by day for the last 30 days. Compare the total with the figure in the analytics section for the same dates.
4. A period comparison. Ask for revenue this month against last month. Check both totals, and check what it counts as revenue (before or after refunds, with or without shipping).
5. A product question. Ask which products have no image or no description. Spot-check five of the list against the catalogue. Count missed items as well as wrong ones.
6. A catalogue browse. Ask it to show your five products with the lowest stock. Compare with your inventory view.
7. A navigation request. Ask it to take you to the place where you edit your return policy page. Note whether it lands on the right page.
8. An order-status update on a test order. Use a test or old order that is safe to change. Ask it to mark the order as shipped. Check that a preview appears, that it names the right order and the right new status, and that nothing changes before you approve. Then confirm in the order list.
9. A question it should decline. Ask it to change a product's price across a category. If it cannot, you want a clear refusal, not a pretend answer. If it can, check for a preview and use a test product.
10. A vague question. Ask "why were sales down last week?" See whether it asks what you mean, shows a chart, or invents an explanation. A good result shows data and does not claim a cause it cannot see.

Score each task 0, 1 or 2 on each of the three points, for a maximum of 60. As a rough reading: above 45 means it is a useful daily helper for those tasks; between 30 and 45 means use it for navigation and quick questions, but verify numbers; below 30 means leave it for now and revisit when the product has changed. These bands are my suggestion and not a standard. Write down the date, because the product changes.

How the chart tasks should be checked

Charts are the feature most likely to be taken at face value, so test them hardest. An illustrative store: 140 orders in October and 118 in September, a rise of about 19 per cent. Suppose Companion's chart shows 140 and 118 and your analytics report shows the same, then you can trust that question type. Suppose it shows 140 and 110: either the period boundaries differ (a time-zone setting can move midnight orders across days), or the definition of an order differs (cancelled or incomplete orders in or out).

Three checks for any chart you plan to act on. First, same dates: ask the assistant to state the date range it used. Second, same definition: find the control-panel report that matches and confirm the total. Third, same currency and tax treatment, if your store sells in more than one. A mismatch is not proof of an error, but it tells you the number needs a reason before it goes in a decision.

When a chart answers a question you actually have, save the question in a notes document with the date and the result. Over a quarter you will build a short list of questions it answers well, which is what you need to decide how much it is worth to you. For a broader view of where the platform's reports, GA4 and chat assistants each fit, see choosing between store reports, GA4 and ChatGPT for analytics.

A filled-in scorecard for a small gift store

Here is how the scoring might look for an illustrative store selling gifts, about 120 orders a month, one owner and one part-timer. The scores are invented to show the method, not the product's real performance.

Reading it: the store's owner would use it for navigation, charts that have been checked once, and status changes, and would not rely on its product audits without a second look. Notice that the useful insight is in the notes column, "missed 2 of 9", because a list that is nearly right is the most dangerous kind. The owner adds a rule to the team sheet: lists from the assistant are a starting point; counts are checked against the catalogue filter.

Training a new part-timer with it

An onboarding use deserves its own note because it is the lowest-risk, highest-return job. Give the part-timer a list of ten questions to ask Companion on their first day, such as how to find an order, how to refund one, how to add a product, how to create a coupon and where the shipping settings live. They learn the control panel by using it, and you review the answers together before they leave. Where an answer was wrong or unclear, write the correction into your own one-page store guide. Over a month the guide becomes your SOP for the store, and the assistant has shown you where the documentation does not match how your store is set up, which is information a manual never gives you.

Using it for order statuses without making a mess

The order-status update is the one write action named on the page, so it deserves a short procedure. The approval step is what makes it safe, but approval only protects you if you read the preview.

1. Name the order by its number in the request, not "the last one".
2. Read the preview for three things: the order number, the customer's name, the new status.
3. If you are changing several orders, change one first and confirm in the order list before approving the rest.
4. Remember that a status change in many stores triggers a customer email. Check in your store's settings which statuses send one before you use Companion to move a batch.

An illustrative before and after. Before: a store owner opens each of eight orders, changes the status, and saves. After: one request listing the eight order numbers, a preview, a check, one approval. About 20 clicks become four. The saving is small per day and adds up at the end of a busy week. The risk to watch for is a customer email being triggered for the wrong order, which is why the single test order comes first.

Five jobs a small store could give it first

If the ten-task test goes well, these are the jobs to try, in order of lowest risk. Each is illustrative and depends on what your test showed.

Learning the control panel. A new part-time team member asks how to do things instead of interrupting you. Low risk, immediate value.

A Monday numbers check. Three charts: orders last week against the week before, top products, and orders by day. Five minutes instead of opening three reports.

Catalogue hygiene. Ask for products missing images or descriptions each month, then fix them using a general assistant for the copy. The two tools complement each other, and writing product descriptions with AI shows how to brief that second step.

Quick lookups during a customer message. A customer asks about an order; you ask Companion to bring up the product or navigate to the order. Check the answer against the order page before replying.

Weekly status updates. Use the preview-and-approve flow for shipped orders, once your test shows it behaves.

What Companion does not do, and where to look instead

It does not answer shoppers, so questions like "where is my order?" and "can I return this?" stay with you or a separate chat app. If that is your pain point, read whether an AI chatbot can handle order tracking and returns before buying anything. It does not, on the evidence of the official page, write marketing emails, produce ads, or manage inventory forecasts. A general assistant or a dedicated tool is the answer for those.

Check privacy before you paste anything sensitive into any assistant. Companion works inside your own account, and BigCommerce's page says it can only see and change what the signed-in user's permissions allow, so give each staff login the permissions its job needs and no more. The page does not say how long conversations are kept, so read BigCommerce's privacy terms if you hold customer data you are cautious about. For a general assistant, use a business plan where training on your content is off by default.

Finally, plans. BigCommerce's pricing page now lists Core, Growth, Scale and Performance plans. The Companion page says it comes included with your store on free trials and paid subscriptions and names no plan limits, though the pricing page does not list Companion as a plan feature, so confirm it in your own control panel. Availability in free trials makes a test low risk.

When an assistant's answer sounded right and was not

An illustrative case that the test is built to catch. A store owner asks for the best-selling product last month and gets a confident answer. The owner restocks heavily. Later, comparing with the sales report, the owner finds that the assistant ranked by units, and the report ranked by revenue, and the two lists differ. The restock was right for one and wrong for the other. Nothing was broken; the question was ambiguous. The fix is a habit: put the measure in the question ("by units sold" or "by revenue after refunds") and put the date range in as well.

A second case: the owner asks how to set a shipping rule and follows the steps for a menu that exists on a higher plan than the store's. The answer came from documentation that covers every plan. Ask for the answer "on my plan", and check that the menu item exists before you spend time on the rest.

A third: an owner approves a status change from a preview without reading it and marks the wrong order. Catching it takes one minute if you spot it the same day. Reading the preview takes five seconds.

Keeping a log, so you know after a quarter

Keep a simple sheet with four columns: date, what you asked, whether it was right, and the time it saved or cost. After 20 entries you will see which question types work. Add a column for "feature changed?" whenever you notice a new capability, because the page says new ones arrive regularly. At the end of a quarter, add up the time saved. If it is under an hour, it is a convenience. If it is several hours, it has earned a place in your weekly routine and perhaps your onboarding for new staff.

Whatever you learn, resist the urge to connect it to everything. A built-in assistant that does five things well is more valuable than a tool you have stretched into a dozen jobs it handles poorly.

Questions about BigCommerce Companion

Does BigCommerce Companion cost extra?

BigCommerce's own page says Companion comes included with your store, in free trials and on paid subscriptions, with no separate price listed. The current plans are Core, Growth, Scale and Performance, and the Companion page names no plan limits, so look for Companion in your own control panel to confirm.

Can Companion answer questions from my shoppers?

No. It is built for the merchant using the control panel, not for shoppers on the storefront. For a chat tool that serves customers, BigCommerce points to customer-chat apps in its App Marketplace. Test any of those against your returns and shipping policies before switching it on.

Can Companion change my products or prices?

BigCommerce's page names order-status updates as the change it makes, after you preview and approve, and shows 'Issue a store credit to this customer' as an example prompt. It lists catalogue browsing as a read action. Do not assume it edits prices or products; ask it, and watch for the preview step, which is the signal that a change is possible.

Is Companion the same as the Merchant AI app?

No. Merchant AI is a separate app from an outside developer, listed in the App Marketplace, not a part of Companion. Companion is the assistant that ships inside the control panel. Check which one you are looking at, because the features, pricing and data handling are separate and should be read from each vendor's own pages.

Should I trust its charts for decisions?

Treat them as a fast first look. Compare each chart you plan to act on with the matching report in the control panel's analytics section for the same dates. If the numbers differ, trust the report and note the difference.

Further reads

• Ecommerce Product Recommendation AI: Quiz, Widget or Chat Assistant? — Quiz, widget or chat assistant for shoppers.
• ChatGPT Ads Cost for a Small Store: CPC, CPM and a Test Budget — What a small store spends testing ChatGPT ads.
• Best AI Chatbots for Small Shopify Stores (2026) — Shopper-facing chat tools, compared for small stores.
• How Much Does It Cost to Add AI to a Shopify Store? — What adding AI to a store costs, itemised.
• How to Set Up AI Abandoned Cart Emails for a Small Online Store — A job Companion does not cover: cart recovery emails.
• Best AI for Ecommerce Operations: Orders, Stock, Returns, Suppliers — Back-office jobs that need more than a built-in assistant.
• AI for Ecommerce Inventory Management: Reorder Points and Stockouts — Reorder points and stockouts, which Companion does not forecast.
• AI Tools and AI Development: The Complete 2026 Guide — the AI hub, including every tutorial in the AI-for-business series.

Sources: BigCommerce's Companion product page and pricing page, the Merchant AI App Marketplace listing and Shopify's Sidekick page (checked 5 Nov 2026). The test tasks and sample figures are illustrative and were not run against a live store.