B2B Ecommerce With AI: Trade Accounts, Pricing and Reorder Emails

Run the B2B shop in three parts: a trade account application that you approve by hand, two or three price tiers written as plain rules, and reorder emails sent only when a customer is overdue against their own ordering pattern. AI drafts the replies and emails and checks your rules, and you make every approval and pricing decision.

The reason to keep AI out of approvals and prices is that both create commitments. A trade account means credit, discounts and sometimes reserved stock. A wrong price tier means you either lose margin quietly or annoy a good customer loudly. Drafting is where AI saves time safely, which is why it takes the writing and leaves the deciding.

Follow me on Instagram@sagnikteaches

What a trade shop needs that a retail shop does not

A retail shop shows the same prices to everyone and takes payment at the till. A trade shop changes five things, and the platform you pick, whichever it is, should handle them or have an add-on that does.

Connect on LinkedInSagnik Bhattacharya

• Approval before access. Visitors apply for an account, and you approve before they see trade prices or can order.
• Hidden or customer-specific prices. Retail visitors see retail prices, and approved accounts see the trade price for their tier.
• Minimum orders. A first order floor and sometimes a smaller floor on reorders.
• Payment terms. Some accounts pay by card at checkout, others on invoice with terms such as 30 days.
• Reorder convenience. A way to repeat a past order, with an order history the customer can see.

This tutorial keeps the platform choice generic on purpose. Whether you use a general shop builder with a wholesale add-on or a purpose-built B2B tool, the rules and the writing stay the same, and they are the parts you will carry from one platform to another. If you do use a general shop platform, the cost of adding AI to a Shopify store shows how add-on costs build up.

Subscribe on YouTube@sagnikteaches

A worked example: a small wholesaler of clothing to independent shops

Say a small online clothing business has decided to sell wholesale to independent boutiques as well as retail. It has about 120 products. A dress that retails at $48 costs $14 to produce, and its trade price is $26, which is a 46% margin on the trade sale. The owner sets these illustrative rules, which the rest of the tutorial builds on.

Three tiers, each with a trigger anyone could check from the order history. A boutique that orders 40 dresses at $26 spends $1,040, comfortably above the $300 minimum. A boutique that spent $6,200 over the past year moves to Trade Plus and sees $24.70 on its next login. The owner will not have to remember to do it, if the platform supports automatic tier assignment, and if it does not, a monthly check of a short list does the same job.

Trade account applications: a form, a checklist and a reply

Start with the application form. Ask for what you need to decide, and nothing more. For a clothing wholesaler that means: business name, trading address or website, type of business (boutique, online shop, market stall, other), how long it has been trading, an indication of expected monthly spend, how they heard of you, and a contact name and email. If you offer invoice terms, add a line stating that terms are agreed separately after the first order.

Then write the checklist the owner uses on every application. Having it in writing means applications are judged the same way on a Monday morning and a Friday evening, and it gives the AI something to check against.

1. Is it a business and not a consumer buying at trade price? Look for a website, a shop listing or a social presence that shows trading.
2. Does the type of business fit what you sell?
3. Is the expected spend at or above your first-order minimum?
4. Are there any signs of a reseller who would undercut you on a marketplace? Decide your policy first, and write it down.
5. Is the contact information consistent, with a business email rather than a free one where that matters to you?

Applications that pass go to a quick human check. Give the AI the application text and your checklist and ask for a summary against each point, plus a draft reply for one of three outcomes: approve, ask for more information or decline. A prompt:

You help a small clothing wholesaler review trade account
applications. You do not decide. For the application below:

1. Compare it with each point of the checklist and write
   "meets", "unclear" or "does not meet" with one line of reason,
   using only what the applicant wrote.
2. Recommend approve, ask for more information, or decline, and say
   why in one sentence.
3. Draft the reply for your recommendation, under 100 words, friendly
   and plain. Greeting "Hi [first name]". Sign-off "[your name]".
   Never mention discounts beyond the standard trade price list.

Checklist: [paste checklist]

Application: [paste application]

A sample, illustrative, for an application from a boutique with a website, a shop address, two years trading and an expected spend of about $800 a month: all five points come back as "meets", the recommendation is approve, and the draft reads: "Hi [first name], thanks for applying for a trade account. I have set you up on our standard trade prices, and you can log in now to see them. First orders have a $300 minimum. Please let me know if you have any questions about delivery or sizes. Best wishes, [your name]". The owner reads the summary, sees the website was checked, and sends.

For a second application, say one from a sole trader with no website and an expected spend of $150, the summary shows "unclear" on trading evidence and "does not meet" on spend. The draft asks for a link to a shop or social page where products are sold, and mentions the minimum order. That is a sensible reply, and a person decides whether to send it as written. The tutorial on replying to trade enquiries in 60 seconds covers the speed side of the same job.

Writing price-tier rules that nobody argues with

Rules that fit on a page get followed. Write each tier as a sentence anyone can test, and include what happens at the edges. For the example above:

TRADE: any approved account. Wholesale list price.
TRADE PLUS: orders totalling $5,000 or more in the previous 12
  months (by order date, excluding returns and delivery charges).
  5% off wholesale list.
TRADE PARTNER: $15,000 or more in the previous 12 months, same
  basis. 8% off wholesale list.
Tier is reviewed on the first working day of each month. Moving up
  takes effect on the next order. Moving down needs the owner's
  approval and a message to the customer first.
Minimum order: $300 on first order, $250 after.
Exceptions: owner only, noted on the account.

Then use AI as a tester. Paste the rules and ask it to find gaps and contradictions: "What happens to a customer with $5,000 exactly? What about an account that returned half its last order? What if two orders are placed on the same day, one before and one after the monthly review?" It will list edge cases you had not considered. Then test the rules with real or realistic accounts: give it ten customers' 12-month totals and ask which tier each belongs to, and compare with your own working. If it disagrees, one of you misread the rule, and finding out which is the point of the exercise.

The order history figures should come from your shop's own report, not from an AI's arithmetic. The assistant can classify and explain, but a spreadsheet formula that sums orders per customer for the past 365 days is more trustworthy than a sum done in a chat window. Use the AI for the question "does my rule have a gap?", and the spreadsheet for "what is this customer's total?".

Where tier rules go wrong

The trouble is rarely the numbers. It is two things: a rule that cannot be checked from the data you hold, such as "customers who are growing fast", and exceptions that nobody writes down. A boutique that was promised a better price on a phone call, with no note on the account, will be moved back to standard at the next review and will be angry. Put every exception on the account record the day it is agreed. If the platform lacks a field, use a notes column in the tier spreadsheet. A tier system with unrecorded exceptions is a dispute waiting to happen.

Reorder emails built from order history

Trade customers tend to reorder in rhythms. A boutique buys at the start of each season, a restaurant weekly, a salon monthly. A reorder email that arrives when a customer is due is welcome, and one that arrives on a fixed Monday regardless is noise. So the trigger is the customer's own pattern.

The rule, in plain words: look at the gaps between a customer's last four or five orders, take the average, and send a reminder when the time since their last order exceeds that average by seven days. Customers with fewer than three orders get no automatic reminder, because there is not yet a pattern. In the example, a boutique ordered on 3 March, 10 April, 25 May, 5 July and 18 August. The gaps are 38, 45, 41 and 44 days, which average 42. The trigger point is day 49. If it is now 63 days since the last order, the customer is 14 days overdue against the trigger and goes on the list.

The list is simple to build in a spreadsheet from an order export: one row per customer with last order date, number of orders, average gap, days since last order and a flag. Then give the flagged rows to an assistant to draft emails. Include the customer's usual products from their history, because a reminder that says "your usual 12 of the wrap dress, in the sizes you took last time" is far more useful than a generic nudge.

Write a short reorder email for each customer below. A person
reads and sends each one.

Rules:
- Under 90 words. Greeting "Hi [first name]".
- Mention that it has been about [days_since] days since their last
  order, once, in a friendly way.
- Name up to three products they bought most often, with the
  quantity of their last order. Do not mention prices.
- Link text: "Reorder in one click" (the link is added later).
- No pressure, no discount offers, no claims about stock.

Customer data:
code | first name | days_since | top products and last quantities
B-0412 | [first name] | 63 | Wrap dress (12), Linen shirt (8), Scarf (6)

A sample result, illustrative: "Hi [first name], it has been about nine weeks since your last order, so I wanted to check in before the new season gets busy. You usually take the wrap dress (12 last time), the linen shirt (8) and the scarf (6). Reorder in one click if you would like the same again, or reply to this email if you want to change anything. Best wishes, [your name]." Two points to check on review. "About nine weeks" is 63 days exactly, so the timing line is right. "Before the new season gets busy" is the assistant adding a reason that was not in the data; it happens to suit a boutique, but the same phrase in an email to a restaurant account would read oddly, so either delete it or add a "reason to mention" column that you fill in yourself. And if an item they usually take is out of stock, add a note by hand, since the prompt rightly bans stock claims.

Send at most one reminder per cycle. A second goes only if you have a reason, such as a new product that suits their buying pattern. And keep a log of replies: orders within 14 days of an email, replies that say they are closing or switching supplier, and unsubscribes. The log tells you whether the emails help or irritate. Consent matters too: trade customers who have bought from you are usually reasonable to contact about reorders, but email-marketing rules vary, so check the requirements that apply to your customers, and offer an easy way to opt out.

Choosing a platform without choosing a product

You do not need a verdict on platforms to get started. You need a list of features to test, and a few hours with a trial. Score each candidate on these.

The export row is the one that matters most for this tutorial. If you cannot get a clean list of orders by customer out of the platform, the reorder emails become a manual job, and that is worth knowing before you commit. Check also how stock is shared between the retail and wholesale sides. If both draw from one pool, the stock-sync tutorial explains what to watch.

Handling a request for terms or a special price

Sooner or later a good account asks for 60-day terms or a better price on a large one-off order. Treat these as owner decisions, and use AI to prepare, not to answer. Give the assistant the account's 12-month spend, its payment history from your accounts system, and the margin on the products involved, and ask it to set out the cost of saying yes: the margin given up on the order, the cost of carrying the unpaid invoice for an extra 30 days, and the cost of saying no in a likely loss of volume.

Take a boutique asking for 8% off a $2,400 order. The discount is $192. If the products carry a 46% margin, the order was worth about $1,104 in margin at standard trade prices and is worth $912 after the discount, which is still good business if the account is growing and a poor one if it is not. The assistant lays out the sum, you read it, and you decide. Then write the exception on the account record, with a date and an end date, so it does not become a permanent tier by accident.

The week the wrong price showed

Picture an illustrative specialist food importer that opens a trade section and relies on the platform's customer groups. A new customer is approved but never assigned to a group. The platform treats them as an ordinary visitor and shows them retail prices, and they order at those prices without a word. A month later the same customer sees a colleague's trade invoice at a lower figure and asks why they paid more. The owner refunds the difference and, after a rough conversation, keeps the account.

It surfaced through a customer's question, not through any report. The preventive step costs a minute: after approving any account, log in as that customer using the platform's preview-as-customer feature, if it has one, or create a test basket, and confirm the tier price shows. Add "tier assigned and price checked" as the last line of the approval checklist, so approving is not finished until it is ticked.

Running the pilot quietly before customers rely on it

Introduce the pieces one at a time, and shadow-run each before it is customer-facing. For two weeks, let the AI draft application replies but send your own as usual and compare. Then send the drafts after review. Only after that, add the reorder list: for the first month send reminders to five or ten accounts you know well, whose reactions you can read, before widening it. The tutorial on piloting AI in shadow mode explains the method in more detail.

Keep the customer's data in your own system. The assistant sees a code, dates, product names and quantities for the email, not an address, a phone number or a bank detail. If you would like help setting up the tier rules and the reorder list around your own order data, my AI implementation consultation covers that kind of work.

Measuring whether the B2B side is paying its way

• Time to reply to an application: same working day, every time. Note the average.
• Approval accuracy: approved accounts that placed a first order within 60 days, as a share of all approvals. Low numbers mean the checklist is too loose or the follow-up is missing.
• Tier disputes: the number of price complaints each month. It should fall to near zero after the first quarter.
• Reorder email results: orders placed within 14 days of an email, as a share of emails sent, compared with the same customers' rate before you started. Count only customers who were overdue.
• Margin by tier: blended margin on Trade Plus and Trade Partner orders against the target, so a discount tier never quietly takes the margin below what the business needs.

After a quarter, review the tiers and the minimums with real figures in front of you. If most accounts sit at the first tier and none reach the second, the threshold is too high or the second tier is not worth having. If a handful of accounts are doing most of the volume, look after them personally, and let the system handle the long tail.

Questions from small wholesalers about trade accounts

Should AI approve trade account applications?

No. Let it check the application against your checklist and draft the reply, but a person approves. Approval commits you to credit terms, discounts and sometimes stock allocation. A wrong yes is costly to undo and a wrong no loses a good customer, so keep the decision with the owner.

How many price tiers does a small wholesaler need?

Two or three. One standard trade price, and one or two volume tiers triggered by 12-month spend. More tiers create arguments over where each customer sits and make the rules hard to explain. Start simple and add a tier only when a real customer case demands it.

Do I need a special B2B platform?

Not necessarily. The features that matter are hiding prices until approval, customer-specific pricing, minimum orders and an order history export. Several general shop platforms and add-ons cover them. Write your list of must-have features first, then test whichever platform you are considering against it.

How often should reorder emails go out?

Only when a customer is overdue against their own pattern, not on a fixed calendar. A rule such as the average gap between their orders plus a week works well. Send one reminder, and a second only if you have a reason, so you never nag a good account.

Is it a problem to put customer order history into an AI tool?

Send the minimum: customer code, dates, product lines and quantities, without addresses or contact details. Check the privacy terms of the plan you use, and prefer a business plan. Keep the AI to drafting the email and let your own system hold the data.

Further reads

• How Barbers Can Win Back Lapsed Clients With Automated Messages — Win-back messages built on a customer's own visit pattern.
• How Small Shops Can Use AI to Forecast Stock and Reorder — The stock side of reordering, for your own purchasing.
• Chasing Unpaid Invoices With AI: Polite Reminders for Trades — Polite reminders for unpaid invoices on trade terms.
• Should You Tell Customers You Use AI? A Disclosure Guide — Whether and how to tell trade customers about AI drafting.
• How to Use Shopify Sidekick to Run Your Store Faster — Built-in store assistant tasks on one common platform.
• Free vs Paid AI Tools for a Small Trades Business — Which AI tools are worth paying for in a small trading business.
• How to Use AI in a Bakery Business: Wholesale and Standing Orders — Wholesale bakery workflow: one standing-order sheet, a change form for cafés, AI that reads messy messages, and reminder texts that get answered.
• Paint Store Marketing: Colour Advice Posts and Trade Offers With AI — Colour-advice posts built from your own chart, a coverage sum, honest wording on photo matching and a trade-account email with the margin maths shown.
• Stationery Shop Marketing Ideas: School and Corporate Orders With AI — A step-by-step plan for winning school and small-office orders: the price sheet, the buyer list, AI-drafted emails, follow-ups, samples and quotes.
• Best AI Tool for Pricing Wholesale Goods: Chat Assistant or App? — Chat assistant plus a sheet, an inventory-system price list, or pricing software: the right pick depends mostly on how many SKUs you carry.
• Builders Merchant Software: Trade Accounts, Quotes and Where AI Fits — Buy trade-counter software for accounts, stock and pricing; use AI on the edges: emailed materials lists, statement chasing and stock queries.
• AI Personalisation in B2B: Tailor Outreach to Each Trade Account — Personalise B2B trade outreach and quotes account by account using structured notes sheets and human-reviewed AI prompts to increase reply rates.
• Klaviyo Back in Stock Emails for Shopify: Setup and AI-Drafted Copy — Build the Klaviyo back in stock flow, draft honest copy with AI, and measure how many orders it truly recovers.
• Line Sheet Template: Turn a Product Export Into a Wholesale PDF With AI — Export your products, let AI add wholesale prices, MOQs and case packs in a sheet, then check the maths and produce a PDF buyers trust.
• Minimum Order Quantity for Wholesale: Tiered MOQs Small Retailers Accept — A tiered MOQ model with margin maths, an exceptions policy and a way to test the rules against your own order history before you publish them.
• Wholesale Nursery Availability List: AI Builds and Emails It Weekly — A weekly routine that turns a grower's stock sheet into a priced availability list, emails it to trade buyers and marks lines as sold.
• AI Tools and AI Development: The Complete 2026 Guide — the AI hub, including every tutorial in the AI-for-business series.

Sources: no vendor figures are used and no platform is recommended. All prices, tiers, margins and customer figures here are illustrative, with the arithmetic shown; check your own platform's current documentation for the features listed.