
Course ecommerce for training providers: architecture options for your catalogue
Course ecommerce for a training provider means selling dated sessions with limited seats, not just courses, so the store must share seat counts, prices and tax with your scheduling system. There are four ways to arrange that: a built-in store, a web shop in front of a TMS, a headless storefront, or a custom checkout.
The platform guides we reviewed for this topic are written mostly for sellers of self-paced online courses. A training company selling classroom and virtual sessions has other problems: a session that sells out, a client buying twelve seats on a purchase order, a bundle whose parts have different tax treatment. This guide compares architectures rather than products, and it does not rank any platform.
Key takeaways
- Model three things separately: the course, the session with its seat limit, and the offer that says who pays what.
- Calculate in a fixed order: price list, bundle split, discount, then tax. Taxing list prices over-collects when bundles and discounts exist.
- Pick the option by deciding which system owns seat counts. Two systems that each think they own them will oversell.
- A built-in store is the least work, a shop in front of a TMS is the most common compromise, and headless or custom checkout buy flexibility with engineering time.
What does course ecommerce have to model that an ordinary shop does not?
A shop sells things that exist; a training provider sells a place on a session that has not happened yet. Course ecommerce therefore needs three separate ideas: the course (what it is), the session (when, where, with a seat limit) and the offer (who pays what, under which terms). Merging them is the most common modelling mistake.
In a typical shop the model is product, variant and stock count. A session fits that shape at first, with the date as the variant and the seat limit as stock. It stops fitting when instructors, rooms and time zones decide the real capacity, which is why our TMS data model for sessions and enrolments keeps them in the scheduling system.
B2B adds a second difference. Corporate buyers expect contracted prices, purchase orders instead of cards, and the right to name attendees after paying. A consumer catalogue usually has no place for these.
How should you model the catalogue, prices, bundles and discounts?
Keep six layers apart: course, session, offer, bundle, discount and entitlement. Each has a different owner and rate of change, and mixing them makes price changes, refunds and reports painful later. The table shows what each holds and who should own it.
| Layer | What it holds | Usually owned by |
|---|---|---|
| Course | Title, description, objectives, duration, format, prerequisites | Catalogue or content system |
| Session | Start and end in UTC plus time zone, venue or meeting link, instructor, seat limit | Scheduling system (TMS) |
| Offer | Amount, currency, audience (public, member, contract), valid dates, tax code | Commerce layer |
| Bundle | Component courses, bundle price, rule for splitting that price | Commerce layer |
| Discount | Type, scope, redemption limit, expiry, whether it stacks | Commerce layer |
| Entitlement | The right to a seat, who may attend, when it expires | TMS registration record |
If you use Stripe, its docs say that “if two options appear as different rows on your pricing page, they must be different products,” while currency or billing interval are prices on one product. Stripe also documents inline prices, defined when you create a checkout session or invoice, for stores that manage the product catalogue outside Stripe. So a payment catalogue can mirror your courses or just receive the final amount.
In what order should price, discount and tax be calculated?
Resolve the price list first, then split any bundle price across its components, then apply discounts to the split amounts, and calculate tax last. Taxing list prices and subtracting the savings afterward overstates tax, and it puts the discount on the wrong lines when components are taxed differently, which makes a line-by-line refund hard to reverse.
The worked example below is illustrative. A bundle sells a two-day workshop (list price 1,800.00, taxed at 8%) with an exam voucher (list price 1,300.00, taxed at 0%) for 2,700.00, with a 10% promotion on top. The rates are made up to show the mechanics, not real tax rules.
from decimal import Decimal as D, ROUND_DOWN, ROUND_HALF_UP
CENT = D("0.01")
def allocate(total, weights):
# split total across weights in whole cents; parts sum exactly to total
raw = [total * w / sum(weights) for w in weights]
parts = [r.quantize(CENT, ROUND_DOWN) for r in raw]
short = int((total - sum(parts)) / CENT)
for i in sorted(range(len(raw)), key=lambda i: parts[i] - raw[i])[:short]:
parts[i] += CENT
return parts
def price_order(lines, bundle_price, discount_pct, seats):
# lines: (name, list_price_per_seat, tax_rate); order: split, discount, tax
net = allocate(bundle_price * seats, [l[1] * seats for l in lines])
out = []
for (name, _, rate), n in zip(lines, net):
disc = (n * discount_pct / 100).quantize(CENT, ROUND_HALF_UP)
tax = ((n - disc) * rate).quantize(CENT, ROUND_HALF_UP)
out.append((name, n, disc, tax, n - disc + tax))
return out
lines = [("Two-day workshop", D("1800.00"), D("0.08")),
("Exam voucher", D("1300.00"), D("0.00"))]
for row in price_order(lines, D("2700.00"), D("10"), 1):
print(row)
The snippet was run before publishing, and the split sums exactly to the bundle price. Here is the result for one seat.
| Line | Bundle share | Discount (10%) | Tax | Line total |
|---|---|---|---|---|
| Two-day workshop | 1,567.74 | 156.77 | 112.88 | 1,523.85 |
| Exam voucher | 1,132.26 | 113.23 | 0.00 | 1,019.03 |
The order total is 2,542.88. Taxing the list prices instead would charge 144.00 in tax on the workshop alone, 31.12 more than the 112.88 above, which is the over-collection a refund process would later have to unwind.
Pro tip: Store the rule, not just the result. Save the price list, bundle split, discount ID, tax code and rate on every order line. A partial refund or credit note must reverse the same allocation, and recomputing it from today’s catalogue will give a different answer after any price change.
How should bundles, group rates and discount codes be modelled?
Treat each as a different object with its own rules. A bundle is a fixed set of courses at one price, a group rate is a price for many seats, and a discount code is a customer-entered reduction with limits. One “coupon” field for all three is how stacking bugs start.
Bundles raise one decision that drives the architecture: does buying hold seats now, or issue an entitlement the buyer redeems against sessions later? Held seats protect the buyer but strand inventory if sessions are never chosen. Entitlements free inventory but need expiry dates and a redemption screen.
Keep contracted client prices on the client account, not in codes. A code is something a customer types and can share; a negotiated price should apply automatically for that client and no one else. Give codes a redemption limit and an end date, and decide in writing whether two discounts may combine.
What are the four course ecommerce architecture options?
The four course ecommerce options are a store built into your LMS or TMS, a web shop such as WooCommerce or Shopify in front of a TMS, a headless commerce backend with your own front end plus a TMS, and a custom checkout on your own catalogue. They differ mainly in who owns seat inventory.
Option 1: When does a store built into your LMS or TMS make sense?
A built-in store makes sense when sessions, seats and enrolment already live in one system and your buying journeys are simple. Seat counts cannot drift because there is one database. The trade-off is less freedom over page design, pricing rules and tax tooling than a dedicated commerce platform offers.
As an example, the SimpliTrain course catalog page describes a storefront where each course page can show pricing, and where corporate clients see a contracted portfolio separate from the public catalog. That page does not describe discounts, bundles or tax, so test those against your scenarios in any built-in store you evaluate.
Option 2: What does a WooCommerce or Shopify shop in front of a TMS involve?
The shop owns product pages, cart and payment, while the TMS owns sessions and seats. That means two catalogues to keep in step: each session becomes a product, variation or booking slot in the shop, and each paid order must create a registration in the TMS. Seat counts then exist in both places unless you name one authoritative.
Shops can model capacity themselves. The WooCommerce Bookings extension, for instance, lets you count each person against available places by setting a maximum per block. That suits a standalone class, but it duplicates the TMS when rooms and instructors decide how many seats exist.
The integration is the product here. The TMS should push availability to the shop, and a paid order should call back with an idempotent webhook, as laid out in our registration and payment workflow post. Plan for refunds too: a refunded order that does not release its seat is the classic leak.
Option 3: When is headless commerce plus a TMS worth the build?
Headless fits when you want full control of the catalogue front end, for example one site that mixes sessions, self-paced content and a B2B portal, while a commerce engine handles cart and payment. You build the front end and the glue to the TMS, so you also own them for years.
Shopify describes the idea this way: “Headless commerce separates your storefront’s frontend from its backend.” Its Storefront API serves products, collections, search and the cart, and hands customers off to checkout. One workable design reads live session availability from the TMS and uses the commerce backend only for cart and payment.
That design only works if the TMS exposes sessions and seat availability through an API you can rely on. Our API-first platform checklist covers how to test that before you commit.
Option 4: When should you build a custom checkout?
Build a custom checkout when your purchasing rules do not fit a shop: contracted client prices, purchase orders, carts that span several sessions, or an invoice-first flow. You own the catalogue, cart and order model, and use a payment provider for card capture. Every rule from the pricing section becomes your code.
A provider’s tax tooling can still help. Stripe Tax, for example, lets you set a tax code per product (the docs give pet grooming, e-books and SaaS as examples) and a tax behavior per price that says whether tax is included or added. Which code applies to your training is a question for your tax adviser, not for the codebase.
The cost is the long tail after release: refunds, credit notes, rounding, failed payments and finance reporting. Treat the order model as a core system and test it like registration.
How do the four options compare on seats, tax and B2B?
Compare them on the questions that break training sales: who owns seat counts, whether prices can differ per session or client, how tax is handled, and where the weak point sits. The table is architectural judgment, not test results for any product.
| Question | 1. Built-in store | 2. Shop in front of TMS | 3. Headless plus TMS | 4. Custom checkout |
|---|---|---|---|---|
| Who owns seat counts? | The LMS or TMS alone | Shop and TMS, unless one is named authoritative | The TMS, read through its API | Your order service plus the TMS |
| Prices per session or client | Whatever the vendor models | Limited by the shop’s product model and extensions | Limited by the commerce backend | Anything you build |
| Tax handling | Vendor’s tooling | Shop’s tax settings or extensions | Commerce platform’s tax features | A tax service or your own rules |
| B2B and purchase orders | Check the vendor’s contract-price and invoice support | Check extensions and manual steps | Built in your front end and glue | Designed in from the start |
| Front-end and SEO control | Template-limited | High with a CMS or theme | Full | Full |
| Build effort | Configuration | Configuration plus an integration | Front end plus an integration | Checkout, orders and an integration |
| Typical failure | Pricing rule you cannot express | Oversold or stranded seats from two sources of truth | Stale cached availability | Gaps in refunds, credit notes and tax edge cases |
Which course ecommerce option fits which kind of training provider?
Match the option to where your sessions already live and what your team can maintain. If one system owns sessions and seats, start with its store. If marketing needs a content-led catalogue, a shop with a clean seat sync fits. Headless and custom checkout suit providers with engineers and rules no product expresses.
Five questions narrow the choice. Answer them for your own business before you watch a demo.
- Does one system own sessions, seats and enrolment today? If so, test its built-in store first.
- Do you need session-level prices, bundles or contracted client prices that store cannot express?
- Does your web team already run WordPress or Shopify and need content-led catalogue pages?
- Can engineers own a TMS integration for years, including its failure handling?
- Is invoice-first B2B your main sales route, or card checkout for individuals?
These are starting points, not a ranking. Many providers combine them, such as a built-in store for corporate portals and a shop for public workshops, provided one system owns each session’s seat count.
Pro tip: Run a last-seat test before you choose. Script three scenarios on each candidate: two buyers paying for the last seat at once, a refund after payment, and a cancelled session with paid seats. Record which system’s count changes, how long it takes, and who is told.
What should you do next?
Write a one-page catalogue spec with the six layers above and your five hardest real sales, such as a client booking twelve seats on a purchase order, a bundle with a refund, and a public discount code. Hard cases expose the gaps that a feature checklist hides.
Run the same cases against each candidate. Anything a vendor answers with “we can customize that” goes on a list with a cost and an owner.
If a built-in storefront is on your shortlist, you can request a SimpliTrain demo and bring those cases along. The spec works for any vendor you compare.
Frequently asked questions
Can you sell instructor-led courses with a standard ecommerce platform?
Yes, if you map each session to something with stock, such as a variant or booking slot. The platform will not know about instructors, rooms or time zones, though, so seat limits, cancellations and waitlists must be synced from your scheduling system, or you risk selling seats that do not exist.
Should seat inventory live in the shop or in the TMS?
Put the authoritative count in the TMS, because rooms, instructors and cancellations decide real capacity there. Let the shop display availability and request a seat hold. If the shop keeps its own count too, reconcile the two on a schedule and alert on mismatches.
How do you price a course differently for corporate clients?
Use price lists by audience, and attach a contracted list to the client account so it applies automatically. Keep discount codes for public promotions, with redemption limits and end dates. This stops a negotiated price leaking through a shared code.
How should course bundles work with limited seats?
Decide whether a bundle holds seats at purchase or issues an entitlement redeemed later. Holding seats protects the buyer but can strand inventory. Entitlements need an expiry date and a redemption step. In both cases, split the bundle price across components so refunds and tax work per line.
Who calculates tax on course sales?
Your system calculates it, but your tax adviser decides the treatment, which can depend on what you sell and where the buyer is. Set a tax code on each item and keep the rate on every order line. Providers such as Stripe offer tax tooling that works from per-product tax codes.
Tags :
