Ecommerce tech stack: the systems a growing store needs and what connects to what
Ecommerce tech stack guide: what to run at each stage, which system owns each record, how one order moves through it and when to add the next tool.
By Rokibul Hasan7 min read
- E-commerce
- Integrations

Very few stores design an ecommerce tech stack. It accumulates. An email tool in the first month, a shipping app in the third, a marketplace connector when Amazon comes along, a spreadsheet to patch the gap between them. Two years later nobody can say which system holds the right stock figure, and every new tool makes that harder.
This post is for owners of growing stores, usually selling on two or more channels, who want to know which systems they need and how those systems should connect. It covers what to run at each stage, who owns each record, what happens to one order as it passes through, and when to add the next tool.
Picking the platform itself is a separate decision, covered in the best ecommerce platform by business type. This post starts after that choice. Tools named below are examples of each category, not endorsements.
An ecommerce tech stack at three stages
| Job | Starting out | Growing, 2 to 3 channels | Established |
|---|---|---|---|
| Storefront and checkout | Shopify or WooCommerce | Same | Same, perhaps headless |
| Payments | Platform default | Plus PayPal, wallets | Several providers by region |
| Stock | Platform inventory | Inventory tool such as Veeqo or Cin7 | WMS or ERP |
| Shipping | Platform labels | ShipStation, Veeqo | WMS or 3PL |
| Email and SMS | Klaviyo or similar | Same, with more flows | Same, with a customer database |
| Support | Shared inbox | Helpdesk such as Gorgias or Zendesk | Helpdesk plus AI answers |
| Accounting | QuickBooks or Xero | Plus A2X or Link My Books | ERP or the same pair |
| Reporting | Platform reports | Combined dashboard | Data warehouse |
Most stores with one channel and one person need only the first column. Moving right is a cost you should be forced into by a specific problem, which is the point of the signals table further down.
Give every record one owner
Tools matter less than this rule. Five kinds of record move around a store, and each needs exactly one system that is allowed to be right. Everything else reads from it.
| Record | Usual owner | What goes wrong with two owners |
|---|---|---|
| Product data | Shopify, or a PIM at scale | A price fixed on Amazon is overwritten overnight |
| Stock | Inventory tool or WMS once you have one | Overselling, or items hidden while still on the shelf |
| Orders | The channel creates them, one system collects them | Orders shipped twice, or not at all |
| Customers | Platform or email tool | Unsubscribes lost, duplicate profiles |
| Money | Accounting software | Two versions of revenue in every meeting |
Write this table down for your own store before buying anything. If you cannot fill in a row, that row is the next thing to fix. The official TikTok app for Shopify shows the idea in practice: Shopify is the source of truth for the catalogue, and edits made in TikTok's Seller Center get overwritten.
Stock is the record that causes the most visible damage. How to keep it right across channels is covered in preventing overselling.
Follow one order through the stack
Tracing a single order shows which connections you depend on. Take a store selling on Shopify and Amazon, shipping from its own shelves.
- The shopper pays. Checkout hands the card to the payment provider, and the platform records the order.
- Stock is committed. Shopify moves the unit from Available to Committed. Amazon does not know yet, so your sync must tell it. If this fails, you oversell.
- The order reaches whoever packs it. A webhook or a poll sends it to shipping software, a WMS or a 3PL. Shopify's webhook docs say delivery is not always guaranteed, that duplicates can arrive and that order is not guaranteed, so a reconciliation job has to run behind it. If this fails, an order sits unshipped and nobody notices.
- A label is bought. Rules choose the carrier, and the tracking number comes back.
- Tracking returns to the channel, which tells the customer and counts towards your marketplace metrics. If this fails, the order looks late.
- Email and support update. Your marketing tool hears about the purchase, and the helpdesk can see the order when the customer writes in.
- Money arrives later. A payout lands net of fees and refunds, and has to be split in your accounts. See ecommerce accounting automation.
- Some of it comes back. A return reverses steps 2, 5 and 7, which is why returns are worth planning for from the start.
That is eight handovers for a single parcel. Walk through it for your own store and mark each step as automatic, manual or unknown. The unknowns are your risk list.
Count the connections before you add a tool
Connections grow faster than tools. If every system talks directly to every other, the possible links are n(n − 1) ÷ 2:
| Systems | Possible direct links | Links through one hub |
|---|---|---|
| 4 | 6 | 3 |
| 6 | 15 | 5 |
| 8 | 28 | 7 |
| 10 | 45 | 9 |
You will not build every possible link, but each real one needs monitoring, credentials and upkeep. Shopify releases API versions quarterly and supports each for at least 12 months, so anything built on an old version has to be revisited. Multiply that by each vendor in your stack.
A hub reduces this. It can be your commerce platform, an inventory tool that already connects your channels, or a small integration service of your own that holds the mappings and logs every message. Pick one early. Most growing stores can use a tool they already own: Shopify's Marketplace Connect covers Amazon, eBay, Walmart and Target Plus (not Etsy), and shipping software usually collects orders from every channel. Shopify Flow, Zapier, Make and n8n handle lighter jobs.
What to add next, and the signal that says so
Buy a tool when a specific, repeated pain justifies it, and not because a list says growing brands have one.
| Signal you can see | Add | Leave for later |
|---|---|---|
| You oversold more than once this quarter | Inventory sync across channels | A full WMS |
| Someone copies tracking numbers by hand | Multichannel shipping software | Custom label code |
| Your bookkeeper rebuilds payouts from CSVs | An accounting connector | An ERP |
| The same five questions fill the inbox | A helpdesk with saved replies | An AI agent |
| Nobody agrees what last month's margin was | One combined dashboard | A data warehouse |
| Picking errors rise with order volume | Barcode scanning and bin locations | Any robots |
The right-hand column is what vendors will try to sell you first. Each of those may be right eventually, and the cheaper step usually tells you whether it is. The same reasoning applies at the customer end: checkout changes, new payment providers and personalisation should each answer a measured problem.
Audit your ecommerce tech stack in an afternoon
A simple spreadsheet is enough. For every tool you pay for, record:
- Monthly cost, including per-order or per-label fees that scale with growth.
- Who owns it inside the business, by name.
- Whose account it is. Tools in a former contractor's or agency's name are a serious risk.
- What goes in and what comes out, in terms of the five records above.
- What happens if it stops for a day. Would you notice? Who would?
- How you get your data out. Check for a full export before you need one.
- Whether anything else does this job. Overlap is where the savings are.
Most audits find a tool or two nobody uses, plus one undocumented connection that everything depends on. Document that one first.
Off the shelf first, custom where it fits
Maintained software beats custom code for nearly every job in the first table. Payments, tax, email and helpdesks are hard problems that vendors solve well for a modest fee, and you should not rebuild them.
Custom work earns its place in the gaps between systems: a rule your connector cannot express, a channel it does not support, or one view across data that lives in four places. That is what I build, on each platform's official API. A single connection or automation is the kind of job the Marketplace & e-commerce apps MVP plan covers, from $2,500. Two or more channels with stock and orders kept in sync, plus a dashboard, start from $5,500 on the Flexible plan. Both are starting prices with a fixed quote after a short brief, as listed on the pricing page, and you own the code, with the accounts in your name.
FAQ
Does a small online store need an ERP?
Usually not. An ERP brings finance, purchasing, stock and orders into one system, and it suits businesses with several warehouses, manufacturing or complex purchasing. Most small stores get what they need from their platform, an inventory tool and accounting software, joined by a connector. Consider one when those tools disagree so often that fixing the data takes real staff time every week, and get a quote that includes implementation as well as licences.
Should I use one all-in-one platform or separate specialist tools?
Start with as much as your platform does well, because every separate tool adds a connection to maintain. Move a job to a specialist when the built-in version is the thing holding you back, such as email segmentation or multichannel shipping. Most stores end up with a mix. What matters is that each record has one owner, so tools can change later without your data falling apart.
How do I know if my ecommerce systems are connected properly?
Trace a real order from payment to payout and check each handover: stock adjusted on every channel, order received by whoever packs it, tracking back on the channel, fees recorded in the books. Then test a failure, such as a tool being offline for an hour, and see whether anyone is told. A healthy setup logs each message, retries failures and runs a daily check for records that do not match.
If your systems have grown into something nobody can fully explain, send me a short brief: your channels, the tools you pay for, how orders reach whoever packs them and where things go wrong. You can start a marketplace MVP enquiry, see marketplace integrations, or read about warehouse automation for online stores.


