The right ecommerce infrastructure depends on the stage your business is in. We regularly see two expensive mistakes that run in opposite directions. One is a $10M brand paying for a system built for a $100M company, while the other is a $40M brand still running orders through spreadsheets.
Most growing DTC brands run four systems: the store platform, data and reporting, fulfillment and customer service. Each one has a version that fits your stage now and versions that can wait. Here is the short map.
- Launch (roughly under $5M): a hosted store platform and the reports that come with it.
- Early growth (roughly $5M to $25M): one revenue number your whole team agrees on.
- Scaling (roughly $25M to $100M): inventory counts that match across every channel and stock location.
- Established (past $100M): a named owner for each system, with custom pieces only where the platform hits a real limit.
The revenue bands are rough guides. How the business runs places you more accurately than what it earns.
Ecommerce Infrastructure Is the Set of Systems That Take, Fill and Support Every Order
Ecommerce infrastructure is the store platform, the data and reporting, the fulfillment operation and the customer service tools, plus the connections that pass orders, inventory and customer records between them. Each system has its own job. The platform takes the order, fulfillment packs and ships it, reporting counts it and customer service answers questions about it.
The connections matter as much as the systems. An order that lands in the platform has to reach the warehouse, the books and the help desk without anyone retyping it. Those handoffs run through APIs, the links one piece of software uses to send data to another.
At launch, the platform’s built-in connections usually cover the basics with a little setup. As a DTC brand grows, the backend systems behind its store multiply. The connections between them carry more work at every stage.
Your Store Platform Is One Piece of Your Infrastructure
A store platform such as Shopify or BigCommerce takes the order, holds the catalog and runs checkout. The rest of your ecommerce infrastructure fills that order, counts it and answers questions about it. Treating the platform as the whole system can lead a brand to blame it for problems that live in fulfillment or reporting.
That’s why replacing the platform rarely fixes a reporting or shipping problem on its own. The difference between platform and infrastructure matters most right before a brand signs a replatform contract.
Your Stage Depends on How the Business Runs More Than on Revenue
Revenue is a rough guide to your stage. How the business runs is a better one, because two brands at $20M can need very different systems.
One might sell a single product on its own site. The other might sell 4,000 SKUs across its site, Amazon and wholesale accounts.
Four Signals Show Which Stage You’re In
Knowing how to scale ecommerce infrastructure starts with four signals about how the business runs today.
- Order volume per day: A team packing a few dozen orders a day can absorb manual steps that break down at ten times the volume. High order volume turns every manual step into a daily cost.
- Sales channels and stock locations: Every added channel or warehouse is another place inventory has to match. Once you sell in more than one place, a mismatch turns into an oversell.
- People who pull or act on the numbers: When one founder reads the reports, the platform’s dashboard is enough. When five people make decisions from different tools, the business needs one agreed set of numbers.
- Product and order complexity: SKU count, bundles, subscriptions and preorders each add rules the systems have to follow. A catalog of 30 products and a catalog of 3,000 put very different weight on the same tools.
Most Brands Sit at Different Stages in Different Systems
It’s common to run scaling-stage fulfillment next to launch-stage reporting. A brand that moved to an outside warehouse early can still be closing the month from a spreadsheet.
So locate each system on its own before deciding what to build. The system furthest behind the others is usually the one costing you money today.
If your systems sit at different stages and you can’t tell which one holds the others back, we can help you pick the system to fix first.
Most Brands Can Stay on a Hosted Store Platform Longer Than They Expect
For most brands, a replatform is a decision for the scaling stage or later. The reason to move should be a limit you can name, such as a checkout you can’t change or wholesale ordering the platform doesn’t support.
At launch, build on a hosted platform and keep the setup close to its defaults. In early growth, spend on the store’s speed and checkout before you spend on a new platform. At scaling, the real signals are business needs the platform can’t meet, such as B2B ordering, several regions and currencies, or checkout rules your team needs to change.
Some established brands outgrow the hosted model once their requirements reach enterprise scale. For those brands, a modular setup (covered below) can pay off.
Too Many Apps Usually Show Up Before the Platform Runs Out
Most brands hit app trouble long before they hit a platform limit. Each app solves one problem. After a few years a store can carry dozens of them.
Each one adds code that can slow the pages and keeps its data in yet another place. So before you price a new platform, list your apps and what each one does. Cutting the apps that overlap can speed up the store and delay the need for a new platform.
Overlapping apps can also skew the numbers. We once found a Shopify app that had quietly turned its own Google conversion tracking back on and made itself the main source of purchase data. It had probably been double-counting the client’s purchases for who knows how long.
Get One Revenue Number Everyone Trusts Before Buying More Reporting Tools
The first data build for most brands is an agreement on what counts as revenue and margin. Dashboards come after that.
At launch, platform reports and a spreadsheet are enough. In early growth, the team agrees on one source of truth for revenue and margin and on which ecommerce KPIs to track.
At scaling, store, ad and fulfillment data meet in one central place, usually a data warehouse (a database that pulls in a copy of each system’s data). Established brands often give that warehouse a dedicated analytics owner or team.
Even large marketing teams struggle here. Salesforce’s 2026 State of Marketing report, a survey of 4,450 marketing decision makers, found only 58% have complete access to service data, 56% to sales data and 51% to commerce data. So roughly half of marketing teams work without full access to commerce data, the record of what customers actually bought.
Different Revenue Numbers in Different Tools Are a Connection Problem
Your store, your ad platforms and your email tool each report revenue their own way. The store counts orders, the ad platform counts sales it believes its ads caused and the email tool counts sales within a set window after someone opened a message.
None of them is wrong. They answer different questions. Agree on which number the business runs on and what each tool’s number is for before adding a dashboard.
The same Salesforce report found 69% of marketers struggle to respond to customers promptly because they can’t access the context they need. Systems that don’t share data produce that delay every day. Connecting the systems closes that gap, which a new dashboard can’t do.
Fulfillment Needs to Change When Shipping Speed or Accuracy Starts to Slip
The signal to change fulfillment is slipping performance: late shipments, wrong items or stock counts the team no longer trusts. Order volume alone doesn’t tell you.
At launch, most brands pack in-house. In early growth, many move to a third-party logistics partner (a 3PL, a company that stores, packs and ships orders for you). Others add warehouse software to their own space.
Early growth is also when demand forecasting starts to pay, because a stockout on a best seller now costs more than the time spent planning. At scaling, inventory spreads across several locations and channels. Established brands run several warehouses or partners and send each order to the closest one with stock.
Reliable delivery dates tend to matter more than fast ones. McKinsey’s consumer research, as reported by Retail Dive, found delivery speed fell from shoppers’ top delivery priority in 2022 to fifth in 2024. So a fulfillment build that makes your delivery promise dependable usually earns more than one that shaves a day off it.
Macy’s supply chain director Vinny Pagliuca put it plainly to Retail Dive: “If we tell you we’re going to get something to you in three days, and every time we deliver in three days, you’re going to keep coming back.” That’s a goal a 3PL contract or a warehouse system can be held to.
Fulfillment programs can also outlive the stage they were built for. Warby Parker launched Home Try-On in 2010, mailing shoppers five frames to try for five days at no charge, because people didn’t yet trust buying glasses online. In August 2025 the company said it would end the program by year end because most recent users lived within 30 minutes of one of its 300-plus stores, as Retail Dive reported.
The wind-down came with a $2.5 million inventory write-down. Warby Parker built the program to win shoppers’ trust online and its stores later did that job. A growing brand can retire a fulfillment program the same way once the need it served is gone.
Inventory Has to Match Across Every Channel Once You Sell in More Than One Place
Selling on your site, Amazon and wholesale from one pool of stock means every sale has to reduce the same count within minutes. This is where an ERP starts to earn its cost. An ERP (enterprise resource planning software) is one system that ties together inventory, orders, purchasing and accounting.
The signals for integrating an ERP with ecommerce infrastructure are oversells, a month-end close done by hand across systems, and purchasing decisions made from counts nobody trusts. Revenue alone is a poor trigger. An omnichannel brand with three channels can hit those signals well before a larger brand that sells only on its own site.
A Shared Inbox Works Until Order Questions Outnumber the People Answering Them
Customer service tools should grow with question volume. A shared inbox is the right tool while two or three people can answer everything the same day.
In early growth, a help desk tool connected to order data lets agents see the order without switching screens. At scaling, self-service answers for the top repeat questions (where is my order, how do I return this) take load off the team. Established brands staff tiers of support and set coverage hours.
The stakes rise with volume. Zendesk’s customer service research found more than half of consumers say they’ll switch to a competitor after a single bad experience. For most brands, the next service build is whatever stops the most common bad experience, which is often a slow answer to “where is my order.”
The Two Expensive Mistakes Are Building Too Early and Connecting Too Late
Both mistakes come from choosing systems by revenue or ambition instead of by the signals above.
A $10M Brand Paying for a System Built for a $100M Company
Buying ahead makes sense on paper. Owners who’ve lived through one painful migration want to do it once. A bigger system also feels like room to grow.
The cost shows up in three places: license fees sized for a larger company, a setup that can run many months, and a team too small to run what it bought. We’d rather see that money go into the one system holding the business back today, which your team can learn to run now. The bigger one will still be there when the signals call for it.
A $40M Brand Still Running Orders Through Spreadsheets
The opposite mistake is quieter and often costs more. Staff spend hours re-keying orders and stock counts between systems, small errors turn into oversells and refunds, and the leadership team argues about whose revenue number is right.
If a team member’s main job is copying data from one screen to another, then the connection between those systems is overdue.
How Your Systems Pass Data Matters More With Every Stage
Orders, inventory and customer records have to move between all four systems. The cost of a broken link grows with each stage. At launch, the platform’s built-in connections handle most of that movement.
By scaling, a broken connection shows up as wrong stock counts, mismatched revenue and customers told an item is in stock when it isn’t. The fix is usually an integrations plan: one list of what data moves where, how often and who owns each connection. Many brands never write that list down, which is why the same mismatch keeps coming back.
If you can’t tell whether your next build is a new system or a fix to the connections you have, we can sort that out with your team in a working session on your systems.
Modular and Headless Setups Pay Off for Some Established Brands
A headless commerce setup separates the storefront shoppers see from the system that runs orders behind it, which lets a team build the front end however it likes. A monolithic platform, the standard hosted store, keeps both in one system. MACH architecture (microservices, API-first, cloud-native and headless) goes further by making each function its own swappable service.
A headless or MACH setup pays when two conditions hold. The business needs a front end the platform can’t deliver. It also has a technical team (in-house or through a partner) that can run several systems at once.
Most brands below the established stage get more from fixing the connections they already have. Whether to build or buy each piece of your ecommerce technology stack is a separate decision. Make it after you know which stage each system is in.
The Team You Have Decides Which Systems You Can Run
A system performs only as well as the people who know how to run it. At launch, the founder owns most of the systems. In early growth, an operations lead usually takes fulfillment and service while marketing owns the data.
By scaling, each system needs a named owner. Established brands add specialists such as an analytics lead or an ERP administrator.
So before you buy, name who will run the new system and what they’ll stop doing to make room. Your ecommerce organizational structure should show that owner before the contract is signed.
Plan Your Next Infrastructure Build From the Stage You’re In
Your next build is the one system whose limit costs you money today. Everything else can usually wait a year. The sequence below works at any stage.
- Locate each system’s stage: Use the four signals, one system at a time, and mark where each one sits.
- Name the limit that costs money today: Oversells, late shipments, numbers nobody trusts or messages that wait a day each count. Pick the one with the largest cost.
- Build that first and let the rest wait: The other systems can usually run another year at their current stage while the team learns the new one.
- Put the build into the yearly plan: Write it into an ecommerce plan your team will follow with a range for time and cost. The range moves with your team’s bandwidth, the state of your data and how many systems the change touches.
Some teams want an outside view before they commit to a build. An ecommerce growth consultant can work through the stages with your team and show them how to run the sequence on their own next time.
Infrastructure built for your stage costs less to run and is easier for your team to own. Build the system that’s holding you back first, then check the signals again at your next planning cycle.
