E-Commerce & Tech Stack
Platform, Apps, Tech Stack
- Use Shopify - it's the least wrong choice for 95% of DTC brands from $0 to $50M+.
- Keep the app stack lean, under roughly 15 apps, and audit quarterly; every app adds JavaScript and slows the site.
- Run Meta Pixel plus CAPI with deduplicated event IDs, or you're training the algorithm on fiction.
- Site speed is money: conversion drops roughly 1% for every 100ms of added load time.
On this page
Your website is your shopfront, your sales team, and your brand experience rolled into one. If your tech stack slows you down, it's costing you more than you know. Every tool, every platform, every integration should make the business faster and more efficient. The moment tech becomes a blocker instead of an enabler, it needs to change. Tech should be a liberator. It should never hold you back.
This section is about making pragmatic tech decisions that increase speed, reduce failure points, and give your team leverage as the business scales.
The Only Platform Decision That Matters
Use Shopify. Not because it's perfect. Because it's the least wrong choice for 95% of DTC brands doing $0 to $50M+.
Quad Lock started on Shopify at zero revenue and was still on it at a nine-figure acquisition, growing from a single store to twelve storefronts (six regional DTC sites plus six business-to-business (B2B) wholesale stores), and running pop-up retail through Shopify Point of Sale (POS) at events globally, without ever re-platforming. Picking a platform that's good enough and learning its limitations beats re-platforming every two years to chase features. The platform doesn't make the brand. But a bad platform choice will absolutely slow you down.
When Shopify Plus came along, we moved to it, but it wasn't a dramatic upgrade. We'd been on the platform for so long that we'd already worked around most of the missing features. The real value of Plus is in checkout customisation, lower transaction fees at scale, and dedicated support, not a fundamentally different platform experience.
Before Shopify, we learned this lesson the expensive way. In the early days, Ashton Kutcher tweeted about our first product Opena, when he was the third most followed person on Twitter. It crashed our WordPress site. We were on an "unlimited" hosting plan, ringing the provider while traffic kept coming and orders kept bouncing. By the time they'd added enough capacity, we'd lost hours during the biggest spike we'd ever had.
Your infrastructure needs to be ready for your best day, not just your average day. One viral moment, one celebrity mention, and if your platform can't handle it, you're losing customers in real time. That's one of the underrated benefits of Shopify: you never think about hosting again.
When NOT to Use Shopify
- Digital-only products (Gumroad, Lemon Squeezy, or Stripe direct)
- Pure marketplace (Sharetribe or custom)
- Complex B2B with custom pricing tiers per account
- Already at $500M+ with legacy systems that work
One of the most impactful decisions we made was bringing web development fully in-house. Early on, standard Shopify themes worked fine. But as the product range grew, outsourced agencies couldn't keep up. They didn't understand our products or customers deeply enough.
Once development was in-house, we could go from idea to execution in days instead of months. Custom product pages, guided buying journeys, smart bundling, smart upsells, all proprietary. It became a genuine competitive advantage. Even if a competitor had similar products, they couldn't replicate the buying experience.
The other major unlock was site speed. As we scaled, we stripped out apps and rebuilt those features in-house. This isn't a day-one investment, but around the $10M-$50M stage, it's one of the biggest productivity gains you can make.
Payment Methods
Must-have: Credit/debit (Shopify Payments), Shop Pay, Apple Pay / Google Pay, PayPal.
Should-have: Buy Now Pay Later (BNPL) options (Shop Pay Installments, Afterpay, Klarna). Increases AOV materially, and a large proportion of Gen Z/Millennial shoppers use it.
Market-specific: Afterpay (AU dominant), Klarna/Clearpay (UK), iDEAL (NL), Bancontact (BE).
Always have a backup payment provider configured and ready to go. Chris and I were in China when we got featured on a major American TV show. I checked our PayPal account from a taxi, and the Chinese IP address triggered an immediate freeze. PayPal shut us down at the exact moment national TV was driving traffic to our store.
We got lucky. We knew the Chief Marketing Officer (CMO) of Shopify from winning their Build a Business competition. He was at the pub near HQ. He went back, logged in, and added a backup PayPal account to our store so we could keep transacting. That relationship saved us thousands in lost sales. A support ticket wouldn't have fixed it that night. Have your backup payment method ready before you need it.
Tech Stack by Revenue Stage
Your tech needs scale with revenue. Don't over-engineer at $500K, don't under-invest at $5M.
Most problems at $500K aren't tool problems. They're process problems.
Don't upgrade based on revenue alone. Move to Plus when at least one of these is true:
- Transaction fee savings cover the monthly cost. Do the maths. If your payment processing savings at your current volume exceed the Plus subscription, it pays for itself.
- You need checkout customisation. Checkout extensions, custom scripts, post-purchase flows that standard Shopify can't support.
- B2B functionality matters. Wholesale pricing, company accounts, net payment terms.
- Organisational complexity justifies priority support. Multiple markets, large teams, high-stakes launch windows where downtime costs real money.
If none of these apply, standard Shopify with the right apps gets you further than most founders expect.
The Core Stack
Keep your app stack lean.
Each tool does an obvious job: reviews build trust, loyalty lifts repeat rate, subscriptions smooth revenue. But the real value is downstream, in the data they feed back into the rest of your stack. A review left after delivery should flag that customer for a Klaviyo cross-sell segment. A subscription about to lapse should trigger a churn-risk flow before the customer is gone, not a 'we miss you' email a month too late. A loyalty tier should be a segment your email and ads can target. If a tool sits in a silo and doesn't talk to Klaviyo and Shopify, you've bought a feature, not an asset.
Nice-to-Have (Add When Ready/Needed)
Software Evaluation Matrix
When evaluating new software or apps, use this framework:
Example Software Evaluation Matrix
| Requirement | Provider A | Provider B | Provider C |
|---|---|---|---|
| Core feature 1 | 9 | 7 | 3 |
| Core feature 2 | 8 | 8 | 6 |
| Integration with Shopify | 10 | 5 | 3 |
| Reporting/analytics | 4 | 9 | 3 |
| Price/value | 7 | 6 | 8 |
| Support quality | 3 | 8 | 5 |
| Total | 41 | 43 | 28 |
This makes it clear which providers can deliver vs those that claim they can. Multiple people can contribute across different meetings. A similar matrix works for all software and services decisions.
What NOT to Install
- Keep it under ~15 apps. Don't install 40. Each one adds JavaScript, slows your site, and creates potential conflicts. There's no magic number, but once you're past 15-20 the audit gets harder than the value most apps add. Audit quarterly and remove anything you're not actively using.
- Multiple analytics apps doing the same thing
- "Speed optimisation" apps (fix your theme and images instead)
- Apps with heavy JavaScript for minor features
- Anything with forced branding visible to customers
Alongside the quarterly audit, maintain a dated change log of every app install, theme update, and site configuration change. When conversion or site speed suddenly moves, the first question is always "what changed?" - a one-line log entry turns that from archaeology into a lookup.
Site Speed Is a Revenue Lever, Not a Vanity Metric
Most founders treat site speed as a developer problem. It's a conversion problem. Google's Core Web Vitals are the three numbers that actually matter, and they map directly to whether a visitor sticks around long enough to buy. Mobile is where this bites hardest: it's 70-80% of DTC traffic and it consistently converts below desktop (Littledata's Shopify-wide benchmark puts mobile at 1.2% against desktop's 1.9% - established DTC brands typically run higher, see Section 11: Website & Conversion Optimisation), so a slow mobile experience is taxing your largest and most fragile traffic source.
The three to watch are LCP (loading, <= 2.5s is good), INP (interactivity, <= 200ms) and CLS (visual stability, <= 0.1). Section 11: Website & Conversion Optimisation owns the full Core Web Vitals treatment; here the lens is the stack. When you're outside the green band, fix it in this order:
Your store loads instantly on the office wifi on the latest iPhone. Your customer is on a three-year-old Android on patchy mobile data. Test there. The gap between those two experiences is the gap between your reported speed and your real conversion rate.
Your ERP is already the system of record. But right now most brands treat it as a glorified spreadsheet. That's about to change. AI agents will sit on top of your ERP, making real-time decisions across demand forecasting, reordering, pricing, supplier management, and financial reporting. The ERP becomes the brain, not just the filing cabinet.
Nobody knows exactly how this plays out, but when you're choosing an ERP, look beyond today's features. Look at the API, the data model, and the AI roadmap. The providers willing to rebuild around AI-native architecture will win. The ones bolting AI onto legacy systems won't.
When You Actually Need an ERP
For most brands, the answer is: not yet.
Symptoms that you've outgrown Shopify + accounting software:
- Weekly reconciliation errors between systems that should agree
- Inventory mismatches across DTC, wholesale, and marketplace channels
- Manual reporting consuming 10+ hours per week
- Purchase orders and demand forecasting done in spreadsheets that only one person understands
The ladder: The symptoms typically appear at $5M-$10M revenue with multi-channel operations (DTC + wholesale + marketplace) - that's when to start evaluating, because the manual workarounds are creating real errors and consuming disproportionate time. From there: a mid-market ERP/OMS from around $10M, a full ERP as the system of record from $15M+ (Stage 4 in the maturity model below), and an enterprise ERP with a dedicated demand planning team from $50M+. Section 7: Supply Chain & Operations runs the same ladder from the inventory side.
The warning: Premature ERP implementation can cost $50K-$200K+, take 6-12 months, and change every workflow in the business. Most brands under $5M are better off with Shopify + Xero or QuickBooks + tight processes. Get the processes right first.
See Section 26: Finance & Unit Economics for the financial reporting framework that should be working before you consider an ERP.
Data & Tracking Infrastructure
iOS 14.5 broke the old model. Before Apple's App Tracking Transparency, Meta and Google could follow your customers across the internet and report conversions with reasonable accuracy. That world is gone. Roughly 40-60% of browser-side conversion events now go untracked depending on your iOS traffic share, and that number only moves in one direction.
This isn't a Meta problem or a Google problem. It's a data infrastructure problem. If you can't capture and connect customer signals reliably, every decision downstream, ad spend, attribution, forecasting, is built on incomplete data.
The Minimum Viable Tracking Stack
At any stage, you need these fundamentals working:
- Meta Pixel + Conversions API (CAPI) - Dual tracking is non-negotiable. The Pixel captures browser-side events; CAPI sends conversion data server-to-server, recovering 20-30% of lost signals. See Section 13: Meta Ads Setup for implementation details.
- GA4 - Your source of truth for site behaviour, traffic sources, and conversion paths. Configure it properly from day one, retroactive fixes are painful.
- Event deduplication - When you're running both Pixel and CAPI, pass matching event IDs from both sources. Without deduplication, you'll double-count conversions and feed bad data to the algorithms.
Server-Side Tracking and Event Match Quality
The Pixel-plus-CAPI setup above is the floor. The next level is how well the platforms can actually match your events to real people - Meta's Event Match Quality (EMQ) score, which Section 13: Meta Ads - Setup defines and targets in full. Here the question is which tooling gets you there. A browser Pixel alone typically lands an EMQ of 3-5; you want 7-8 or higher, and you climb by sending more identifiers server-side: hashed email, phone and name, plus IP address, user agent and browser ID. Higher EMQ means better-trained algorithms, lower CAC, and conversions that don't quietly vanish.
Where you sit on the tooling ladder depends on spend:
| Setup | Best For | Notes |
|---|---|---|
| Shopify native CAPI | $0 - $10k/mo paid media | Free, built in, covers the basics. Start here. |
| Elevar / Stape | Paid media above ~$10k/mo | Dedicated server-side GTM. Elevar from ~$200/mo. Better EMQ, multi-platform (Meta, Google, TikTok) from one data layer. |
| Custom server-side GTM | $10M+ / eight-figure spend | Full control, engineering-owned. Only when data quality directly moves large ad-spend decisions. |
Running the Pixel and CAPI together without matching event IDs is worse than running neither cleanly. Both sources fire, neither knows about the other, and you report inflated conversions that train the algorithm on fiction. Pass a shared event_id from both the browser and the server for every event, then verify the dedup is working in Meta Events Manager before you trust a single number.
Consent Management
If you're selling into the EU, UK, or increasingly anywhere with privacy regulation, you need a consent management platform (CMP). Section 9: Compliance & Regulatory owns the legal obligations and penalties; here's the tooling that satisfies them. Brands selling primarily into AU/US have more runway, but the direction of travel is clear globally.
At minimum: a cookie banner that actually blocks tracking scripts until consent is given (not just a notification that does nothing), and a system that respects opt-out signals. Shopify apps like Consentmo or Pandectes handle this for most brands under $10M.
If you run Google Ads or GA4 and serve any traffic in the EEA, Google requires Consent Mode v2. It passes the visitor's consent state to Google's tags, and where consent is withheld it uses modelling to recover a portion of the conversions you'd otherwise lose to opt-outs. Most Shopify CMPs (Consentmo, Pandectes, Cookiebot) configure it for you. Scope your CMP to where you actually sell: the EU and UK are strict now, and US state privacy laws (California and a growing list) are tightening fast. Build for a first-party, consent-respecting world regardless.
First-Party Data as a Compounding Asset
Every email address, purchase history, quiz response, post-purchase survey answer, and loyalty programme interaction is first-party data you own. Unlike platform data, nobody can take it away with a policy change.
| Data Type | Where It Comes From | Why It Matters |
|---|---|---|
| Email + SMS consent | Signup flows, checkout | Direct communication channel you control |
| Purchase history | Shopify | LTV modelling, segmentation, personalisation |
| Product preferences | Quizzes, browsing, wishlists | Better targeting, product development signals |
| Acquisition source | Post-purchase survey | True attribution data, independent of platform reporting |
| Support interactions | Helpdesk | Product quality signals, retention risk |
Store it cleanly. Your Email Service Provider (ESP), Shopify, and analytics tools should share a consistent customer identity. When these systems disagree about who a customer is, you get garbage segments and wasted spend.
Common Mistakes
- Trusting platform-reported Return on Ad Spend (ROAS) as truth. Meta and Google both over-report. Always validate against blended metrics (Marketing Efficiency Ratio (MER), nCAC). See Section 27: Measurement & Data.
- Not deduplicating events. This inflates reported conversions and trains the algorithm on bad data.
- Ignoring consent requirements. One General Data Protection Regulation (GDPR) complaint from a European customer can cost more than your entire tracking stack.
- Over-building too early. A $300K brand doesn't need a customer data platform. Get the basics right first.
Integration Architecture
Your tech stack isn't a collection of tools. It's a system, and how the pieces connect matters as much as which pieces you choose.
The core data flow for most DTC brands: Storefront → ESP/SMS → Analytics → Support → 3PL → Accounting. Orders flow out, customer data flows through, financial data flows into your books. When this chain works, operations run cleanly. When it doesn't, you get manual exports, data discrepancies, and people spending hours reconciling spreadsheets.
Connected vs. Integrated
- API-native integration: Data flows automatically, in real time, no manual steps. Shopify to Klaviyo is a good example.
- Middleware (Zapier, Alloy, Mesa): Bridges gaps between tools that don't talk directly. Fine for low-volume workflows, but fragile at scale.
- Manual sync: Someone exports a CSV and imports it somewhere else. This is not integration.
Warning Signs of Integration Debt
- Revenue in Shopify doesn't match revenue in your accounting tool, and nobody knows why
- Customer data in your ESP is days behind what's in Shopify
- Your 3PL and Shopify inventory counts disagree regularly
- Someone on the team has a "Monday morning reconciliation" ritual
If you recognise two or more of these, you have integration debt. It doesn't mean you need to hire a developer tomorrow, but it means you should stop adding tools and start connecting the ones you have.
Architecture Maturity Model
Most brands move through these stages. The mistake is jumping ahead before the pain justifies it.
Workflow Automation: Wiring the Gaps
Where two tools have no native integration, orchestration is how you close the gap without writing code. The manual copy-paste between Shopify and your accounting software. The supplier email someone retypes into a stock sheet. The Monday-morning dashboard someone pulls by hand. None of it is hard work. All of it is a person doing what software should.
Workflow automation, or orchestration, is the glue. It wires your tools together so data moves itself: an order triggers an accounting entry, a low-stock signal fires a reorder alert, a daily sales pull lands in a dashboard before you're awake. On top of that, AI copilots can now handle scoped judgement work inside those flows, drafting the supplier follow-up, flagging the reconciliation that doesn't tie out, summarising the week. The three workhorse platforms sit on a complexity ladder:
| Tool | Complexity | Use it when |
|---|---|---|
| Zapier | Low-code, visual | You want simple A-to-B automations live this week with no engineer |
| Make | Mid, steeper curve | You need multi-step branching logic and more control over the data flow |
| n8n | Self-hosted, open source | You've got technical help and want to own the stack and the data |
Orchestration is for the gaps, not the backbone - the middleware caveat above still holds, so as volume grows, push the high-traffic flows onto native API integrations and keep automation for the long tail.
One thing to get right first: you can't automate what you haven't documented. Write the SOP, get the process stable and repeatable, then automate it. Wiring up a messy, undefined process just bakes the mess in faster.
CMS & Product Data Management
One of the biggest operational unlocks at Quad Lock, and something we should have done years earlier, was moving to a custom CMS using Sanity. Every time Apple or Samsung released new devices, we needed to create multiple new cases that then propagated across all mount kits, across six Shopify stores.
Before Sanity, every iPhone launch meant a 5am start for the whole team, manually updating content across stores with potentially hundreds of product combinations. The web team eventually got this down to a two-person, one-hour process. It reduced errors, increased efficiency, and meant the website was live faster at a time of very high purchase intent.
For most brands, Shopify's built-in product management is perfectly adequate. If you have fewer than a few hundred SKUs, sell in one or two markets, and don't have complex variant structures, Shopify's admin does the job.
It stops being enough when:
- You're managing large catalogues with hundreds of variants across multiple product lines
- You're selling into multiple markets with localised content, pricing, and translations
- Product launches require coordinated updates across many pages simultaneously
This is where a headless CMS or PIM starts earning its place. A headless CMS like Sanity, Contentful, or Storyblok manages all your content in one place and pushes it to Shopify (or multiple storefronts) via API. We moved to Sanity because the operational pain was real and measurable, as covered earlier in this section.
Worth noting: we did much of this before Shopify Markets existed. Today, Markets handles a lot of the multi-region complexity that used to require separate stores and custom tooling. If you're starting now, evaluate what Shopify gives you natively before building around it.
PIM: What It Is, When You Need One
A PIM centralises all your product data, descriptions, specs, images, translations, attributes, in one system. Changes propagate everywhere: your website, marketplaces, retail feeds, printed catalogues.
Most brands under $10M don't need a PIM. The overhead isn't justified until you're managing 500+ SKUs across multiple channels or markets, or your product data quality is actively costing you sales. If you're at that point, tools like Akeneo, Salsify, or Plytix are worth evaluating. But be honest about whether you have a tooling problem or a process problem.
Before investing in a PIM or headless CMS, audit your actual pain points. If the biggest issue is "updating product descriptions is slow," the fix might be a better internal process, not a new platform.
For how your data and reporting stack should work once the systems are connected, see Section 27: Measurement & Data.
With the tech stack in place, the next step is making it convert. Section 11: Website & Conversion Optimisation covers how to turn your store into a selling machine.
Section 10 Checklist
Go from reading to doing.
You've read the section. Sign in to score your ecommerce brand with the Health Check, track your progress across 472 checklist items, and get your tools and history in one dashboard. Free, and always will be.