Question 1

When is headless commerce worth the added technical complexity?

Accepted Answer

Headless commerce is worth the operational complexity when standard themes cannot support complex merchandising logic, multi-region catalogues, or strict site performance targets.

Question 2

Should product variants have their own URLs?

Accepted Answer

Usually not as separate indexable pages. The default is one canonical product page with variants selectable on it, and variant URLs canonicalising to the parent — otherwise twenty near-identical pages compete with each other. The exception is a variant with real independent search demand, such as a named colourway people search for. Then give it a genuinely distinct page with its own title, copy and images, not a duplicate.

Question 3

Does structured data improve search rankings?

Accepted Answer

Not directly. It decides whether your result displays price, availability and ratings, which changes click-through at the same position, and it is increasingly how AI answer engines extract facts about your products. The rules that matter: never mark up something the page does not display, never invent an aggregate rating, and keep identifiers consistent so the site reads as one entity rather than a pile of unrelated snippets.

Question 4

What are Shopify metafields used for?

Accepted Answer

Metafields attach typed, structured, queryable data to products, variants, collections, customers and orders — specifications, materials, compliance information, size guides, certifications. Metaobjects go further and let you define whole new content types with their own fields. Together they extend Shopify's product model far enough that many merchants who think they have outgrown the platform have actually just not modelled their catalogue properly.

Question 5

Can Shopify checkout be customised?

Accepted Answer

Yes, through checkout extensibility rather than by editing checkout templates. Checkout UI extensions add components at defined points, Shopify Functions run server-side logic for discounts, shipping and payment rules, and the branding API controls appearance. Several of these require Shopify Plus. What you cannot do is replace the flow itself — a quote-and-approve cycle or staged payments sits beyond what extension supports.

Question 6

Should I install a Shopify app or build a custom one?

Accepted Answer

Install when the problem is generic and the app is well-maintained and uses app blocks rather than writing into your theme. Build when the logic is specific to your business, when you are stacking three apps to approximate one behaviour, or when you need a private integration with an internal system. And check first whether Shopify or Shopify Flow already does it — native capability has grown faster than most merchants' picture of it.

Question 7

How many Shopify apps is too many?

Accepted Answer

There is no fixed number, but the questions worth asking are how many inject scripts into your storefront, how many are on your critical path, and how many overlap. Three apps approximating one behaviour is a stronger signal than any count. Audit by removing one at a time on a duplicate theme and seeing what actually breaks — the answer is often less than expected.

Question 8

Is headless commerce faster than a traditional storefront?

Accepted Answer

It raises the performance ceiling but does not raise the floor. A headless storefront that fetches on the client can easily be slower than the theme it replaced.

Question 9

What is the difference between headless and composable commerce?

Accepted Answer

Headless separates the storefront from the commerce engine. Composable goes further and separates every capability — search, content, cart, payments, product information — into independently chosen services. Headless is one decoupling; composable is many, and the sum is a distributed system requiring an orchestration layer and a platform team. Composable suits large organisations and is a poor default for a mid-sized retailer.

Question 10

Can I go headless with Shopify and keep Shopify checkout?

Accepted Answer

Yes, and it is the version of headless with the best risk profile. You build the storefront against Shopify's Storefront API and hand off to Shopify's own checkout at the end. Payments, fraud screening, PCI scope, tax and order management stay Shopify's responsibility while you own the front end completely. You can use Hydrogen or any framework — Next.js against the Storefront API is a common and sound choice.

Question 11

What does a headless storefront need that a theme does not?

Accepted Answer

A rendering strategy per route, a cache invalidation mechanism driven by platform webhooks, SEO output written as application code — canonicals, sitemaps, structured data, redirects — a preview environment for merchandisers, and hosting with monitoring and on-call. All of it was handled by the theme layer before, invisibly. Budget for it as engineering, because that is what it is.

Question 12

How do you stop AI automation from making expensive mistakes?

Accepted Answer

Design the failure mode rather than trusting the output. Write to a draft state instead of the live system, auto-approve only above a confidence threshold, keep a human between the system and anything a customer sees, make every action logged and reversible so a systematic error can be undone as a set, and sample outputs weekly rather than verifying once at launch. Safety comes from the workflow, not from the model.

Question 13

Do I need AI to automate my ecommerce operations?

Accepted Answer

Often not. A large share of what gets scoped as an AI project turns out to be an integration problem — two systems exchanging a spreadsheet because nobody connected them. Ordinary code that moves and validates data solves those completely, costs less, and never produces a confidently wrong answer. Rule that out first; use a model for the part that genuinely requires interpreting unstructured input.

Question 14

What are good Core Web Vitals scores for an ecommerce site?

Accepted Answer

LCP ≤ 2.5s, INP ≤ 200ms and CLS ≤ 0.1, at the 75th percentile of real visits — assessed separately for each URL group.

Question 15

Why is my Largest Contentful Paint slow on product pages?

Accepted Answer

Usually the product image is lazy-loaded when it should not be, is served far larger than displayed, or is discovered late because it is referenced from CSS or injected by JavaScript. Break LCP into its four parts — server response, load delay, load time and render delay — and you will normally find one dominant. A rotating hero carousel is an LCP problem by construction and is worth replacing outright.

Question 16

How do I improve Interaction to Next Paint on a storefront?

Accepted Answer

Start by auditing third-party scripts, which are responsible more often than application code. Remove everything nobody can name an owner for, then load the survivors after first interaction where possible. After that, break long synchronous tasks into chunks that yield to the main thread, move heavy product filtering to the server, virtualise long lists, and reduce how much of the page hydrates.

Question 17

Do Core Web Vitals actually affect ecommerce rankings?

Accepted Answer

They are a ranking signal, but a modest one — relevance and content quality dominate. The stronger case for fixing them is commercial rather than algorithmic: they measure how long a customer waits, whether a tap responds, and whether the button moves before they hit it. Those affect conversion directly, which is a better reason to act than a small ranking adjustment.

Question 18

Is a headless storefront faster than a Shopify theme?

Accepted Answer

Not inherently. A well-built Liquid theme renders server-side on Shopify's infrastructure and is genuinely fast. Headless raises the ceiling — you control exactly what ships and can serve pre-rendered HTML from an edge cache — but it also makes it easy to build something slower by fetching data on the client. And it carries the same third-party tags across, which are usually the actual problem.

Question 19

Which ecommerce tasks are worth automating with AI?

Accepted Answer

Product data enrichment, support triage, content drafting, return classification, feed normalisation and internal search — tasks with volume, clear inputs, checkable outputs and recoverable failures.

Question 20

Should AI handle customer support conversations directly?

Accepted Answer

Not autonomously. A model answering policy questions from its own knowledge will occasionally invent a returns window or a delivery promise, and a customer told something by your website will reasonably expect it honoured. The version that works retrieves your actual policy text, answers only from it, cites the source and escalates to a human when confidence is low. Triage and routing are safer and capture most of the value.

Question 21

What is the difference between an AI workflow and an AI agent?

Accepted Answer

Who decides the sequence. In a workflow you wrote the steps and the model does one bounded job at each — classify, extract, draft — which is predictable and testable. An agent chooses its own steps, which is more flexible and has open-ended failure modes. For ecommerce operations, workflows are right for most cases; reserve agents for genuinely variable investigation, with read-only access and a human committing any change.

Question 22

What is an AI CRO agent and how does it work?

Accepted Answer

An AI CRO agent is an automated tool that analyses site interaction data and visual hierarchy to generate prioritised audit reports and actionable fixes.

Question 23

What kinds of custom software do you build?

Accepted Answer

B2B ordering platforms, customer and partner portals, internal operations tools, and the integration layers between ERP, accounting, logistics and commerce systems.

Question 24

Which technologies do you use for custom software?

Accepted Answer

Laravel or Node.js over PostgreSQL, with Redis, React and Next.js — mainstream choices, so you can hire others to maintain it later.

Question 25

How is InvoiceKit delivered and installed on our servers?

Accepted Answer

InvoiceKit is delivered as a Git repository with complete Laravel and Vue source code that you clone and deploy directly to your web hosting environment or container platform.

Question 26

Do I need a developer to customise a Shopify theme?

Accepted Answer

Not for layout and content changes on a 2.0 theme — sections and JSON templates cover those. You need one when behaviour changes or when new structured data has to reach the storefront.

Question 27

How do you stop an AI system inventing answers?

Accepted Answer

By grounding every answer in retrieval from your own data, measuring confidence against an evaluation set, and escalating to a person below the agreed threshold.

Question 28

When should a business choose headless commerce over a standard Shopify theme?

Accepted Answer

Choose headless commerce when monolithic themes constrain your render speeds or when you need multi-region content routing and complex custom layouts that native theme editors cannot support.

Question 29

What is the primary difference between Hyperband and Bayesian Optimization?

Accepted Answer

Bayesian Optimization uses a probabilistic model to predict top configurations, while Hyperband uses early stopping and multi-armed bandit resource allocation.

Question 30

Who is eligible for this free Shopify store audit?

Accepted Answer

This audit is designed for established e-commerce brands generating over £1m in annual revenue on Shopify or Shopify Plus with active marketing traffic.

Question 31

Who qualifies for this free Shopify store audit?

Accepted Answer

This audit is designed for established brands generating over £1M annually on Shopify or Shopify Plus experiencing flattened conversion or technical debt.

Question 32

Can I use Shopify Markets if I have separate warehouses in the UK and the EU?

Accepted Answer

Yes. Shopify Markets supports multi-location inventory mapping to route orders to specific regional warehouses, provided your accounting structure allows consolidated financial reporting.

Question 33

Can InvoiceKit be hosted on standard server infrastructure?

Accepted Answer

Yes. InvoiceKit runs on standard Web hosting or virtual private servers running PHP, a relational database, and background job workers.

Question 34

How do you choose between Next.js, Vue.js, Node.js and Laravel?

Accepted Answer

Next.js where SEO and server rendering matter, Vue.js where your team already runs Vue; Laravel for a rich admin and mature ecosystem, Node.js for high-concurrency APIs sharing types with the front end.

Question 35

What is the difference between Shopify and Shopify Plus?

Accepted Answer

Plus adds checkout extensibility, Shopify Functions, B2B company accounts, Markets and higher API limits — the capabilities that matter at scale.

Question 36

Should we build a custom ecommerce site or use Shopify?

Accepted Answer

Shopify unless your catalogue, pricing or checkout genuinely cannot fit it. Coder71 builds both, so the recommendation is not a sales pitch.

Question 37

Do we own the code and the store?

Accepted Answer

Yes — you own the repository, the store, the hosting and the analytics, and documentation is handed over with the build.

Question 38

Who needs a custom ecommerce website rather than a template?

Accepted Answer

Businesses whose catalogue, pricing or fulfilment rules a hosted platform cannot express, and stores whose existing code has become too risky to change.

Question 39

When is Shopify the wrong choice?

Accepted Answer

When weight-based pricing, delivery slots, substitutions, multi-vendor payouts or entitlement models are core — then a custom or hybrid platform fits better.

Question 40

Can you work alongside our in-house Shopify team?

Accepted Answer

Yes — either as a scoped project alongside your roadmap, or as staff augmentation with a Shopify engineer embedded in your team.

Question 41

What technology stack do you recommend for a custom online store?

Accepted Answer

Our default is Next.js and React for the storefront, Node.js or Laravel for the commerce and integration layer, and PostgreSQL for data. The database choice is the least negotiable: ecommerce data is deeply relational, and constraints and foreign keys prevent the kinds of inconsistency that are expensive to reconcile later. The front-end choice matters less than whether someone owns the caching strategy.

Question 42

Do you work with businesses outside your own timezone?

Accepted Answer

Yes. Coder71 works across the US, Canada, UK, Australia, Europe and the Middle East, with a senior engineer overlapping your business hours.

Question 43

Are you a Shopify Partner?

Accepted Answer

Yes, Coder71 has been a Shopify Partner since 2017 and works on Shopify and Shopify Plus stores weekly.

Question 44

Can you customize Shopify checkout?

Accepted Answer

On Shopify Plus, yes — through checkout UI extensions and Shopify Functions. Below Plus the platform limits what can be changed.

Question 45

Should we build a custom app or use one from the App Store?

Accepted Answer

Use a listed app where one fits. Build custom when no listing matches your logic, or replace it with a Shopify Function, which adds no page weight.

Question 46

How long does a custom ecommerce build take?

Accepted Answer

It depends on the six drivers in this article — catalogue complexity, integration count, migration scope, checkout requirements, content tooling and internationalisation — and any number quoted without knowing those is guesswork. What is predictable is the shape: the data model and the hardest integration should be settled early, and a project that leaves either until later will run longer than planned.

Question 47

Can you migrate our store to Shopify without losing rankings?

Accepted Answer

Yes — a complete URL map, permanent redirects, structured data rebuilt before launch, and customers and order history migrated alongside the catalogue.

Question 48

When should a business build custom software instead of buying?

Accepted Answer

When the business rules cannot be configured into any product, and the workaround already has a person's time permanently attached to it.

Question 49

What happens if we want to change developers later?

Accepted Answer

You can. Documentation, runbooks and tests live in your repository, and the stack is deliberately mainstream so other teams can pick it up.

Question 50

What should not be automated?

Accepted Answer

Goodwill decisions, damaged items, angry customers and anything with a legal or safety dimension — the cost of getting those wrong exceeds the saving.

Question 51

How is success measured?

Accepted Answer

Resolution without human contact at thirty days, not tickets deflected. A closed ticket that did not solve the problem has only moved the contact.

Question 52

Does it work with our helpdesk and store?

Accepted Answer

Yes. It integrates with your existing helpdesk, store and ERP rather than requiring your team to work in a new tool.

Question 53

Why are my product pages not being indexed?

Accepted Answer

The three usual causes are crawl budget consumed by faceted URLs before crawlers reach the products, products reachable only from deep paginated listings with no other internal links, and pages whose content arrives by client-side JavaScript so crawlers see an empty container. Check Search Console's "Crawled – currently not indexed" report, then load a product page with JavaScript disabled. One of those two will usually explain it.

Question 54

How many words should an ecommerce category page have?

Accepted Answer

There is no threshold, and chasing one produces the keyword blocks at the bottom of category pages that nobody reads. What matters is whether the page answers the questions somebody choosing within that category has — how to pick between sub-types, what the differences are, what fits what. Two genuinely useful paragraphs above the grid and a longer guide below it beats a thousand words of filler.

Question 55

Do we own the source code?

Accepted Answer

Yes — the repository, schema and hosting are yours from the first commit, with documentation, runbooks and tests handed over.

Question 56

How do you handle integrations with our existing systems?

Accepted Answer

With retries, idempotency and reconciliation built in, so failures are visible and recoverable rather than silently lost.

Question 57

What does custom software development cost?

Accepted Answer

It depends on the number of business rules, integrations, data migration and user roles. Coder71 scopes it in a free consultation before quoting.

Question 58

What can AI actually automate in an ecommerce business?

Accepted Answer

Order status, returns eligibility, product questions, ticket triage and routing, and catalogue content — work that is repetitive, high volume and checkable against your own data.

Question 59

Do we keep ownership of our analytics and ad accounts?

Accepted Answer

Yes. Analytics, Search Console, ad accounts and tag manager stay in your ownership from the start.

Question 60

How do you approach AI search and answer engines?

Accepted Answer

By structuring content around questions and entities, with matching structured data, comparison tables and named expertise — and being honest about what is measurable.

Question 61

Is there a long contract?

Accepted Answer

A three-month minimum, then rolling month to month. No long-term lock-in.

Question 62

Can you do SEO if you did not build our store?

Accepted Answer

Yes. Coder71 implements directly on mainstream platforms, or hands ready-to-ship changes to your own developers.

Question 63

What is IT staff augmentation?

Accepted Answer

Placing vetted engineers into your existing team under your process and roadmap. You own delivery; the supplier supplies and manages the people.

Question 64

How is this different from hiring a freelancer?

Accepted Answer

The engineer has a team behind them and a manager who is not you — so escalation, leave cover, payroll, compliance and replacement are the supplier's problem.

Everything clients ask, answered.

General

Tell us the goal. We’ll scope the work.