How to Choose an Ecommerce Development Company

Search for an ecommerce development company and you get two kinds of page. Agencies describing why they are the right agency, and “top 10 agencies” lists where placement was bought. Neither is a buyer’s guide. Both are sales collateral with a table of contents.

TopTut sells no development services, lists no agencies here, and takes no referral fee from anyone in this market. That is the only reason this page is worth more to you than the ones above it: nobody at the end of it is trying to be the answer.

The thesis is simple. The platform decision determines the agency decision, not the other way round. Almost every bad ecommerce build traces back to hiring a builder for a job that needed an architect, or an architect for a job that needed a builder. Get the scope right and the shortlist writes itself.

How to choose an ecommerce development company, in one pass

Settle what kind of project you have, write one brief with a named budget band, send it to three suppliers of the right type, and score their answers to ten specific questions. That is the whole method. Everything else in this post is detail on those four steps.

The first question to settle before you contact anyone is this: are you buying a store, or are you buying a system? A store is a catalogue, a checkout, and a set of templates that look like your brand. A system is a store plus logic that does not exist off the shelf, such as configurable products, trade pricing tiers, ERP-driven stock, or a subscription model your platform does not natively support. Those are different purchases from different firms.

On money: the shape of this market is public. Clutch’s ecommerce developers directory filters suppliers into hourly bands of under $25, $25 to $49, $50 to $99, $100 to $149 and $150 to $199, and into minimum project sizes of $1,000, $5,000, $10,000 and $25,000 and up. On the platform side, published starting prices for migration work in the Shopify Partner Directory ran from around $500 to $11,000 per partner when checked in August 2026, with one listing advertising a range that topped out in the six figures. Together they describe a market spanning roughly two orders of magnitude for work that all gets called “an ecommerce build”.

Decide the platform first, because it decides the supplier

There are four distinct project types, and each one needs a supplier with different skills, a different team shape and a different price. Confusing them is the single most common cause of a build going wrong, because you cannot tell from the outside whether the firm quoting you has ever done the version of the job you actually have.

Project type	What it actually involves	Typical cost band and where that band comes from	Who to hire
Themed build on an existing platform	Buy or license a theme, configure it, load the catalogue, set up payments, shipping and tax, launch. Little or no custom code.	Low four figures to low five figures. Shopify Partner Directory setup and migration listings start around $500 to $2,000; Clutch’s smallest minimum project size band is $1,000 and up.	Freelancer or small team. A full-service agency will price this at three times what it needs to cost.
Customised platform build	A theme or starter stripped back and rebuilt, custom templates, two to five integrations, some app or plugin development, a real design phase.	Mid five figures. Worked example: 400 billable hours at Clutch’s $100 to $149 band is $40,000 to $59,600, before content and photography.	Boutique specialist agency that works on your platform daily.
Replatform and migration	Everything above, plus data migration, URL mapping, redirect implementation, integration rebuilds and a cutover plan with rollback.	Same build cost as above plus a migration workstream. Shopify Partner Directory migration listings alone advertise starts from $500 to $11,000 depending on catalogue size.	Specialist agency with named migration references on your source platform.
Genuinely custom	Headless or bespoke architecture, custom order management, complex B2B logic, a development team on a roadmap rather than a project.	Six figures and ongoing. Clutch’s top hourly band runs to $199 an hour; sustained work at that rate is a retained team, not a project fee.	A development partner with a named architect and a contract that assumes years, not weeks.

If you are in the first row, you probably do not need an agency at all. A capable freelancer plus a good theme gets a small catalogue trading, and our guide to building a WordPress ecommerce site covers most of what that supplier would be doing for you anyway.

If you are in the fourth row, be honest about why. “Custom” is often a way of saying nobody has done the work of mapping requirements to what the platform already does. The genuine cases are real, though, and they tend to look like operational logic rather than visual ambition. Our note on where WordPress stops being the right tool is a fair sketch of the moment a platform build turns into a systems build.

Platform licence cost feeds this decision too. Shopify’s published pricing runs from $39 a month on Basic to $399 on Advanced, with Plus starting at $2,300 a month. A supplier who quotes you a custom build without asking which tier you are on, or whether the tier is the constraint, has not read your problem.

The three supplier types, and what each one will not do

Suppliers come in three shapes: the freelancer or small team, the boutique specialist agency, and the full-service or offshore agency. Each is genuinely the right answer for some projects and genuinely the wrong answer for others, and the failure modes are predictable enough to plan around.

Freelancer or small team

Strength: you talk to the person doing the work, decisions take minutes, and the price reflects one person’s time rather than a company’s overhead. For a themed build or a focused piece of custom work, this is usually the best value available.

Real risk: single point of failure. Illness, a better-paying client or a quiet disappearance leaves you with a half-finished build and no second person who knows how it works. What they typically will not do is a structured discovery phase, formal QA across devices and browsers, or support with a response time attached to it.

Boutique specialist agency

Strength: they do one platform, all day. That means they have already hit the edge cases your build will hit, they have internal tooling for the repetitive parts, and their estimates are based on having done it rather than on hope.

Real risk: capacity. A ten-person shop with two good developers takes on a project the week both are booked, and your build queues behind whoever signed first. What they typically will not do is stray far from their platform. Ask a Shopify boutique for a headless build against a legacy ERP and you get a yes that costs you a year.

Full-service or offshore agency

Strength: bench depth and price. A large agency can staff design, development, QA and project management simultaneously, and an offshore team can do that at a rate a domestic firm cannot match. The capability argument against offshore delivery is mostly snobbery, and the arithmetic explains why the rate gap exists without implying anyone is worse at the job.

The US Bureau of Labor Statistics puts the median annual wage for web developers at $92,650 as of May 2025, with web and digital interface designers at $104,000. Divide $92,650 across a 2,080-hour year and you get about $44.50 an hour in salary alone, before employer costs, overhead, non-billable time and margin. That is why nobody billing under $25 an hour is paying US salaries: it is a labour-market fact, not a quality signal.

The real risk with offshore delivery is communication overhead and timezone-lagged iteration. A question that takes ninety seconds to answer in person takes a day when your working hours barely overlap, and a build is thousands of small questions. Each round of “not quite, move it left” costs a calendar day instead of a coffee. What a large or offshore agency typically will not do is give you an account manager who can make a technical decision on the call, which is precisely the thing that keeps a build moving.

Fix it with process, not prejudice. Insist on a two-hour daily overlap window, one named technical contact rather than a relay through account management, and a staging URL that updates continuously.

Writing a brief that gets comparable quotes

A brief gets comparable quotes when every supplier is pricing the same scope, which means you have to supply the scope rather than asking them to guess it. Three quotes for three different imagined projects tell you nothing except who is most optimistic.

Include these, specifically: your current platform and monthly traffic; the business outcome you are buying, stated as a number or a behaviour rather than “a modern site”; every integration that must work on day one, including ERP, payment gateway, shipping rate provider, tax calculation and any PIM; catalogue size and complexity, meaning SKU count, variant depth and how many products are configurable; who is supplying copy and photography; your launch constraint and the reason behind it; and your budget band.

Here is a template you can paste into an email and fill in. It is deliberately short. A brief nobody finishes writing is worth less than a page of specifics.

ECOMMERCE BUILD BRIEF

1. Current state
   Platform and version:
   Monthly sessions / monthly orders:
   Average order value:
   Live URL:

2. Outcome we are buying
   The one thing this build must change:
   How we will know it worked:

3. Must-work integrations (day one)
   ERP / accounting:
   Payment gateway(s):
   Shipping rates and labels:
   Tax calculation:
   PIM / product data source:
   Email and CRM:

4. Catalogue
   SKU count:
   Variants per product (max):
   Configurable or made-to-order products? Y/N, describe:
   Data currently lives in:

5. Content and design
   Copy supplied by:
   Photography supplied by:
   Brand guidelines exist? Y/N
   Design from scratch, or theme customisation?

6. Constraints
   Launch date and the reason for it:
   Anything we cannot change (hosting, ERP, payment provider):

7. Budget
   Band for build:
   Band for first-year support and iteration:

8. What we want back
   Fixed-price proposal broken into phases, hours per phase,
   named team members, and anything you consider out of scope.

That last line does more work than the rest of the document. The out-of-scope list is where a supplier tells you what they are quietly assuming, and comparing three of those lists is more revealing than comparing three prices.

Ten questions to ask, with the answer you want

Ask these ten on the first technical call, before any proposal exists, and write down the answers. None of them require you to be technical. All of them are questions a competent supplier answers without hesitating, because they have answered them before.

Question	The answer you want	The red-flag answer
Who owns the code and design files when you are paid?	You do, assigned in writing on final payment, including source files.	“We retain the framework” or “the design files stay with us”.
Is there a staging environment I can see?	Yes, a password-protected staging URL from week one, updated continuously.	Screenshots, a slide deck, or “we will show you at the end”.
Do you use version control, and where does the repo live?	Git, in a repo you get access to, ideally in your own organisation account.	“We keep backups”, or a repo only they can see and will not transfer.
How do you handle URL redirects in a migration?	A full crawl of the old site, a mapped 301 for every URL, signed off before cutover.	“The platform handles that” or “we redirect the main pages”.
What happens to my search rankings during a replatform?	A named plan: redirect map, structured data parity, staging blocked from indexing, post-launch crawl.	“SEO is a separate service” or a promise nothing will change.
Who is actually doing the work, employees or subcontractors?	Named individuals, their role, and whether they are staff or contracted, stated plainly.	Evasion, or “our team” with no names offered.
What handover documentation do I get?	Written docs covering architecture, integrations, credentials, deploy process and known limitations.	A recorded walkthrough call and nothing else.
What is the post-launch warranty period?	A defined window, typically 30 to 90 days, with defects fixed free.	No warranty, or bugs billed as change requests from day one.
How are change requests priced?	A stated hourly or day rate, with a threshold under which small changes are absorbed.	“We will work it out as we go.”
What does support cost after launch?	A monthly retainer figure or a rate card, given before you sign the build.	Deferred to later, which means priced once you cannot leave.

The SEO clause nobody writes into the contract

Put the redirect map in the contract as a named deliverable with a sign-off date, because a replatform without one is the most expensive routine mistake in ecommerce and most agencies treat redirects as out of scope by default. You lose the rankings, then you pay someone else to work out which URLs used to exist.

Google’s own site move documentation is unambiguous about the mechanism: use server-side permanent redirects, 301 or 308, wherever technically possible, and keep them in place for as long as you can, generally at least a year, so signals transfer to the new URLs. It also tells you to update internal links to point directly at the new URLs rather than relying on the redirect chain, and to submit the new sitemap in Search Console. None of that happens by itself during a platform migration.

Five things to require in writing, worded as deliverables rather than intentions:

• A full crawl and URL inventory of the existing site, delivered before build starts, covering products, categories, paginated series, blog posts and any legacy URLs still receiving traffic.
• A 301 redirect map, one row per old URL, signed off by you before cutover, with a stated rule for what happens to discontinued products.
• Structured data parity on launch. Whatever Product, Offer, BreadcrumbList and Review markup the old site emitted, the new one emits, validated against Google’s product structured data documentation.
• No staging site left indexable. Password protection during build, and an explicit check that the live site is not shipping the staging robots.txt or a stray noindex header.
• A post-launch crawl within 72 hours, comparing the new URL set against the redirect map, with 404s and redirect chains fixed inside the warranty window at no cost.

If your supplier says SEO is somebody else’s job, the clause above is still yours to insist on, because it is not optimisation. It is not losing what you already have. Our ecommerce SEO checklist covers the crawl and index control work in detail, and if the migration itself is still undecided, the comparison of Shopify alternatives goes through what moves cleanly between platforms and what does not.

Contract terms that decide who owns what

Six terms determine whether you own a business asset at the end or rent one indefinitely, and all six are usually absent from a standard agency proposal because nobody asked.

• IP assignment on final payment. Written, covering code, design files and any custom app. Suppliers may reasonably carve out their pre-existing frameworks and libraries; get that carve-out named rather than implied.
• Milestone payments tied to deliverables, not dates. “40% on staging sign-off” creates an incentive to reach staging. “40% on 1 November” creates an incentive to reach 1 November.
• Licences, domains and hosting registered in your accounts. You pay, from your card, on your login, with the agency added as a collaborator. Every horror story about a hostage website starts with the agency owning the registrar account.
• Source code and asset handover as a named deliverable. Repository access transferred, plus layered design files and original images, delivered before the final invoice is settled.
• Defined acceptance criteria. A written list of what “done” means, ideally as test cases: a guest checkout completes, an order reaches the ERP, a tax rate calculates correctly for three named jurisdictions.
• A named warranty window. Thirty to ninety days in which defects against the acceptance criteria are fixed at no charge, with the distinction between a defect and a change request written down before it is contentious.

Red flags in a proposal or a first call

Six signals reliably predict a bad engagement, and they show up before you have spent anything. Any one of them is a conversation. Two or more is a decline.

Red flag	Why it predicts trouble	What to do
No staging environment	You cannot review work in progress, so every problem is found at launch when changing it is most expensive.	Make a staging URL a condition of the first milestone payment.
A quote given without looking at your current site	They have priced a generic project, so the specifics of yours become change requests later.	Send the URL and ask what they found. A health check of your existing site gives you a baseline to compare their reading against.
Portfolio links that are dead or rebuilt	Either the work was long ago or the clients did not stay. Both are worth knowing.	Ask for two live sites launched in the last twelve months and contact those clients.
Refusal to name the team doing the work	Usually means it is being subcontracted, which is fine if disclosed and a problem if hidden.	Ask directly. The answer matters less than the willingness to give one.
A fixed price for an undefined scope	The margin is protected by narrowing the scope later, and you will be the one arguing about what was included.	Insist on a written scope and out-of-scope list attached to the price.
Pressure to sign before a technical discovery call	Discovery is where the estimate becomes real. Skipping it means the estimate is a guess with a signature under it.	Pay for a short paid discovery instead of signing the full build.

A paid discovery is the best money in this process. It buys you a specification you own, which you can then quote to three suppliers on identical terms.

The total cost you did not budget for

Build cost is typically a fraction of first-year cost, and the difference lives in five places that almost never appear in a proposal because they are your work, not the agency’s.

Content migration and product data cleanup. Exporting a catalogue is quick. Fixing it is not. Inconsistent attribute names, three spellings of the same brand, descriptions written by four people over eight years. Somebody sits with a spreadsheet for weeks, and it is usually you, because nobody else knows which of two conflicting values is correct.

Photography. A new theme with consistent image ratios exposes every product shot taken at a different crop against a different background. Either you reshoot, or the new site looks worse than the old one on the pages that sell.

Integration edge cases. The integration works. Then a partial refund on a multi-currency order with a discount code fails to write back to the ERP. Budget for the second and third pass on every integration, not just the first.

Training. Your team knew the old admin. Order processing slows for a month, and if you did not book training the questions arrive as support tickets you are billed for.

The first six months of support and iteration. Launch is where you find out what you got wrong, and fixing those things is where a lot of the value is. If you spent every available pound on the build, you will spend the next six months unable to act on anything you learn from it. Ring-fence a support budget before you sign, and make the supplier quote it as a monthly figure at the same time as the build.

Verdict: which ecommerce development company suits which build

Match the supplier to the project type, not to the impressiveness of the pitch deck. Here is the position, stated plainly.

Under about $10,000, themed build, small catalogue: hire a freelancer or a two-person team, on a fixed scope, with staging and repo access written in. An agency at this budget is buying you overhead, not capability.

Roughly $15,000 to $75,000, customised build or replatform: hire a boutique specialist that works on your target platform every day, and pay for discovery separately before committing to the build. This is the band where the SEO clause and the redirect map earn their keep several times over.

Six figures, complex B2B logic, ERP-driven operations, headless: hire for architecture rather than delivery, insist on a named technical lead who stays on the account, and structure it as a retained team on a roadmap. A fixed-price project contract at this complexity produces a fight, not a system.

Offshore or large full-service: a legitimate choice at any of these bands if you can supply the process. Daily overlap window, one named technical contact, continuous staging. Without those three, the rate saving is spent on elapsed time.

When you should not hire anyone at all

Do not hire a development company if you cannot state, in one sentence, what the build must change about the business. That is not a briefing problem, it is a strategy problem, and no supplier can solve it for you at their hourly rate.

Three other cases where the answer is no. If traffic is small and conversion rate is the constraint, a rebuild changes nothing you could not change with copy, photography and a faster checkout today. If your catalogue data is a mess, fix the data first: migrating chaos costs the same as migrating order and leaves you with chaos. And if you have not kept your current site updated, a bigger custom site will be harder to maintain, not easier, and the bill arrives monthly.

The best outcome in this market is often a smaller purchase than the one you set out to make: a themed build done properly, launched in eight weeks, with the money you saved spent on the product data, the photography and the first six months of iteration. Very few agencies will tell you that, which is more or less the point of this page.

Frequently asked questions