How to Choose an Ecommerce Web Design Agency
Commissioning a shop is buying a relationship that outlives the invoice, because somebody has to look after it afterwards. Most buyers study the portfolio carefully. Almost none ask what the following two years look like.
What An Ecommerce Agency Actually Does
An ecommerce agency covers rather more than design. Structure and category planning, the design of every page type, the build itself, product data, integrations, content, testing and launch. The variation between agencies is not in that list. It is in where the list stops.
What sits inside the work. Deciding how the catalogue is organised. Designing each page type around what it has to do. Configuring the platform. Preparing and loading product data. Setting up basket, checkout, payments, delivery rules and tax handling. Connecting whatever has to connect. Populating content. Testing properly. Launching.
What usually falls outside it. Ongoing search and marketing work, which is normally bought separately. Photography. Writing product descriptions, which almost always returns to the client. Trading decisions such as pricing and range. Hosting, on a hosted platform, since the platform does it.
Those boundaries move considerably between suppliers, which is why block two exists and why comparing headline prices tells you so little.
There is a larger omission running through all of it. Everything above describes a project with an end date, while the thing you are actually acquiring keeps running afterwards. Who does all of this in year two is the question nobody raises at this stage. It is the one this guide keeps returning to.
What A Service Should Include
Eleven items, each with the reason it matters. Read it against a quote and the gaps become visible immediately.
Discovery and structure planning. Because it sets the ceiling on everything the shop can reach afterwards.
Design of every page type. Not a home page and one product page, with collection pages, search results and error states assumed.
Build and platform configuration. The part everybody includes.
Product data preparation and loading. The largest hidden cost in most projects, also the item most often excluded without being flagged.
Basket, checkout, payments, delivery rules and tax. Delivery rules in particular are more involved than they sound once bulky items or overseas orders appear.
Integrations, priced individually. Each behaves differently, so each should be quoted separately rather than bundled.
Content population. Somebody has to put the words in. It is worth knowing which somebody.
Testing, including a genuine order and a genuine refund. On real devices rather than a resized browser.
Launch. Redirects where a site is being replaced, tracking confirmed working, sitemaps submitted and the construction block removed.
Training. So somebody in your business can add a product without telephoning anybody.
A support period stated in weeks. Not described as ongoing help.
Anything absent from a proposal has not been removed from the project. It has been moved to you.
The Questions To Ask
Six questions specific to commissioning a shop. General agency vetting, meaning references, stability, team size and working style, belongs in our guides on choosing an agency and is not repeated here.
Who owns the finished site and the code? The answer should be one word. Any elaboration is worth following carefully.
What happens if we part company? What you leave with, whether the shop continues working, how a handover happens. A supplier who has done this before describes a process. One who has not describes a feeling.
Who holds the accounts and the domain? Domain, hosting, platform, analytics and payment accounts, in your name from the start rather than transferred later as a courtesy.
What is included after launch, stated for how long? Weeks, not reassurance. Then what it costs once that period ends.
Is search structure in scope or excluded? Category structure, address patterns and filter handling specifically. This is the item most commonly missing, for the reasons in how to build an ecommerce website with SEO built in.
Who prepares the product data? Answered precisely rather than with client to supply products, which is a sentence that has absorbed a great many unbudgeted weeks.
Five of those six are about what happens after the invoice is paid. That is not an accident of ordering.
Warning Signs
Six, in order of expense rather than in order of how alarming they sound. None proves bad faith. Each is a reason to ask one more question.
Ownership retained by the agency. The most expensive by a wide margin, because it can cost you the entire shop. Arrangements where the supplier holds the site, the code or the accounts and you effectively rent access are legitimate when chosen deliberately. They are ruinous when discovered during a disagreement.
No discovery before a quote. Second, because it sets a permanent ceiling. A price produced without anybody asking about your catalogue is a price for a structure nobody has thought about. Structure is the one thing a rebuild is needed to change.
No written specification. Third, because every subsequent disagreement resolves against whoever cannot point at a document. That is usually you.
A fixed price with no scope attached. Fourth. The price is not fixed. It is a deposit. The remainder arrives as variations once the work reveals what was never defined.
A portfolio where every site is the same template. Fifth. You will receive that template, whether or not it suits your catalogue, because it is what the process produces.
Guaranteed search positions. Sixth in direct cost and first in what it tells you. Nobody controls the systems producing those results, so the guarantee cannot be honoured. Its real cost is that it puts every other claim they made in question.
Judging A Portfolio Properly
A portfolio is the main thing buyers use and among the least informative things available, unless it is interrogated rather than admired. Four checks, none taking longer than a few minutes.
Are those shops still trading? Visit each one. A portfolio containing businesses that have closed is not disqualifying, though it is worth knowing before somebody presents it as evidence of results.
Are they still on that build? Compare the live site with the portfolio image. A portfolio where several entries have since been rebuilt by somebody else is telling you something. It is worth asking about rather than assuming the worst.
How do they perform on a phone? Open one on your own phone and try to buy something, stopping at the payment step. Four minutes here teaches you more than an hour of presentation.
Can they say what the project achieved commercially? This is the question that separates suppliers. It is the one to lean on.
An agency genuinely involved beyond delivery can tell you something concrete. That the client migrated several thousand products without losing visibility. That order volumes moved. That the support enquiries about delivery stopped. They will not always have the figures to hand and they will know what the project was for.
An agency that can only describe how it looks was a supplier of pictures. That is a real service and not the same service. The difference matters most in the two years the portfolio does not cover.
What Happens After Launch
This is the block the whole page has been walking towards. Buyers raise it at the end of the conversation, once the interesting decisions are made. It describes the arrangement they will live inside for years.
Support. What is covered, for how long, at what cost once the included period ends. Stated in weeks and in figures rather than in willingness.
Updates. Platforms change, apps change, security fixes appear. Somebody applies them or nobody does. Establish which.
Hosting. Who provides it, whose account it sits in, what happens to it if the relationship ends.
Then the question that matters more than the other three together. It is worth asking in exactly these words.
Who fixes it at nine o'clock on a Saturday evening when the checkout stops taking payments?
Not whether somebody is available. Who, by name. Through what route. Within what time. At what cost at that hour. Whether there is anybody else if that person is unreachable.
Every shop has that Saturday eventually. It is the moment the relationship is actually tested. It is entirely predictable. It is almost never discussed while a proposal is being written. A supplier who answers it precisely has thought about being your supplier rather than about winning your project, which is the distinction this entire page is built on.
Pricing Models
Four arrangements are common. Each one pays a supplier to behave in a particular way, so the useful question is not which is cheapest. It is which behaviour you are funding.
Fixed project fee. You know the number in advance. It also pays the supplier to define scope tightly and to resist additions, since every change erodes the margin. That is the model working as designed rather than bad faith. It suits well defined projects and produces friction wherever requirements are still moving.
Phased. Split into stages, each priced and approved before the next begins. It reduces risk on both sides and is the sensible answer when scope genuinely cannot be settled at the outset. It costs more in administration and in decision points.
Retainer. A monthly amount for continuing work. It suits running a shop rather than building one. The risk is scope quietly thinning while the fee holds steady, which is why the work needs defining rather than describing.
Build plus ongoing. A project fee to make it, then a monthly arrangement to run it. This is the only one of the four whose shape matches how a shop actually behaves, since a shop does not stop needing attention on launch day.
We work on the last of those. Build work is quoted per project. Everything afterwards sits inside a fixed monthly fee rather than being billed by the hour, which we do because hourly billing after launch discourages clients from reporting small problems until they become large ones.
Comparing Two Proposals
Two ecommerce proposals are almost never describing the same project, which makes comparing their headline figures an exercise in comparing two different things and calling one cheaper. Five steps make them comparable.
Put the two exclusion lists side by side. Not the inclusions, which always look similar because everybody includes design and build. The exclusions are where the projects genuinely differ.
Price whatever one excludes and the other includes. If proposal A leaves product data with you and proposal B does not, establish what that work costs, either in money or in weeks of somebody's time.
Add the running costs each implies. Platform subscription, transaction charges, apps, hosting and maintenance. A build using several apps carries a monthly total that continues indefinitely.
Add what happens when each support period ends. One proposal including twelve weeks and another including four is a difference with a price attached.
Then compare over three years rather than at launch. That horizon is roughly how long a shop runs before needing serious attention. It is the only comparison including the relationship rather than only the project.
Done properly this regularly reverses the apparent order. A proposal excluding product data, content and post launch support is not cheaper. It is smaller. The difference has been moved onto you. The full method sits in how much ecommerce web design costs in the UK.
Ask us about
the Saturday.
We will tell you who answers, by what route and how quickly, before you ask us anything about design. Accounts go in your name from the start. Everything after launch sits inside a fixed monthly fee rather than an hourly rate.
What a build covers:
Ongoing management afterwards sits inside our fixed monthly fee. Build work is quoted per project.
Twenty guides.
One subject.
This page covers choosing a supplier. The rest of the series covers what a build costs, how long it takes, which platform suits which business and why the search ceiling is set during the build.
