8 recommendations before starting your eCommerce project

Before starting an eCommerce project, a company should settle eight things in this order: the processes the store will automate, the platform that will run it, the buyer it is built for, the usability of the shopping path, the conversion routes, cross-browser testing, search, and a consistent design system. Decisions made here determine how much revenue the store can capture and how much it costs to operate.

Picture the monthly executive meeting of a consumer goods manufacturer that launched its online store nine months ago. The CEO asks why visits are up while revenue per visit is flat. The CFO points out that the operations team still re-keys web orders into the ERP by hand. The commercial director admits that nobody can say which customers buy online and which ones only browse. The store exists, it is attractive, and it was delivered on time. Yet no one in the room can say what it is supposed to contribute to the business. That scene is common, and it almost always traces back to decisions that were never made before development began.

An eCommerce project is not a website project with a shopping cart attached. It is the digitization of a commercial process, and the quality of the outcome depends on how well that process, the platform, the buyer and the experience were defined before the first screen was designed. The eight recommendations below come from ICX Consulting's work with companies that sell online, and they are ordered the way a steering committee should take the decisions.

Why do eCommerce projects fail before development starts?

The first recommendation is the least glamorous and the most decisive. Software automates processes, so a company that is unclear about its own process will automate the disorder. If the quoting rules live in one salesperson's head, if returns are handled by whoever happens to answer the phone, if three departments keep three different price lists, the new store will faithfully reproduce all of it, only faster and in front of the customer.

Consider an industrial distributor that decides to open a B2B portal for its repeat buyers. In the workshops, it turns out that credit approval depends on an email exchange between sales and finance, that delivery dates are promised by the warehouse supervisor from memory, and that discounts are negotiated order by order. None of this can be coded into a catalog and a checkout. The distributor has two options. It can launch a store that only displays products and sends every real transaction back to email, or it can spend a few weeks defining the credit, availability and pricing rules that the platform will then enforce. The second option delays the launch and produces a store that actually sells.

ICX recommends taking the time to define and clean the processes that will be automated before any design or programming begins. A business process that is written down, owned and measured becomes the best possible input for the platform configuration, because it tells the developers what the system must do and tells the executives what a good outcome looks like. For the commercial flow, the SIPOC diagram is a useful instrument. It forces the team to name the suppliers, inputs, process steps, outputs and customers of the order cycle on a single page, and it exposes the handoffs where orders usually stall. The SIPOC consulting service from ICX applies this exercise to the order-to-cash flow of the store.

How do you choose an eCommerce platform your own team can run?

Choosing the platform is the second decision, and it should follow the process work rather than precede it. The common mistake is to select a platform because a competitor uses it or because a vendor demonstration was impressive. The question that matters is whether the marketing and commerce teams can manage design and content on their own, without opening a ticket every time a banner, a coupon or a landing page has to change. Many platforms make that difficult, and the cost appears months later as a queue of small requests to developers.

At a minimum, the platform should allow integration with third parties, simple creation of coupons, landing pages and forms, and self-management by the marketing team. Beyond that minimum, the advanced considerations are whether it can handle B2B and B2C processes on the same base, and how easily it connects to invoicing, ERP, inventory, databases and the CRM. A store that cannot exchange data with those systems becomes an island, and islands are expensive to maintain.

The product landscape has also changed since many of these projects were first conceived. Magento is part of Adobe's portfolio, and the commercial edition is now called Adobe Commerce. According to Adobe's own documentation, it can be deployed as a fully managed platform-as-a-service on cloud infrastructure, as a software-as-a-service offering, or installed on premises, while the open-source line continues as Magento Open Source. ICX's practical guidance has been consistent. A smaller company running a B2C operation with moderate integration needs can start from the open-source edition. A mid-size or large company running B2B or demanding B2C operations, with heavier requirements for integration, security and vendor support, should evaluate the commercial edition and its cloud options. The right answer depends on the process requirements defined in the first step, which is why the order of the decisions matters.

Selecting a platform is also selecting a long-term relationship with a technology and an implementation partner. The Technology Selection Toolkit offers an evaluation matrix for comparing candidate platforms against the requirements that the business has defined, which keeps the conversation anchored in criteria rather than preferences.

Who is the buyer when there is no salesperson in the room?

In a physical store, an experienced salesperson reads the customer, asks a question, adjusts the offer and builds trust in a few minutes. On a website there is nobody to do that. The design has to do it, which means the company must know whom it is designing for. This is the role of the buyer persona.

A buyer persona is more than a demographic summary. It is a detailed portrait of each type of buyer, including tastes, preferences, interests, the metrics that matter in their work, and what they expect from the shopping experience. Defining it properly lets the company decide which communication channels to use, which products to emphasize for which profile, and how to allocate budget so that it is not spread evenly across audiences that behave very differently. It is also the first step in building trust, because a site that visibly understands its audience feels reliable in a way that a generic site does not.

Imagine a regional pharmacy chain building an online store. The occasional buyer who needs an over-the-counter product wants speed and a clear price. The caregiver who reorders medication for a relative every month wants reminders, a saved cart and a delivery window. The corporate buyer who supplies first-aid cabinets to offices wants invoices, approval flows and negotiated prices. The three journeys share a catalog but not much else, and a store designed for an average of them serves none well. ICX's buyer persona development work exists to make these differences explicit before they are expensive to retrofit.

Does usability decide the sale?

Usability is where the work on process and persona becomes visible to the customer. Making navigation easy and intuitive is a discipline, not a matter of taste. Products belong in categories and subcategories that make sense to the buyer rather than to the company's internal org chart. They should sit on a clean grid with clear labeling that is easy to read and scan. Methods from user experience design, especially testing with real users and collecting their observations, allow the team to refine the path to purchase before launch rather than guess at it.

The evidence on what goes wrong in the path to purchase is consistent. The Baymard Institute, which has studied checkout behavior for years, documents an average cart abandonment rate of 70.22%, calculated across 50 studies, and its most recent update lists extra costs such as shipping, tax and fees as the leading reason after shoppers who were simply browsing, followed by slow delivery, distrust of the site with card information and forced account creation. None of those reasons is about the product. They are about how the experience handles money, time and confidence, which is exactly what a professional design and a thoughtful checkout address.

That is why a professional design that gives the user confidence should not be treated as a cosmetic line item. A shopper who cannot tell whether the site is trustworthy, or who is surprised by a charge at the last step, leaves without complaining, and the company never learns why. The cost shows up only as a conversion rate that nobody can explain.

The eighth recommendation, consistency, belongs to the same discussion. Less is more. Avoid more categories than necessary in the navigation, use white space to let content stand out, and build on a solid grid. The header, footer and product grid should look and behave the same on every page. A color palette should be defined and respected, the tone of photography and copy should follow the brand identity, and the calls to action should have a fixed color, size and style across the site. Consistency reduces the effort a visitor spends interpreting the interface and leaves more attention for the decision to buy.

What conversion paths does an online store need?

Users are not afraid of clicking a link or a call to action, provided the path makes sense and leads where they expect. A well-built call to action guides the visitor through organized content toward a destination in a natural progression. A cluttered page with no obvious next step does the opposite. Because conversions rarely happen on the home page, the store needs deliberate routes into the sales funnel from every page a visitor might land on.

A simple and clear main navigation, visible calls to action and well-chosen links to popular content are the first routes. A clearly placed site search is another, and it deserves its own attention.

The conversion paths also generate data, and that data needs somewhere to go. Managing the information an online store produces by hand, in spreadsheets, or from orders that arrive by email is not viable beyond a trivial volume. A well-chosen CRM, integrated with the store, is what turns a path of clicks into a record of who the customer is, what they looked at and what they should be offered next. ICX's article on the four factors that improve sales conversion in an eCommerce develops this logic further, and the piece on product pages that boost conversions looks at the page where most buying decisions are finally made.

Why does on-site search deserve its own budget?

ICX's experience is that visitors who use the search bar on a site tend to be those with the strongest purchase intent, since they arrive knowing what they want. For that reason the search field should be prominent, properly designed and able to retrieve every product and piece of content on the platform. A search box that returns nothing for a misspelled word is the digital equivalent of a store clerk who turns away from a customer who asks a question.

The industry data suggests that many stores fall short here. Baymard's 2026 benchmark of ecommerce search finds that 56% of the sites it evaluated deliver a mediocre or worse search experience. The same research also shows nuance that executives should respect. Search is not equally important for every business. Niche direct-to-consumer sites with small catalogs may need it far less than a distributor with thousands of references, and apparel retailers often do better investing first in category navigation and curated paths. The decision about how much to invest in search should therefore follow the catalog size and the way customers actually shop, which brings the discussion back to the persona and the process.

How should you test an eCommerce site before launch?

The sixth recommendation concerns the range of devices and browsers on which the store has to work. Some customers will arrive on older computers and browsers, and that does not mean the store should be designed around them. Equally, it should not be designed only for those with the latest equipment. The goal of any eCommerce site is to increase revenue and conversions, so the design has to be flexible enough to present its content clearly on most devices without a large amount of extra development.

The basic tests are done in Google Chrome, Apple Safari and Mozilla Firefox, and on desktop, tablet and mobile phone. To that baseline, a modern project should add performance. Google's Core Web Vitals define what counts as a good experience, assessed at the 75th percentile of page views, with the main content loading in 2.5 seconds or less, interactions responding within 200 milliseconds and visual layout shifts staying at or below 0.1. A store that passes functional tests but fails these thresholds on a mid-range phone is a store that loses customers in the first seconds, long before any persuasion can happen.

How should the project be phased?

A final practical point concerns the way the project is delivered. Large eCommerce initiatives that promise everything at a single launch date tend to slip, and they delay the moment at which the company learns from real customers. ICX recommends methods such as Growth Driven Design, which delivers the project in short phases while still allowing the company to go to market quickly. The first release carries the essential path, from catalog to payment and order management, and later iterations are prioritized by what the data shows about visitor behavior.

The advantage is financial as much as technical. Each phase has a smaller budget, a clearer hypothesis and a measurable result. If a feature does not move revenue or reduce cost to serve, the company finds out early and redirects the investment. This also protects the relationship between the executive sponsor and the project team, since progress is visible every few weeks rather than promised for a distant date.

What this means for the C-level

For the CEO, the central point is that an online store is a commercial channel with its own economics, not an IT deliverable. The questions to ask before approving the budget are what share of revenue the channel is expected to carry, at what cost to serve, and how it will coexist with distributors, branches or field sales without creating conflict. If those questions have no answers, the project is not ready to start.

For the CFO and the operations leader, the most valuable early work is process definition and integration. Every manual step between the store and the ERP is a recurring cost and a source of error, and the integration design determines how much of the efficiency promised in the business case is realized. For the commercial and marketing leaders, the persona work and the conversion paths determine whether the store converts the traffic that marketing pays for, and whether the data it captures strengthens retention and loyalty through better follow-up. The store becomes a customer data source that feeds the CRM, and what the company does with that data over the following years is where most of the return is generated.

Where to begin on Monday morning

A seasoned consultant would start with a half-day session that brings together commerce, sales, finance, operations and IT around one concrete question, which is what exactly happens today between the moment a customer decides to buy and the moment the cash is collected. Draw that flow on a wall, mark the points where people improvise, and note the systems that touch it. That single picture usually settles several of the decisions in this article at once, because it reveals what the platform must support and where integration will be hardest.

From there, the next two weeks should go to the buyer. Interview a handful of real customers from each segment, and ask the sales team which objections they hear most. The persona that emerges is the brief for the designers. Only then does it make sense to compare platforms against a short list of criteria, to agree on a first release that can go live in weeks and to schedule the cross-browser and performance tests before the date is announced. A project organized this way spends the first month on definition and the following months on delivery, which is the reverse of what most troubled projects do.

How ICX helps

ICX Consulting approaches eCommerce as a commercial and operational transformation first and a technology implementation second. Through its B2C eCommerce consulting and B2B eCommerce consulting practices, ICX helps leadership teams define the processes, the buyer and the platform requirements before building, then guides implementation and the integration with CRM and ERP. The work draws on ICX's business process and user experience practices so that the store is designed around how customers buy and how the company operates.

If you want a structured way to compare platforms and partners, use the ICX Technology Selection Toolkit, which includes an assessment checklist, an evaluation matrix and a process diagram. Is your project at the point where these decisions are still open? Request a diagnostic with an ICX consultant and review them before the budget is committed.

Frequently asked questions

How long should the planning phase of an eCommerce project last?

For most mid-size companies, four to six weeks of structured definition is a reasonable range, covering the order-to-cash process, the buyer segments, platform criteria and the scope of a first release. That is an ICX planning guideline rather than a fixed rule. Larger organizations with several business units or legacy systems may need longer, but the phase should end with decisions, not documents.

Should the CRM be integrated from day one?

Yes, at least for the core customer and order data. Retrofitting the integration after launch means migrating records, reconciling duplicates and rebuilding automations, which costs more than designing the data flow at the start. Advanced features such as scoring or lifecycle campaigns can wait for a later phase once the foundation is stable.

Is the open-source edition enough for a B2B store?

It depends on the integration, security and support requirements. A B2B store with negotiated prices, approval flows and ERP connections usually benefits from the commercial edition, now called Adobe Commerce, or from a comparable enterprise platform. The decision should be made against documented requirements and total cost of ownership rather than license price alone.

What should be measured in the first ninety days?

Conversion rate by segment, abandonment at each checkout step, search success, page performance on mobile and the share of orders that flow to the ERP without manual intervention. The last measure is often neglected, yet it is the clearest indicator of whether the process work paid off.

An online store earns its place in the business when the decisions behind it were made on purpose, from process to platform to buyer. If your team is weighing an eCommerce launch or trying to rescue one that is underperforming, Talk to an ICX consultant about what the store should contribute to revenue and margin before the next phase is funded.