How to build an e-commerce portal that customers love

In 2020, PepsiCo brought a pair of direct-to-consumer websites from ideation to launch in just 30 days in an attempt to capture market share as customers moved online during the pandemic. This era saw consumer product companies rush to put up an e-commerce front, but three years on it’s clear that many are falling short of customer expectations. Outdated, fragmented, and poorly integrated portals frustrate the consumer. How did so many companies get it wrong?

The vision of e-commerce infrastructure (see figure 1) as merely a digital “portal” for customers misses a crucial piece of the undertaking: e-commerce platforms aren’t just an interface, they comprise the full consumer experience, driving offsite brand interactions, consumer awareness, and both online and instore sales. That misconception means that most backend design efforts start by collecting business requirements from business users rather than the consumers who actually use the platforms.

In this, the final piece in our series on e-commerce, we will explore the common reasons for failure, and the questions companies need to ask themselves as they stand up an online store, such as whether to buy or build the platform, or look at some combination of the two.

Using the approach we outline, we’ve been able to help our clients launch successful e-commerce platforms within six months. But first, let’s look at what many companies miss in the rush to the digital market.

Reasons for failure

1. Lack of an intuitive interface

Users adapt as technology and design approaches progress, so older portals quickly become dated. One AlixPartners client had built their customer portal in the early 2000s, and even they couldn’t recall how to perform certain functions—how would the customer?

Solution: To design a new user interface, we did repeated usability testing with actual customers to detect where users paused or selected the wrong path. Contrast with Amazon which is so intuitive that it doesn’t require user training (the client did in fact produce a library of training videos to facilitate the user, enhancing opportunities for the user to find and discover). Prototyping tools like Figma allow you to review platform requirements with customers and collect their feedback. Then, you can remove the obstacles between their shopping cart and the checkout.

2. Poor product information

Working with a client in the grocery wholesaler business, we found many customers reported a “lack” of product information—and they weren’t wrong. Through a review of their top-selling products, we found that only half had product images and none had ingredients or product descriptions. These gaps in product information don’t only hurt the user experience, they impact the SEO performance of the site. Customers were googling the product to get correct descriptions, since the e-commerce portal contained one-line cryptic descriptions better suited for invoicing rather than marketing.

Solution: Many e-commerce implementations focus narrowly on sales technology (e.g., Shopify, SalesForce Commerce, Magento), but a successful digital shopfront needs four major components: the commerce platform itself, as well as the customer relationship management (CRM), product information management (PIM) and data asset management (DAM) systems. For example, generative AI can be incorporated into PIM to generate marketing descriptions better suited for everyday users. Whatever the tools, the implementation effort must deliver a comprehensive solution that includes all four components.

3. Relevancy

A UK consumer research company called Which? surveyed customers in 2022 about product substitutions in online grocery orders, finding that major outlets often did a poor job of suggesting relevant or likely substitutions: one customer had a box of breaded fish fillets substituted for a pint of Ben and Jerry's Phish Food ice cream, and another had a citrus household cleaner substituted with orange soda. Even best-in-class consumer shopping sites perform poorly on search, ad placement, and substitute recommendations. Which means there is a huge opportunity to convert lost transactions into revenue.

Solution: There are commercially available products that can help you get started (for example, Algolia, Adobe Target, Amazon Personalize, and Salesforce Einstein), but no organization will get It perfect immediately. The key is to make sure your site has appropriate analytics embedded to understand customer behavior – not just at the high level (pages visited, time on site) but in terms of how they navigate the site and at what point they abandon the sale. That information can be used to drive improvements.

4. Internal silos

For all intents and purposes, internal organizational dividers are irrelevant to the external consumer. For example, within one company that partnered with AlixPartners, the service parts organization was totally separate from the original equipment sales organization, each reporting to different senior executives. That would be fine, except that each had its own service portal and design language, and customers—dealerships—had to interact with both. This not only guaranteed a poor user experience, it was a missed opportunity for the service parts portal to take advantage of the information the sales organization had on customers’ purchases with an integrated customer relations management (CRM) system.

Solution: A team dedicated to optimizing the end-to-end experience, with authority in the decision-making process, can prevent this failure by leveraging the CRM and looking at where the consumer experience overlaps different parts of a company.

Building—or buying—a platform that works

Existing e-commerce software like Shopify and Squarespace allow companies to get to market fast, and upgrade site functionality relatively often. They also often require a portion of revenue to be fed to the vendor.

The alternative to buying access to an existing platform is to build one from scratch, which may work in situations where consumer needs are fairly static and predictable. There are pros and cons for each (see Figure 2), as well as the potential to run with a hybrid model of build and buy.

In fact the right tech stack is almost always a combination of build/buy once you add in elements like the DAM, PIM, and warehouse management system.

Ideally, you want the flexibility to build custom enhancements tailored to your specific business, with the potential to set it apart from competitors—think AI models that give customer-specific product recommendations.

Once you know what you want the customer experience to be, you have time to finesse it. Amazon took over 20 years to finetune its blockbuster e-commerce platform, and it is still evolving. You won’t typically be able to launch all possible features and capabilities on day one, but with knowledge into the how and why of your tech stack, you’ll be able to build something powerful over time.