You will be redirected back to your article in seconds
Skip to main content

Deep Dive: Experts Debate When to Build or Buy Retail Software

To build, or not to build—that is the question some retailers are grappling with today.

In a sense, retail is often synonymous with “legacy systems,” a reference to the reality that for decades major organizations in the industry built their own heavily customized tech stacks in house. But the data age and the era of the cloud have changed that, ushering in the software-as-a-service model that gives retailers significantly more flexibility in deploying purpose-built systems without doing the heavy IT lifting themselves—freeing them to do what they do best: sell.

Though it’s become the de facto norm for many retailers to hand their tech needs over to software vendors, for some the IT pendulum is swinging back in the other direction. Dick’s Sporting Goods recently decided to go against the grain by building some of its critical software systems in-house to gain an edge over competitors.

So what, exactly, are the pros and cons of building software in house versus deploying commercial systems?

Some schools of thought believe choosing to build in-house is a rejection of cloud-based solutions, said RSR Research managing partner Brian Kilcourse, who previously spent nine years as CIO of pharmacy chain Longs Drugs. But the truth is that RSR’s data shows that once retailers reach a certain size in sales and turnover, they’re more inclined to tap their IT teams to develop code proprietary to their organization.

“And the reason they’re doing that is they believe there are things about their systems that are unique and competitive,” Kilcourse explained.

Darin Archer, CMO of microservices software solutions provider Elastic Path, agreed that size is an important factor in the question of whether to build or buy, but he also described both business model and business case as influencers, as evidenced by TechStyle Fashion Group’s decision to create an operating system tailored to its needs.

Related Stories

“Because we pioneered a new membership model, there was no platform that could sufficiently meet the needs of our employees and our customer base,” said Tim Collins, CTO of TechStyle, which operates the Fabletics, JustFab, FabKids, ShoeDazzle and Savage X Fenty brands.

Kilcourse points to changes in the DevOps environment as a factor in why retailers are more comfortable outsourcing many of their software and systems. In the past, commercial off-the-shelf systems served as a “framework for customization” subject to what he described as the “multiplier factor,” meaning a $1-million purchase price, for example, really mushroomed into a $5-million investment once the retailer finished modifying the software to its specific needs.

Now, rather than a purely either/or decision, many retailers foster a hybrid IT environment that mixes and matches old and new technology.

“They have serious investments and legacy applications that are not crap,” Kilcourse said. “And many of them are either custom-built systems or they are commercial systems that have been modified heavily over the years.”

It’s not easy to walk away from systems that, while older and perhaps not as user-friendly as modern technology, often serve as the fundamental backbone of the retail organization. However, many retailers lack the IT resources—both budget- and talent-wise—to execute on the demands of today’s consumer.

That’s why Aptos VP of retail innovation Nikki Baird says it’s the aggressive pace at which systems must be updated that prompts many merchants to hand over the IT reins to software vendors.

“Retailers have to be committed to an innovation cycle that keeps up with what software companies offer,” she said. “Most retailers are still in a pattern of upgrading every two to three years, usually because it’s a big lift due to integration or customizations they’ve made.”

Total cost of ownership

There’s also the question of total cost of ownership, said Luke Tuttle, VP of international product for Klarna, a payments solutions provider.

“In today’s world, time to market and the ability to iterate rapidly are the real differentiators of success,” he said. “Often times a purchased solution can be deployed in days, while a build will take months, if not longer. Given the rapid innovation and quality of many of the solutions on offer, the question becomes: can you afford the time it takes to build?”

Experts agree that taking software in-house is a sensible approach if the market doesn’t offer what you really need. If you “have to give up too much to get what you want out of the market,” Baird said, then it’s better to chart your own software course. She points to clothing rental as an example; whereas a decade ago companies that wanted to offer apparel subscription boxes would have had to build the technology themselves, today the market has expanded to the point where vendors like CaaStle now offer turnkey rental and subscription software platforms.

Systems that are a differentiator in a retailer’s business also are ripe for in-house development.

“If your customers are willing to drive through traffic to experience your brand, then you have something that’s differentiating. For everything else, you should try to find a commercial solution,” Kilcourse said.

Retailers in need of “very specialized” features that aren’t commercially available would be wise to leverage their IT staff to build the system themselves, said David Dorf, VP of product strategy for retail at business software vendor Infor.

“‘Build’ is an ongoing commitment to maintain the software as long as it is in use, so it should not be taken lightly,” he explained. “The long-term costs, including startup and ongoing operating costs, should be considered to determine the return on investment (ROI). This normally only makes sense when the functionality is unavailable on the market and is very valuable to the business.”

Baird agreed that those long-term costs should factor into the decision to build or partner with a vendor’s existing solution.

“Offering packaged solutions requires a lot more than writing code you plan on being the only one to use, but a lot of the things that go into packaged solutions are things retailers can sometimes take for granted,” she said, noting that they often overlook deployment tools to ease the transition plus data maintenance tools and documentation, or “the stuff that explains how the software works and how to deploy it.”

One retailer Aptos is talking to operates a custom promotions engine written more than 10 years ago by someone who has since left the company, Baird said. “Now no one knows how it works or why it works because there is no documentation, and yet this is a central piece of the company’s strategy—to be strategic with the promotions it offers consumers,” she said. This kind of “brain drain” can hinder retailers from innovating and developing new, differentiating capabilities.

TechStyle’s in-house build

For its part, TechStyle has enjoyed the advantages of a tech stack designed not for mass appeal but to serve a young, emerging company with a non-standard business model. Collins said the company was “determined to provide our workforce with a system that would enable seamless collaboration among all groups.”

With FashionOS, Collins said the art of fashion and the science of technology “thrive together in tandem, resulting in a more efficient approach to launching and scaling fashion and apparel brands.”

The fashion company’s nearly 2,000 global employees have all had a hand in developing FashionOS’s six systems applications, to some degree.

“At this point fashion designers working alongside data scientists—something unheard of at most fashion companies—is status quo to us,” Collins said.

Though retail tech vets weigh and measure the benefits and drawbacks of taking software development in house, “the efficiency advantages of building over buying are not to be discounted,” Collins added.

TechStyle wanted to avoid a lengthy and confusing onboarding process Collins described as “a common pain point of third-party software systems.”

“We developed FashionOS with our employees in mind, so learning new applications is a much more seamless process,” he noted.

Take TechStyle’s warehouse management systems (WMS), for example, built on the iOS operating system already used by many facility staffers on their personal smartphones—thereby limiting the training time they needed to get up and running on a new system. The California-based fashion group claims to be the first e-commerce company to build an in-house WMS on iOS “and the ROI has been incredible,” Collins said. TechStyle credits the proprietary WMS with slashing shipping costs by 400 percent.

When TechStyle began opening Fabletics stores, it needed a backend system that could seamlessly tie in-store activity to online behavior—capabilities not readily available at the time in off-the-shelf software. Patent-pending OmniShop, an integral component of its OmniSuite software, personalizes the customer shopping journey with custom greetings in fitting rooms where shoppers are shown tips on digital displays demonstrating how to style the clothing they’re trying on.

Customers can request assistance via a dressing room iPad and if they decide to hold off on purchasing, OmniShop stores the items they tried on—as well as any feedback about those products—so they can buy online later when they’re ready.

Still more granular features help TechStyle avoid some common pitfalls in fashion retail.

“Internally, the dressing room scanning allows TechStyle to track inventory movement and activity in its brick-and-mortar stores down to the smallest level, including SKU, size, color or even individual item,” Collins said.  “By leveraging this framework, Fabletics can drill down into the manufacturing and design details of each individual product to determine why specific products are selling or not selling and react in real time.”

Collins credits this proprietary technology with reducing the “traditional retail feedback cycle to hours, hugely optimizing business efficiency.”

At the moment, much of TechStyle’s energy is focused on developing Albert, the newest part of FashionOS comprising data science systems to drive machine learning and automate decision-making across business units.

“One aspect of what Albert is enabling is dynamic merchandising, where customers are presented with a personalized shopping experience tailored to their preferences based on their online behavior,” Collins said. “Given artificial intelligence (AI) is only as good as the data you provide it, the journey to this sort of individualized customer experience has been a long one as we’ve collected and made sense of a vast amount of information.”

The innovation imperative

Regardless of the path they choose, merchants must keep an eye on what really matters today: customer-centric innovation.

“Every retailer should prioritize the ability to innovate whether the platform components are built internally or bought,” said Tuttle, the Klarna VP. “In most cases today, ‘bought’ means that there is a subscription, and the ability to rapidly integrate, test and swap out components based on price, performance and innovation is critical.”

Merchants shouldn’t make the mistake of ceding control of their entire platform, Tuttle continued, but focus on areas that offer a competitive advantage. He pointed to last-mile delivery opportunities as an example. A retailer might want to sell inventory fulfilled in near-real time by couriers for ridesharing companies like Uber or Lyft.

“That innovation might require a specialized logistic module and integration because it doesn’t exist today,” Tuttle noted. “However, it should be built in a way such that once that innovation becomes commonplace, it can be swapped out for another ‘buy’ solution that might provide better and improved features because of investment from many retailers interested in that [same technology].”

There are certain things best managed in house versus left to third parties, Baird said. Because core platforms like ERP and supply chain planning are used by external parties like suppliers and factory partners, it’s “much better to be standardized,” she said, adding, “standards and ease of connectivity are critical there.”

Dorf agreed that “stable, mature processes” like planning, merchandising, e-commerce, WMS and point-of-sale are best served by solutions on the market.

“Most of the building blocks of retail are already available from competing vendors, so there are lots of choices available,” he said.

Instead of fooling around with business-critical systems, merchants are often lured in the opposite direction. There’s so much open-source content and capabilities in the AI world, Baird noted, that some retailers are tempted to build AI tools themselves and run into perhaps the biggest problem in retail IT today: skillsets.

“We’ve heard from retailers who are experiencing 200 percent to 300 percent turnover in their AI developers, because the market is so competitive they can’t keep them for long,” she said.

Though in-house IT teams have become more sophisticated, large-chain retailers still get “hung up” in deploying new technology across sprawling fleets.

“As much as retailers want to get to the speed of digital, they’re always going to be governed by the slowest link in the chain, and that’s the store,” Baird said.