# SHOPIFY: Built to adapt

Published: 2026-11-22

Canonical URL: https://yaffamedia.shorthandstories.com/Shopify_BuiltToAdapt/

---

![](https://yaffamedia.shorthandstories.com/Shopify_BuiltToAdapt/assets/XCf9elcFN1/shopify-logo-white-400x200.png)

# **Built to adapt**

Shifting customer expectations and technology are pushing brands towards more agile commerce operations. Shopify’s [*Built to Adapt*](https://www.shopify.com/enterprise/webinar/built-to-adapt?ref=.p2PHnqH2D5DxoWJ05xIzug%2A%2A.tjU98ZERanytrV5WMPdNBWbeT-RB_Wqex1op0DhokBcsspLXKW0tP0eh8Yt8KUGZPKVOtieiaI2tpunHaEEK8R663EI2FwXyJ21j_utUUW4%2A&syclid=dbv6v8paag5c73e53fv0&utm_medium=email&utm_source=mozart) webinar brought together Forrester’s Emily Pfeiffer, Kathmandu’s Craig Mildenhall and Shopify’s Hannah Udina and Kole Mahan to examine what it takes to keep pace.

*This is part of a Ragtrader partner series*

## **Three problems, ranked**

**Fashion retailers have spent a decade building out the stack. A platform. Integrations. Extensions. Contracts to renew. Every piece solved a problem at the time. What has changed is the cost of holding it together.**

Forrester’s study of retail and wholesale digital business and strategy decision-makers asked what they intend to fix in customer-facing technology in the year ahead. Three answers came back ahead of everything else: too many systems in the ecosystem, technology too old to move at the pace of the business and internal tools staff will not use properly.

Consolidation ranked first. Forrester principal analyst Emily Pfeiffer said the finding marks the end of an expansion cycle, in which retailers added specialist platforms, modules and integrations one requirement at a time until the combined ecosystem cost more to run than it returned.

The expense is operational rather than licensing. Every additional system carries an integration to maintain, a contract to renew and a point at which something can fail. “Moving parts are things that break. There are things that need management. There are things that create risk,” Pfeiffer said.

The same study found nearly a third of respondents plan to add geographies to their existing commerce experiences, selling into regions where they do not currently operate, either by standing up a new site or running a multi-site structure within one platform. Retailers are therefore reducing system count and adding market complexity in the same budget cycle, which is the tension the consolidation finding sits inside.

## **The cost of legacy systems**

**Modernisation ranked second and the study put the scale of the problem at systems untouched for five or six years, with some retailers running technology that has been in place for 15 or 20.**

Pfeiffer identified two tests that separate technology that is merely dated from technology actively costing a business money. The first is data speed. If a retailer cannot send through inventory files for items sold across the course of the day and get the updated position live as often as it wants, the limitation reaches customers as inaccurate availability.

The second is the maintenance crossover, the point at which the expense of keeping a platform running exceeds the benefit of avoiding a replacement. It is rarely marked by a single failure. “If you’re keeping your solution on life support unnaturally, it might be time to let it go,” Pfeiffer said.

Retailers reaching that second test arrive at Forrester with a consistent set of questions. Its client inquiries fall into five areas: contract negotiation at renewal across commerce platforms, order management systems and search providers, removing or consolidating technology out of the ecosystem, improving a single high-level function such as inventory management, simplifying environments with too many integrated pieces to manage cost-effectively and adding a commerce search product to close a functional gap.

The timeframe attached to those conversations has shortened. Retailers now look for a return on a replacement in months rather than years and in some cases weeks.

The standard Pfeiffer applies to a replacement is her own FIRE framework, which judges commerce technology on whether it is flexible, inexpensive, rapid and easy. She described it as a perpetual goal rather than a destination, with flexibility becoming a priority in early 2020 when businesses found their ecosystems too stiff to adapt.

### “Moving parts are things that break. There are things that need management. There are things that create risk.”"

**Emily Pfeiffer**

![](https://yaffamedia.shorthandstories.com/Shopify_BuiltToAdapt/assets/A9dvYcuACE/emily-pfeiffer-headshot-enhanced-680x680.jpg)

![](https://yaffamedia.shorthandstories.com/Shopify_BuiltToAdapt/assets/18lGqVAK5m/pexels-ron-lach-9783831.jpg)

## **The hidden operating cost**

**A platform is costed on licences, implementation and support. The hours a team spends inside the interface it is given are not on that list and neither is the work that does not get done because of it.**

That's why the third finding is internal tooling. Pfeiffer said it is consistently underestimated because businesses treat the interface as a one-off training cost rather than a permanent operating one, on the argument that staff only have to learn a system once. The interface a vendor provides is the one a retailer is stuck with for as long as it runs the solution, turnover means new people to train against it and internal technology teams absorb the retraining and the support.

The benchmark has moved outside the category. Staff do not compare a commerce administration system with competing enterprise platforms, because most only ever use one. They compare it with the consumer applications and websites they use every day.

The cost lands in productivity, retention and execution. A task running to as few as five clicks can carry enough friction to stop the work being done at all.

“If it’s annoying, there’s a psychological impact. They don’t want to do it,” Pfeiffer said. “So if you’re starting to find users are reticent, that’s a great time to start evaluating a switch.”

## **Kathmandu's four-month fix**

**Most retailers in this position take a year to decide and another to deliver. Kathmandu set itself four months and chose to land the work in the middle of its biggest trading period.**

The outdoor retailer was running a Magento core built up over approximately a decade, with what general manager of digital Craig Mildenhall described as about a billion other things plugged into it.

No single failure triggered the decision to move. The clearest illustration of the cost was a request to change the website menu, which came back with the answer that security updates would block the work for the next couple of months.

“It just took too long to do anything,” Mildenhall said.

The migration ran four months. The work was built to land before the winter trading period that delivers Kathmandu its strongest growth and highest customer volumes, on the reasoning that the biggest return sits where the volume is. It set a deadline the team could not move, because missing the window meant waiting for the next cycle.

Mildenhall went through team one at a time, all 30 of them, to understand how they wanted to work and what they thought the business needed, before any technology decision was taken. He then took the commercial case to the wider business and parent company KMD Brands, covering cost implications, potential savings and customer engagement, with the decision requiring alignment across the group, its finance stakeholders and its partners.

“Change is always hard when it’s being done to you,” he said.

The work was staffed so trading continued through it. Kathmandu split into two squads, one on the migration and one on day-to-day trading and agency Collective worked alongside the team in New Zealand during the process. A bell in the office was rung at each milestone and is still in Mildenhall’s office now.

The site went live during a flash sale and customers kept buying. Kathmandu reported 26 per cent growth in trading revenue following launch.

Release cycles moved from monthly, in some cases two-monthly, to almost daily or weekly. Prototypes for peak trading are now built and tested in stores and in the office inside days and the team has been reorganised into squads built around a customer or business problem rather than a function.

Mildenhall’s advice to retailers sitting in the same position is to stop waiting for a clear window. “There’s always a thousand reasons to wait for something,” he said. “Waiting can hurt. Waiting can cause a whole bunch of issues further down the line.”

His second point is that the constraint is rarely the technology.

“You can have the greatest, shiniest Ferrari sitting in your great garage, but if you don’t know how to drive it and if you don’t know how to get everyone behind it, then it’s really hard to operate it,” he said. “I always tell my team, take the shot.”

Shopify ANZ head of partnerships Hannah Udina puts the cost of ageing commerce technology in three places, none of which appear as a line item.

“It shows up in how long a simple change takes,” she said. “It shows up in the people whose whole job it’s become to hold the current tech stack together, and it shows up in the trading moments where you couldn’t move fast enough to catch.”

### "I always tell my team, take the shot."

**Craig Mildenhall**

![](https://yaffamedia.shorthandstories.com/Shopify_BuiltToAdapt/assets/9diokhJ3NY/craig-mildenhall-headshot-enhanced-680x680.jpg)

![](https://yaffamedia.shorthandstories.com/Shopify_BuiltToAdapt/assets/ahNccPgl3k/pexels-ron-lach-9784240-4096x2731.jpg)

## **Can the stack keep pace**

**The next question facing retailers is what their stack will be asked to support next. Most of what the market is calling agentic commerce does not meet the definition.**

Pfeiffer said agentic means an AI agent taking an action a human did not explicitly instruct it to take. Most of what is running today is executing instructions it was given, which makes it assistive, and the dominant use is product discovery rather than transaction.

Where it happens is the distinction that matters commercially. The owned environment is a brand's own website and app. The non-owned environment is everything else: answer engines, marketplaces, search engines and social channels. Conversational product discovery is live in both.

Native checkout is coming and going from individual answer engines. Agents purchasing on a customer's behalf remain minimal, with Amazon's buy for me function letting a shopper send an agent to buy an item on another website, and personal AI agents have only become more mainstream in recent weeks. Agents transacting with each other is at the earliest stage.

"Anybody who says they've got it down is just guessing," Pfeiffer said.

## **Where fashion actually sits**

**The scale of it sits in the adoption data, which covers the shoppers already in those environments rather than the market as a whole.**

Forrester's research among US and UK adults who already use answer engines regularly found three in five research products or services in them and two in five discover products they did not know they wanted. One in five have completed a purchase.

Those figures describe existing regular users rather than the general population. The group skews younger, higher income and higher educated, with men statistically more likely to be regular users and it is not a measure of Australian retail adoption.

Pfeiffer scores the opportunity on five factors: customer, control, commodity, consideration and cost. For fashion, two of them matter.

Control is how much say a brand has over where its products are sold. A label selling only through its own stores and site has little to gain. A label sitting on a dozen stockists' sites at a dozen prices is one shoppers can compare, which is what answer engines are good at.

Consideration is how much research a purchase takes. Her example of a strong fit is consumer electronics, where a shopper works out which features they need before buying. Fashion is mostly not that, and she said a styling decision is the kind of purchase shoppers are least likely to hand over.

"What should I wear to this event? Very personal, very subjective," Pfeiffer said. "That's a shopping process that probably we won't offload as frequently to an agent."

Pfeiffer also expects autonomous buying to reach business purchasing before consumer purchasing. A business order carries fixed parameters a machine can act on, being a set quantity of a specified product by a date at or below a price.

For a retailer scoring low on the test, her advice is to wait six months and run it again as customers and costs change, or to test the channel on a subset of products, a sub-brand or a single market rather than the full catalogue.

On cost, paid placement has already begun in answer engines and Pfeiffer expects them to follow search engines into sponsored positioning. Her standing advice is a profit and loss sheet per channel rather than an aggregated view, so a retailer can see what each channel costs it and what it makes there. She also warned against taking existing product data and pushing it across unchanged, because the content has to answer how a product is used rather than list its attributes.

## **Choosing the next platform**

**A vendor selection starts with a list of functionality. What that list does not show is the two kinds of gap Pfeiffer tells clients to check.**

"I always recommend looking for the current gaps, like the things you know you have to improve today," she said. "But also the created gaps, like the stuff that's good today, but you're going to replace it, so you need to make sure you don't lose it."

She also recommends looking at what each vendor has released recently and speaking to reference customers with similar requirements, which she said do not have to be competitors.

The factor she says gets overlooked is whether the two organisations can actually work together. Clients have told her they knew a vendor was right as soon as they walked into the room and others have said the opposite about the same vendor.

The published evaluation retailers will be shown in a pitch is The Forrester Wave: Commerce Solutions, Q3 2026, which scores vendors on the strength of the product and on the vendor's strategy for it. Some carry a halo, which marks customer feedback rather than product capability. It is the first time Forrester has assessed B2B and B2C commerce in a single report.

Shopify account director Kole Mahan said the case for change should be built from a retailer's own profit and loss, line by line across headcount, platforms, applications and retail, separating spend that drives growth from spend that keeps the business running as it is.

"If 90% of my P&L is just on maintenance, how am I actually going to grow?" he said.

He said the push to run that exercise is now coming from outside technology teams, with finance, chief executives and procurement asking why platforms costing hundreds of thousands of dollars do not come with AI built in. He frames the decision on a 12 to 18 month view rather than six months.

Udina's framing is that retailers have already moved on from the question they used to ask.

"For a long time, the question was, is changing platforms too risky?" she said. "I actually think that question has moved. It's now, can the stack we've got actually keep pace?"

She put the same point to the room a different way earlier in the session.

"This is a conversation about a decision, not about software," she said.

### "This is a conversation about a decision, not about software."

**Hannah Udina**

![](https://yaffamedia.shorthandstories.com/Shopify_BuiltToAdapt/assets/dBBXLMIcur/hannah-udina-headshot-hires-enhanced-681x680.jpg)

![](https://yaffamedia.shorthandstories.com/Shopify_BuiltToAdapt/assets/q4MTk9uH9s/pexels-ron-lach-9783835-4096x2731.jpg)

Click above to catcth the full webinar

![](https://yaffamedia.shorthandstories.com/Shopify_BuiltToAdapt/assets/eT9mEM29VN/shopify-logo-white-400x200.png)
