Composable computing is most advanced in the field of digital commerce and content, also known as digital experience (DX). This is a discipline that values the flexibility of an IT architecture based on Microservices, APIs, Cloud-native SaaS and a 'Headless' approach to supporting heterogenous clients — four features that make up the acronym adopted by the MACH Alliance, an industry body that promotes this model. At its MACH TWO conference last week, there were plenty of global brands on hand to talk about their experiences with the model.
It's still early days — a board meeting of the organization last November estimated that adoption is around the 10% level, reflecting the slow pace at which enterprises tend to upgrade their computing platforms. But therefore the learnings of these early adopters is particularly valuable in understanding some of the challenges.
Interestingly, those early adopters seem to be more prevalent in Europe than in the US, in contrast to earlier technology waves. A survey of global IT decision makers carried out for the MACH Alliance earlier this year found that adoption is fastest in Germany and the UK. In a discussion at last week's event, Balakrishnan Submaranian, VP Digital Demand at global confectionery to petfood giant Mars, speculated that European adoption is being driven by the need to serve the needs of disparate local markets with differing requirements for logistics, warehousing and regulation. He says:
Each country has such different needs in Europe ... The US I would assume is more standardized. Although it's a big market and a big revenue base, the variety of what you need to do across the whole map is not as much as in Europe. So probably necessity is the mother of invention.
In retail, some of the UK's leading brands in particular were early to move to omni-channel. Submaranian says that because this was difficult to do using traditional monolith solutions, they built their own composable stacks. He explains:
The MACH journey was recognized probably in the 2012-2013 period by the likes of advanced retail companies like Tesco, to cater to that omni-channel need, click and collect, the grocery advancements ... The time between 2013 to I would say 2017/2018 was when these organizations built up large-scale tech force inside to actually decompose these capabilities ...
So you see the MACH journey for these companies started long ago. The advantage I see now with composable and MACH is, for especially companies like Mars, which are trying to get closer to consumers — we're typically not a retailer but we need to be on the journey — we don't want to do the same thing that Tesco did, by standing up large-scale engineering teams. What we are trying to do is ... not build stuff, but assemble stuff. That I see is a big differentiator of MACH.
A composable architecture is destined to become the mainstream choice because of the pressure to adapt faster while reducing costs, says Anca Iordanescu, VP Engineering at home furnishing chain IKEA. She comments:
The only way to get to economies of scale is actually to become composable and actually do the standard thing. We are evolving from what we called a layered architecture into a composable architecture. That's where we see that's the only way we can create this experience that we want to bring our customers, from omni-channel to stores, also our coworkers, so we can create bigger adaptive journeys for our customers. So it's really a standard for the future.
Obstacles on the path to composable
The accounts of these early adopters reveal plenty of obstacles along the way. Moving to a more agile, composable architecture means adapting to a different mindset in relation to IT — one that challenges many of the preconceived notions people have about project priorities and decision-making. Therefore it's important to build trust right from the start by showing quick wins and progressing incrementally. Submaranian says:
You need to start getting the belief in the organization, that this is the right approach. To keep the executive backing to pursue on this track, it is critical to show results in three months, six months, nine months. Our entire project plan ... was not based on, 'In three years time, we will go live.' The project plan is based on showing what can be done in shorter sprints, which is production ready [and] actually showing results.
Those results need to be tangible to the business, rather than being measurable solely in technology terms. Sportswear brand ASICS is in the midst of moving from an existing e-commmerce platform to a more broadly-based digital experience ecosystem that will help it build lifetime engagement with consumers. An important part of this is to take more of a product-based approach rather than being project-based, says Mindy Montgomery, Associate Director, Product Management, Consumer Platforms at ASICS Digital. For example, the group found that technology projects to add new payment methods had delivered no real business value. She says:
Based on the data analysis we've done, adding a new payment method doesn't bring in new customers. It just moves their payment method from one to the other. So we need to get out of that project-based, very small feature-based approach, and into a more product-based approach and thinking about how do we engage with users.
Instead, the focus needs to be on "the job to be done," whether that's a merchandizer adding new product data, or a consumer looking to buy some new running shoes. This requires trust and collaboration between product teams and technology engineers as they work together to figure out what's really required — "Having that trusted partnership where we can kick around ideas, and really iterate on what the solution is going to be," she says.
A similar change of mindset was needed at Interflora when its European operation began to unify a highly fragmented technology stack on a composable platform. Nicholas Pastorino, Group Chief Product and Digital Officer, comments:
Progressively, people are understanding that if you want to allocate your resources wisely and spend our money well, we need to find what the problem is really well — rather than jumping to a conclusion, designing a solution, testing it out and discovering maybe four or five months later, or after years of investment, that's not solving our problem.
Data instead of intuition
Another change of mindset revolves around cleaning up data and being prepared to make decisions based on what the data says. In the past, the data has been so unreliable or fragmented that people had no choice but to rely on intuition. But now that habit has to be unlearned. Pastorino says:
Making informed decisions based on data and not on your intuition is very new thinking in the organization. That's been a major roadblock in transforming, around communication.
At ASICS, the desire to bring together product information for consumers led to the realization of how much variation there is in the details held for each product. One measurement that's a key criterion when choosing running shoes is the difference in height between the heel and the sole, known as the heel drop. Montgomery says:
When we were doing the analysis of designing what product information we wanted to put on the screen, we found out that only 14% of the running shoes have heel drop information in products that we have available. So we're undergoing a massive project right now to identify where the gaps in the data are, not only in running shoes, but in all other apparel, to really say 'Okay, here's the complete product data. This is where it needs to live. This is where it needs to be cleaned up.'
The other big change is being ready to abandon projects or investments when they aren't working out. Pastorino says:
If we want to go fast, fail fast — think build small, fail fast — you need to build a lot of things. And 90% of it is going to be dismissed because it's not working or it's not interesting for users, not bringing value.
The sunk cost fallacy
This principle applies to larger projects too. At ASICS, a project to build a 'shoe finder' application for guided shopping was put on hold recently in light of the emergence of ChatGPT. Montgomery explains:
We invested probably about 18 months' worth of development time with a couple of new technology solutions to build out our shoe finder application. That is a series of six questions that apply to things like, 'What's my relationship with running? What surface do I like to run on? How much do I run a week?' and build out a list of products that are suitable for you, based on the product information we have.
Then along comes ChatGPT and others. So now we're taking a pause on all of our product discovery and guided shopping experience roadmap to really think, Okay, do we want to consider continue to invest in that shoe finder solution, which is very transactional? And is our guide to how we think you would find the product? Or do we want to look at building out a curated ChatGPT-like experience where you can ask the questions, or you can input the information based on what your needs are?
It may seem drastic, but it'll be worth it if the end result is a solution that lasts longer and works better because it can give more context-aware advice to help the customer find the shoes they need. Montgomery adds:
I've seen in my career, people double down because they put money down once. So it's like, 'Well, we did all this investment here. If we walk away from it, that's lost money.' Well, if you continue to invest in it, and it's not fruitful, that's more lost money. So we're really taking a harsh look at what we have to do in order to move forward from a business perspective. Even though the sunk cost may be in technology, it's really a business function to start looking at the dollars and cents.
I was intrigued by the finding that MACH adoption seems to be more advanced in Europe than in the US, which is the reverse of the usual order in technology adoption. But perhaps that has been due to earlier technologies benefitting from the massive scale of the US market, whereas the greater variety of local requirements in Europe becomes a forcing factor for adopting a more composable approach. This is an exception to the rule, however — look at AI, and scale becomes the key factor once again. So Europe's lead in MACH may turn out to be short-lived.