Two-Speed IT Architectures

I recently read an interesting McKinsey article, Organizing for digital acceleration: Making a two-speed IT operating model work.  The article argues that born digital  Internet companies, - e.g., Amazon, Google, Facebook, - are setting the standards in the quest to understand and optimize the customer experience, challenging traditional companies to keep up.  These older companies have to both leverage the back-end systems, processes and overall capabilities that have served them well over the years, while adopting innovative consumer-facing technologies to enhance their customers’ experiences.

The article starts out by summarizing the best practices of born digital companies:

• Provide the appropriate information and platforms so customers can find what they need quickly and reliably. 
• Align technology and production systems with business goals. 
• Limit the size of application-development teams. 
• Focus on agile product development to enhance customer engagements. 
• Encourage product managers to think about digital experiences rather than discrete applications or components. 
• Collect and learn from consumer data.
• Engage in frequent testing and experimentation.
• Don’t be afraid to fail.
  

It then lists some of the challenges faced by traditional companies as they struggle to keep up: 

•  Reliance on legacy enterprise resource planning systems.
•  System changes that generally take months of development and testing.
•  Inadequate culture, governance models, tools and skills, making it difficult to pursue the agile, test-and-learn approaches used by their digital competitors.

To overcome these challenges, traditional companies should deploy two key, mutually reinforcing steps:

•  Adopt a two-speed IT architecture - “one that decouples the management of customer-centric front-end systems and applications from the management of existing transaction-oriented back-end systems.”
•  Embrace the digital product management used by successful digital companies, which “empower managers to incorporate users’ feedback into product-development efforts actively and systematically and hold managers accountable for results.”

To me, this feels like “déjà vu all over again.”

Traditional companies have been struggling to keep up with their younger Internet competitors ever since the explosive growth of the Internet and World Wide Web in the 1990s.  Many of these older companies were pioneers in the use of data processing applications to automate back-end business processes, - e.g., financial transactions, payroll, and inventory management.  They later deployed more sophisticated applications to help better manage their overall operations, including enterprise resource planning and customer relationship management.

In the 1980s, they embraced personal computers, workstations and client-server systems along with personal productivity and front-office applications.  By the time the Internet era came along in the 1990s, these traditional companies owed much of their success to the considerable investments they’d made over the years in their IT infrastructures and applications.

The open standards and open source software of the Internet made it much easier to interconnect their disparate systems, - regardless of vendor, architecture or operating systems, - including their PCs, client-server systems and enterprise mainframes.  By integrating their existing databases and applications with a web front end, they could now reach their customers, employees, suppliers and partners, - anytime, anywhere.

But, the Internet was also highly disruptive to established companies.  Internet startups brought to market all kinds of innovative applications, - many didn’t make it, but some turned out quite successful.  Older companies were now facing new competitors, whose new technologies, talents and culture enable them to quickly get their offerings to market, and once in the marketplace, to continue to rapidly enhance them.  As the dot-com bubble gathered intensity, the buzz in the air was that in the Internet-based new economy, older companies, - with their legacy infrastructures, - could not possibly keep up with the new generation of fast-moving startups.

In the fall of 1995 IBM made the decision to embrace the Internet across the whole company, and created the IBM Internet Division to orchestrate this company-wide effort.  Our job was to figure out the implications of the Internet to our clients’ businesses.  It became increasingly clear to us that the universal reach and connectivity of the Internet was truly transformative.  But what differentiated our e-business strategy from what many others were saying was our belief that every business and institution would benefit from embracing the Internet, not just startups.

We were convinced that the brand reputation, customer base and IT infrastructure that companies had built over the years were even greater assets when properly combined with the new capabilities of the Internet.  And they could start quite simply, by web-enabling their existing systems and applications.

Underpinning e-business was a two-speed IT architecture: the relatively slow moving, mostly back-end legacy assets, combined with the much faster moving, mostly front-end new Internet-based offerings.

IT has continued its dramatic advances in the intervening years.  We are seeing a seemingly perfect storm of technology and market innovations, - e.g., mobil, Internet of Things, social, cloud, big data and analytics, artificial intelligence.  And, as the McKinsey article points out, older companies are once more challenged to keep up with the ongoing digitization of the economy.  They must figure out how to best compete with born digital companies, a number of which are now very large and successful.

Let’s look in more detail at the two key steps that McKinsey recommends.

Two-Speed IT Architectures

A two-speed IT architecture would once more enable traditional companies to take advantage of their existing back-end processes and infrastructures, while deploying state-of-art digital practices for their front-end, consumer-facing applications.

The biggest challenge in implementing such a two-speed architecture is aligning the organization on how to best connect front- and back-end systems, and where and when to decouple them in the interest of speed.  Setting up a totally separate and independent digital organization is a good way to get things moving quickly, but only if the independent digital organization works closely with the rest of the business and is properly aligned with the back-end IT developers.

“Leaders in both traditional and digital camps must designate clear roles and responsibilities, and they must build mutual trust among team members… When legacy systems and new front-end innovations are aligned, staff members in traditional and digital groups can better coordinate their efforts…  Each side has a role to play in getting things right.”

Digital Product Management 

It’s not enough to embrace agile development and common front-end platforms.  “Companies must reorient themselves and their product-development processes around customer experiences and insights.”  To do so, they should adopt a digital product management model.

A digital management approach requires that teams be held accountable for implementing improvements to their digital products that incorporate customers’ needs and insights.  Time-to-market is key, so that the team can involve customers in the creation process, eliciting and acting on their feedback.  This is a very good way to both streamline the development process while engendering customer loyalty.

Digital product management should be guided by six key imperatives:

• Clearly define your digital products.  Define the user or business problems being solved through digitization, “identifying which digital activities have been designed to address the problems, and prioritizing those activities according to the amount of development resources required for each.”
• Assign discrete parts of the customer experience to product managers. “They should be given freedom not only to refine existing digital products but also to redefine them in ways that create better customer experiences.”
• Test prototypes with customers. “Digital product managers must be given the leeway to get in front of customers with minimally viable products…  Ideas must be tested early and often with actual target users, whose views on the utility and viability of the proposed digital experience are likely to be more accurate than those of internal stakeholders.”
• Recruit differently.  “Digital product managers must interact with professionals from across the product-development spectrum.  So they must be well versed not just in technology concepts but also in engineering, user-experience design, agile product development, finance, marketing, and many of the other professional domains that enable the rapid creation of an online product or service.”
• Strengthen relationships between product managers and product engineers.  “A digital product management approach and two-speed architecture both must be supported by a product-management organization that interacts seamlessly with the members of the IT group who are designing and delivering digital products and experiences.”
• Create appropriate incentives.  The performance of digital product managers must be measured according to their ability to support a successful customer experience, such as higher sales and customer-satisfaction scores.  “Product managers’ performance should be evaluated using business metrics that they can influence directly.”

“Having both a flexible IT architecture and a digital product management capability is a prerequisite for competing in today’s fast-moving markets.  By aligning the processes that manage back-end and front-end IT systems and by holding all staffers and stakeholders accountable for adopting new mind-sets and processes, organizations can create compelling customer experiences online and compete more effectively with digital rivals.”

Comments

[William Tetzlaff]

William Tetzlaff

SOSP 2015, was a few weeks ago, and I have a few anecdotal observations. This was the 25th SOSP, which is the premier systems conference, which has run every other year for 50 years. In the set of retrospective talks about contributions to systems technology, for fifty years, IBM was clearly the most mentioned company. On the other hand, by attendance, speakers, poster sessions, and the small exhibit area, the movers and shakers in Systems are now Face Book, Amazon, and Google. These are companies that are not just satisfying customer needs, using the internet, they are also innovators in the boiler room, building scalable infrastructure.