Internet giants are not only shaping creative business models, they are reinventing the way to manage computer science, software applications, and moreover, to better their human capital. As their new products are intrisincally linked to computing, their reengineering unfolds new tremendous paths for innovation.
Octo Technology, a savvy IT consulting firm, has recently reviewed some of the most emblematic firms of the Internet, such as Amazon, Facebook, Google, Netflix, LinkedIn, and extracted ten best practices used to maximize applications delivery.
No matter how young these companies are, it is surprising to discover their creativity in leadership and management: as if these ‘engineers from the Silicon Valley’ truely believed in the ‘power of their men’, in a simular way they trust their computing machines; as if they have realized they rely deeply on their human capital in order to beat the competition.
Their successful patterns are sorted in three thematics: bigger, faster, harder.
The first type of practices is related to scale. How Facebook handles continuously over 1 billion users, Google and Yahoo! manage over 200 millions email boxes, Twitter delivers daily 400 millions tweets, is absolutely impressive. No former computer system had to handle such a massive market.
Octo identifies three untraditional approaches that Internet players have used to scale:
- Build vs Buy, which means the use of open source software vs software package such as ERP (enterprise resource planning) or CRM (customer relationship management) applications package. Naturally software package license fee are an issue for start-up with initially limited resources; the restriction in the use of software package also relies in their difficulty to scale. They unable to help the start-up build a computing system specifically adapted to their business model: software package are generic in essence when start-up are creating a specific activity analyses Octo. At last, Open Source, by activating the intelligence of the many, turns to be an much more economical and powerful tool: Internet giants leverage on Open Source, and share back their developments to Open Source (Octo lists: Cassandra by Facebook, Hadoop and HBase inspired by by Google and developed by Yahoo!, Voldemort by LinkedIn, Bootstrap responsive design toolkit by Twitter, and last but not least, Android, Google operating system for mobile devices).
- Networked commodity hardware : if scaling innovation in the 20th meant manufacturing, in the 21th, scaling innovation relies on network connection. Considering computing, the most effective and economical ways to handle big data (over 10 To) and high-rate transactions (over 1000 transactions/s) seem to manage networks of commoditized servers, rather than setting-up giant servers. This network of servers shall enable easy and continuous access to server capacity through virtualization. It involves mastering new techniques: one remembers that to boost the performance of its network of servers, Google’s 18th employee was a neurosurgeon who had qualified at Harvard and the Yale School of Medicine kinds of databases. Similarly, it leads to moving from central relational databases to distributed databases (Sharding, distributed data management; NoSQL concept; technologies like Hadoop or Cassandra) and new paradigms: availability must prevail, and facing partition, which happens when communications fail, might turn out to be preeminent compared to consistency (re-reading of CAP theorem);
- API ecosystem: we have talked about the power of API in our description of modular design. API is seen here as a way to scale on various dimesnions: catching additional revenues (Google maps), marketing your service through third-party applications (Twitter), building a creative ecosystem, letting others build value on top of your platform. With ‘API first’ pattern, API is used to design its own end-user app, the same way third-party would use it, a kind of “Eat Your Own Dog’s Food”.
That is where the ability to grow does not only rely to your capacity to scale. If some new customers will join as a result of marketing and brand awareness, others will be attracted by new unique features: you need innovative products to fuel the growth engine.
According to Octo research, Internet giants have set out three peculiar approaches to accelerate innovation delivery:
- Minimum Viable Product is linked to Lean Startup approach developed by Eric Ries: MVP corresponds to the most basic release of the product that can be tested by a customer, and generate feedback. You have to select one single or a limited number of features to simplify your product, and put it at the core of your first release. We see that MVP refers to the idea of Focus cherished by Steve Jobs, and Fast Prototyping and Iterating, promoted by Design Thinking by Tim Brown. Similar thoughts are exposed in ‘Keep It Simple and Sexy’ post. Nevertheless, it is not because the product is limited that the core feature should not be not perfectly up and running: otherwise you are sure to fail your test. The challenge is to select the winning core feature, ‘the killer app’ as soon as possible. That’s where the attributes of Meaning, Belief, Social Imaginaries based on user understanding, step in. They lead to designing something which holds meaning for end-users, echoing their social imaginaries through simple metaphor, and funding your innovation on your personal unwavering belief in the fact that it will change peope’s life;
- A/B testing, made popular by Amazon, prolongates the former idea by setting up testing of a similar feature among 2 separate populations with some changes in the lay out (home page, visual of an item, …). LinkedIn is even able to sort 2 populations of lawyers in the only area of Boston. This easy-to-run mechanical approach helps to reach rapidly an optimized release by making successive iterations smooth;
- Unifying development and operations teams facilitates innovation delivery and accelerates go to market; following concepts are in use: Infrastructure as a Code (where usual coding can manage the infrastructure system), Continuous Delivery (where new release are delivered at fast regular pace), Cooperation Culture and common language among the teams. Amazon has extended this concept to a modular organization, each team being accountable for running the system it has shaped: “you build it, you run it“.
Internet giants are fast to innovate and are skilled at scaling, but it would mean nothing without their ability to manage broadly their impact: Internet giants really nail the disruptive formula, leveraging on their step ahead to generate exponential margins. How to reach this efficiency?
Octo highlights four organizational patterns of their operating system:
- A measurement oriented culture: “In God we trust, everything else we test” comments Octo; one could only say: “you can only improve what you can measure”. Internet giants are in race for perpetual improvement of their response time, the number of pages viewed, the items which interact the best with Internet users, or the mean time between failure (Google disk failure experience): these performances directly impact directly their revenue stream. More unexpectedly, measures can also be used to build better HR processes and management rules as in the Oxygen Google initiative.
- Pizza Teams which means keeping teams small: 5 to 15 people (8 being the appropriate size of team you can feed with 2 pizze, according to Amazon). Amazon’s CTO, Werner Vogels claims: ‘Small teams are holy’, they are agile and creative, and limit at the minimum the bureaucratic flow;
- Feature Teams, or multifunctional Project teams concentrating all necessary skills to achieve autonomously a functional module or product;
- Design for failure, assuming that outage is part of the game: Eventual Consistency (identifying ways to keep the system running even if data are not consistent for a given moment of time), Graceful Degradation (degradation keeping the system alive with minor shortcomings), Feature Flipping (activating features independantly from delivery), Simian Army (completing an army of random failure tests).
Other additional patterns from Internet Giants are described by Octo such as:
- fluent customer experience: interface must display in a blink;
- talented and perfectionist coding engineers, referring to Software Craftmanship, and motivation tools: stimulating environment, Hackathons, Dojos;
- perpetual beta: continuous cocreation with the users-developers;
- device agnostic: maintaining optimum experience on any device;
- cloud first.
Crossing with Rapid Innovation model
Though it is focused on accelerating innovation in large companies, the Rapid Innovation model share many communalities with the Internet giants patterns.
Llet’s go through the four principles of the model:
- Setting-up an agile and autonomous innovation entity: this first principle is obviously close to Pizza and Feature teams rules;
- Designing within a framework of “creative tension”, an exciting environement mixing stimulating and stretched goals with a proven expertise in innovation discipline. This second concept meets with the toolbox including MVP, A/B testing, Build vs Buy, Customer experience, Unified teams;
- Naturally, the third principle, aligning with innovation group strategy, across a shared portfolio and permanent connections, is not an issue for a start-up, and not relevant for Internet giants at the beginning; but once the growth has come, the notion of alignement will certainly be precious to share a common vision across lultiple agile entities;
- In the course of seeking adoption and engagement from the core corporation, we came to shape the fourth principal of modular design, an “innovation by component” approach; no doubt it matches the API ecosystem pattern.
Formula to success
If the recipe is so widespread, why doesn’t anyone reach success?
First of all, what a difference between the plan of a recipe and its successful execution! It’s about talent: top chef or great performers show us the gap.
Then, no one can rests on its laurels: you have to reinvent your initial formula, to change while remaining yourself, growing while keeping your identity. In other words, ‘product that made you big may not always be your core business’ noted @JimmyDaugherty in his post about Amazon. And who could demonstrate a better quintessence of continuous transformation than Amazon, which managed changing its business model every three or four years?