From Continuous Delivery to Business Innovation: Mastering the Product Innovation Cycle

For senior engineers, merely having the ability to continuously deliver value is not enough; we must also ensure that what we deliver is useful and creates real value.

Continuous delivery and business innovation are interrelated and complementary. Continuous delivery is the foundation of business innovation, and the benefits of continuous delivery must ultimately be realized through business innovation.

How can we achieve effective business innovation based on a deep understanding of the business? Below is a brilliant sharing by senior Alibaba technical expert He Mian .

Effective business innovation relies on continuous delivery capabilities. Continuous delivery enables faster feedback and is the basis of innovation in the Internet era. What makes Internet‑era innovation different? See the diagram below.

The first image shows icons of some shared‑bike apps, many of which have exited competition in the past year; the second image shows icons of some live‑streaming apps, illustrating the “hundred‑teams, thousand‑streams” phenomenon—massive competition in live streaming that has not yet ended, and now short‑video battles have begun.

In the mobile Internet era, the cost of entrepreneurship and product development has dropped dramatically, and channels to reach users are more direct. Users now have many more choices, and decision‑making power has shifted from producers to consumers. If the delivered product is not chosen or ignored by users, it becomes the biggest waste in the innovation process.

Moreover, we must face the accelerating‑change world, known as the VUCA era. VUCA originally described the post‑Cold‑War international political environment and stands for Volatility, Uncertainty, Complexity, and Ambiguity. Today it is used to describe the technology and business environment, which is the context that product delivery and innovation must confront.

The accelerating environment and the shift of choice power to users pose new challenges for product innovation and delivery models. We need a more effective product‑innovation model.

Based on continuous delivery, we have the opportunity—and the necessity—to introduce systematic innovation methods. This leads to the topic of moving from continuous delivery to product innovation.

Effective innovation starts with clear goals. Where do goals come from? They stem from mission and vision, which are broken down in space and time. Spatial decomposition means analyzing the elements needed to achieve the goal and their relationships; temporal decomposition means sequencing the steps to build those elements.

First, let’s look at spatial decomposition. We introduce the concept of the flywheel effect. The flywheel was first proposed in the book *Good to Great* by Jim Collins, who found that enduring companies share a positive feedback loop—a flywheel—that continuously reinforces business growth. This flywheel also exists in fast‑growing Internet businesses.

Take Amazon as an example. In its 1997 IPO prospectus, Jeff Bezos described Amazon’s flywheel: provide users with abundant choices, which improves user experience; a better experience attracts more traffic; more traffic draws more sellers, which expands choice further, creating a virtuous cycle. Another branch of the flywheel is that larger scale reduces cost structure, leading to lower prices, which further improves user experience and fuels growth.

The difficulty of the flywheel lies in how to start and sustain it. To get a stationary flywheel moving, you must apply a large force repeatedly—each turn is hard, but the effort accumulates, and the flywheel spins faster. Once it reaches high speed, its momentum makes it easier to overcome resistance.

However, the flywheel can stall if any link breaks. In Amazon’s case, if users do not value abundant choice, the flywheel stops, so we must understand the underlying assumptions, validate them, and build a sound business loop.

Many successful businesses—Taobao, Hema, Douyin, WeChat—exhibit similar flywheels. A flywheel is essential for scaling products; without it, competing in today’s harsh environment is extremely difficult.

Let’s imagine building a short‑video app similar to Douyin. The theoretical flywheel would be: better viewing experience → more users → more content creation → more high‑quality content → better experience, forming a reinforcing loop. The key question is where to start the flywheel—chicken or egg?

Better viewing experience brings more users; more users lead to more content creation; more content leads to higher quality, which again improves experience. The problem is where to spin the wheel first. For example, to ensure an initial supply of high‑quality content, we might curate existing content or subsidize creator incentives—these are the “handles” that start the wheel.

Other handles include attracting initial KOLs (key opinion leaders) through content subsidies or external sourcing, and designing an initial viewing experience that encourages user‑generated content.

We also need to consider product tone during growth. Different tones lead to different expansion paths—e.g., Kuaishou and Douyin have distinct tones, affecting their scaling strategies.

Based on business goals, we built a growth flywheel and identified the handles that drive it. This spatial decomposition provides the initial direction for constructing product and business models.

Next, we need to address the temporal dimension—what to focus on first, second, etc. The “gate‑keeping model” (or “gate” model) states that product development must pass through sequential stages, each a prerequisite for the next.

The first gate is empathy—identifying an unmet need that strongly motivates users. Next is stickiness—ensuring users keep using the solution. Then virality—users invite friends. After that comes revenue—sustaining operations and growth. Finally, scaling.

Many teams make the mistake of scaling before achieving stickiness, driven by short‑term KPI pressure. The gate model reminds us that each stage must be validated before moving on.

Returning to the growth flywheel, we must decide where to start. The user experience is the natural entry point because it directly relates to empathy and stickiness.

For Amazon, the core experience is price and choice. Starting with books made sense because books offered abundant choice and competitive pricing, allowing Amazon to quickly build the flywheel.

Experience is not just UI; it is how well a product satisfies core user needs in specific scenarios. For example, Hema’s core experience for urban white‑collar users is: 1) instant delivery within half an hour, 2) one‑stop grocery shopping, and 3) freshness and convenience.

Similarly, Baidu’s MP3 product succeeded because its core experience was “searchable and downloadable.”

For short‑video products, the core experience is a steady supply of high‑quality content and precise distribution, especially for an initial core user segment.

Different teams may define core experience differently, but a clear, focused definition is essential.

Time‑based decomposition emphasizes starting with the experience, then iterating through user‑centric steps.

From the business goal, we built a business map that guides which goals to focus on at each moment. Quantifying goals allows us to know whether we are moving in the right direction.

We define a product “big picture” from the user journey: each block represents a user action—content production (deep blue), content consumption (light blue), and the desired outcome (green) of content publishing, viewing, and relationship building.

The core of the user‑path diagram is the user; product success depends heavily on converting user behavior—metrics such as conversion rate, activation, and retention are crucial.

Based on the impact map, we select the most effective features, plan iterations, and ask whether we need to implement the entire map. The answer is no; during requirement gathering we brainstorm widely, but during planning we must converge.

Features marked “1” form the minimal viable product (MVP); those marked “2” are for later consideration.

Delivering features does not guarantee success; success is measured by goal achievement. Between features and goals lie two assumptions: (1) the features will produce the intended impact on user behavior, and (2) that impact will help achieve the goal. These assumptions must be validated with data.

Impact mapping makes these assumptions explicit, allowing teams to assess impact and goal attainment after feature delivery, fostering a data‑driven, outcome‑focused culture.

In summary, we start from business goals, decompose them spatially (growth flywheel and handles) and temporally (gate model), build an impact map linking why, who, how, and what, and use continuous delivery to iterate quickly. By continuously measuring, validating, and adjusting, we create an effective innovation loop that turns delivery capability into real business value.

Correct questions lead to good results.

Whether we adopt agile, improve efficiency, or any other practice, we must first clarify the problem we aim to solve.