Keys to D2C Success: CX, Agility and Analytics

While much of the critical attention on the most recent crop of D2C platforms is focused on the comparative depth of their respective content libraries and the discovery mechanisms available to navigate those libraries, the long-term success of any of these platforms will be determined by the depth and breadth of the entire consumer experience. Building out a D2C streaming platform with a compelling and competitive consumer experience requires a significant and ongoing investment driven by analytics as well as inspiration. How can media publishers best leverage that investment to win and stay ahead?

In going over-the-top, media publishers are fundamentally altering their revenue ecosystem. In bypassing the traditional intermediaries (local stations and MVPDs), they are breaking free of consumer experience (CX) limitations and gaining direct access to consumer behavior data. For ad-supported programming, the combination of CX freedom and first-party data can open the door to much more engaging, targeted, and lucrative advertising than was ever achievable with traditional linear spots (no matter how addressable). Regardless of monetization model, these services can now leverage CX and user navigational data to directly impact subscriber acquisition and churn.

But this also means D2C media owners have to stand up the infrastructure necessary to scale, support, and maintain millions of direct relationships. They’re also flying in a wider sky, competing not just with other streaming services for share of wallets and eyeballs, but with more interactive forms of media for share of attention. All of which is to say they need to keep innovating at Web, not TV, speed. To do so requires a clear answer to the question: “How will you keep up?”

Challenges

The path forward is complicated by a number of technological and cultural challenges:

• Lots of devices. D2C platforms need to dazzle and delight consumers across a broad and growing range of disparate devices, from smartphones, tablets and watches, to streaming sticks, smart TVs and legacy set-top boxes, and eventually to IoT devices, autonomous cars or anything with a display-capable surface. With few common capabilities, standards, or programming interfaces, spanning such a proliferating media/consumer interface with a brute force approach would be extremely expensive.
• Organizational impediments. At the risk of overgeneralization, product owners, designers, and developers have divergent perspectives, foci, and ecosystems; this has often led to communication barriers, siloed team structures, and sequential (and less than agile) processes producing suboptimal results.
• Legacy integrations. New feature ideation often gets bogged down with the minutiae of integrating innovations into an existing, often messy, set of enterprise systems. This impedes the exploration and validation of new features, requiring significant resource investment before they can be proven viable and thus inhibiting innovation.

Success Factors

From our perspective, long-term success will depend upon:

• A process orientation,
• An approach and tooling that facilitate the generation of validated human-centered feature designs, the transition of those feature designs into mainstream software development, and the deployment of the resulting feature implementations across the full range of consumer devices, and
• Analytics to inform the process.

The Innovation Feedback Loop

With new D2C services appearing almost monthly, it would be easy to assume that these launches are one-time disruptive events, but the reality looks much more like an ongoing war of innovation. Netflix has internalized that viewpoint, building their innovation processes to continually deliver new features as they are developed, eschewing the more traditional release-based model for agility and velocity. Facebook, Amazon, and Google work in much the same way.

But software development agility is just one part of the puzzle; iterating quickly on mediocre product concepts only exposes their mediocrity faster. Stepping back to look at the big picture, we see that the whole ideation-creation-validation-implementation-deployment sequence must become a closed loop driven by feedback and metrics.

Iterate to Innovate
The CX innovation process consists of two sequential cycles that in turn feed the production software development life cycle. Agile software development is driven by a backlog of clearly defined and validated product feature requirements; these two cycles work together to build that backlog.

Cycle #1: Innovation Piloting

The purpose of the innovation piloting cycle is to create new feature candidates through a process of generation, clarification, testing, and evaluation. Led by a product owner, the innovation piloting team should be made up primarily of product strategists and designers, with a few full-stack developers to provide implementation perspective and do the rapid prototyping. This dedicated team should reiterate the piloting process on a regular basis.

Innovation piloting focuses on the following:

• Identifying challenges and opportunities to improve CX. The consumer’s experience defines the product, so CX should be the primary focus of innovation piloting. In this part of the process, the piloting team is looking specifically for ways to reduce friction across the entire range of user interactions, seamlessly augment the core experience with relevant interactions, messaging and media (beyond long-form video), and reinforce the relationship through moments that catch the user’s eye and establish brand identity.
• Generating and refining potential feature concepts. The piloting team should agree on a specific set of identified challenges and opportunities to be addressed in the cycle, then work together to brainstorm new feature concepts that address them. All feature concepts must be consistent with the overall CX, even if they involve a variety of user interactions.
  Prototyping and evaluating new feature candidates. Once the team generates a list of viable potential feature concepts, they should build them out just enough to subject them to human interaction and testing. By putting a basic version of each concept in front of real users, the team can identify the strengths, weaknesses, and nuances of each and evaluate which ones are worth bringing to market.

The output of each cycle iteration is a working prototype of each new feature candidate.

Cycle #2: Scale-Up

The purpose of the scale-up cycle is to flesh out each of the feature candidates produced in the piloting cycle to the level of detailed definition necessary for production implementation. Led by a software architect, the scale-up team should include designers and senior developers, as well as the original product owners from the piloting cycle.

Scale-up focuses on the following:

• Resolving the capability gaps between the reference architecture and production environment. Some feature candidates may require or assume more functionality from consumer devices, back end services, or enterprise systems, so the team will have to decide if the feature should either be detuned or delayed until the production environment can catch up. In other cases, the feature definition will need to be expanded to address or leverage production functionality not available in the reference architecture.
• Thinking through the corner cases and suboptimal situations. Feature piloting will illustrate basic scenarios in which everything happens as it should – developers call these scenarios “happy paths.” For each feature candidate, the team will identify the situations in which the feature may interact with or depend on other features and systems, and then define how the feature should behave when those situations don’t play out as they should.
• Addressing what happens when millions of users are interacting with the feature. Most features should scale cleanly, given that they will be made available through a production environment that already handles a large volume of interactions. But there may be features, particularly those that make use of external systems (e.g., advertising technology, social platforms), that exhibit performance bottlenecks as the number of simultaneous users increases. The team will need to model feature usage to identify these bottlenecks and develop a resolution approach.

The output of each cycle iteration is a loaded backlog of detailed requirements, design assets, and reference implementation.

Leverage Facilitating Tools

Both cycles require the exchange of ideas and work product between designers and developers; this requires a set of tools and approaches that facilitate this exchange as well as the transfer of features from pilot to production. For example, design tools that output detailed development specifications and assets make it possible for designers and developers to work side by side, collaborating in near-real-time rather than in sequential, stovepiped workflows.

A reference architecture will facilitate piloting, allowing any prototyping to be decoupled from the often challenging task of integrating with enterprise systems. As new features are scaled up, a reference architecture will guide and ease the integration of those features with production back ends.

On the other side of feature development, a common development platform, a device-agnostic presentation engine, and an automated test environment will allow one intelligent design to be deployed across every type of device in a high-velocity yet cost-effective manner.

Close the Loop With Analytics

The difference between a procedure and a process is the use of metrics to drive a feedback path, so the ability to close the loop in this case depends upon the selection of useful metrics. Each of the typical monetization models has a set of customary Key Performance Indicators (KPIs): subscription-based services typically look at Cost Per Acquisition and Customer Return Rate, transaction-based services look at things like Take Rate and Bounce Rate, and ad-supported services are measured by CPMs, CPCs, and conversion rates. But these are only first-order metrics – they can be used to measure overall business performance, but they don’t offer insights as to how to improve specific aspects of the product or process.

Advanced analytics can be of great use, providing much more meaningful metrics and insights. Analyzing the user clickstream using machine learning and artificial intelligence techniques can provide a wealth of information:

1. User interaction analysis offers insights into how users are navigating through the experience, allowing the service to isolate interactions that trigger frustration, discover unforeseen usage patterns, and identify under- or over-utilized features.
2. User behavior profiling classifies users by how they consume content or use the service. These profiles can be used to generate recommendations, facilitate social connections, inform content development, focus messaging, and target new features.
3. Audience segmentation builds on user behavior profiles by integrating additional data from opted-in subscribers to generate audience segments that are more interesting to advertisers, and whose past behaviors can be used to forecast future behavior.
4. Demand forecasting analyzes historical user behavior data to forecast future system usage and content consumption. These forecasts can be used to optimize Content Delivery Network (CDN) bandwidth costs, pre-position content to minimize load times on new titles, and inform content acquisition decisions.

The punchline

The key to building a sustainable competitive advantage is to treat product innovation as part of a closed-loop process that uses consumer behavior analytics to continually enhance the consumer’s experience. This is a fundamental shift in perspective from that traditionally held by media owners and publishers, but it is vital to compete against the other players in the Attention Economy.