Operational analytics: The key to service assurance and consumer experience management
With increasing use of video, constant grade-my-paper/ data streams, and soaring customer expectations, operators and content providers are living in a very proofreading precarious world, with customers ready to jump ship at the first sight of trouble. Price wars are also driving down profits, meaning that Quality of Service (QoS) is increasingly critical to differentiation in a crowded market. Gone are the days when operators simply needed to provide reliable video services and quality content; now, data is king. A transformational shift in service assurance has become crucial to competitiveness.
Network failures and service degradation have a huge impact on profits, productivity and reputation. In today’s world of social media, the voice of one unhappy customer travels far, so problems must be fixed quickly. Yet a lack of real-time visibility into network operations is hindering operators’ ability to act, and means that they’re unable to effectively manage costs in their environment. This impacts directly on profits as Total Cost of Ownership (TCO) rockets. To remain competitive, operators need to find a way of driving down costs, while improving QoS.
Vimmi analytics and reporting tools enables operators and content providers to access critical real time traffic and usage information broken out by region, node, account, request, video stream, viewing model (VOD/live/playlist) and platform. The system provides all information needed to optimize resource utilization, traffic patterns and account policies.
Reporting on CDN historical performance and trends is based on data from monitoring systems and the CDN core. Parameters include storage utilization, service availability and load. Operators may monitor and manage log processing and track object distribution statistics. Information about all internal processes is presented clearly, enabling operators to keep constant tabs on system performance.
The Analytics and Engagement platform is a powerful and intelligent set of tools for creating interactive dashboards that enable content owners and operators to extract, analyze and utilize information about the behavior of the consumers in their platforms, including user, subscriptions, media consumption, application interactions, and campaign engagement.
Analytics and Engagement provides real-time and aggregated insights into consumers’ reactions and responses across any device and service. With Analytics and Engagement, operators at content distribution sites can build dashboards enabling the display of reports—in a variety of visually optimized views—containing real-time data that informs decision makers regarding the interactions and consumption of content by end-consumers. This information can prove essential in the taking of key business decisions concerning the acquisition, curation, marketing, and distribution of content. The outcome is enhancing monetization of content and ROI through expanding and maintaining an active and engaged subscriber base.
Analytics and Engagement provides interactive dashboards that enable businesses to segment users into audiences. Based on insights derived from audience data, Analytics and Engagements drives personalized campaigns to accomplish business goals such as content consumption, subscribing to a service, upgrading from free to premium plan, retaining subscribers, and reducing churn.
The Life cycle: How Analytics and Engagement Works
The Analytics and Engagement lifecycle is designed to take advantage of real-time data gathering, coupled with interactive administration screens. Together, these empower your business to take action and make informed responses based on the effectiveness of the actions taken.
Data is ingested into the Big Data system from various sources, thereby allowing for in-depth analysis and segmentation, based on abstract-level figures, as well as drill-downs to the most granular level of per-user or per-video detail
- App events
- Media events
- Consumption events
- User tags are utilized to track information about a user
- Performance events networks
- Campaign events
Analyze and Visualize
Reports are the central informational elements provided by Analytics and Engagement. For a dashboard to be of value to an organization as a whole—and to each user within the organization—it needs to contain informative and actionable data, in real-time, describing the metrics that enable the user to make an effective business decision.
By means of the data gathered across all apps and applications in the Big Data system, and the analytics performed on that data, B2C customers have the ability to communicate and market to their end-consumers directly in the app.
The audiences reached through engagement messaging will be targeted based upon rules set up by administrators, involving any number or combination of parameters and meta data in the Big Data system.
Audiences can be identified and defined by a wide variety of attributes gathered from diverse sets of data, including user preferences, interactions with the applications, consumption of content, subscriptions, and beyond.
In addition, audiences can be saved and used in comparative reports; and, based on the analytical activity stemming from those reports, targeted in structured marketed campaigns. What’s more, saved audiences can be re-used in the future.
Design and Launch Campaigns
Campaigns can involve one or several steps that entail engaging an audience to accomplish a goal. Ultimately, your purpose in running a campaign is to convince the consumers in your network to engage more closely with the platform, spend more time using it, acquire content; and, when the time comes for renewal, to extend subscription (strengthen retention) instead of leaving the network (avoid churn).
When designing a campaign, the engagement manager has the option to consider:
- The tailored audience to target
- Communication channel(s) (Push, Inbox, E-mail, In-App)
- Messaging Type – Deliver Text, Image, Audio, Video Rich Content and then Deep-link into the app for optimal campaign effectiveness.
- Frequency and Timing – Campaigns can be run as one-off events, or alternatively on a scheduled basis. For instance, a campaign could run daily with the aim of engaging users that have not consumed content in the last 7 days.
- Decision Tree – A process flow determining further actions to take once the first communication has occurred. Any number of steps, as well as follow-on steps, can be configured, based on the user’s engagement with the previous messages.
Measure Effectiveness and Re-target
All data derived from campaigns is gathered back into the user profile. This data can, in turn, be used to further refine audience engagements, where an audience is re-targeted based on this response data.
A native advertising module allows you to place relevant ads to the consumer, instead of blanket DSP advertising, which is used as filler. The ads can be targeted to relevant end-consumers based on audience segments, for maximum effect. Ad campaigns can embody a combination of image, audio and video ads, with each having its own pricing models (CPC, CPM, CPL, and CPV).
You are able to define zones on their application. Per campaign, the relevant zones are chosen and weighted based on program preferences.
Vimmi’s Recommendation Engine
Vimmi’s recommendation mechanism provides its recommended content according to three main categories:
- Most Trended – those movies, series and programs shows that gets the highest viewership ratings, transactions and ordering.
- Promoted Content/Editor’s Choice – content items that are marked manually by the service editor as promoted content.
- Profile Based – content items that will be promoted to a certain group of users with the same consumption patterns
Data Collection for Profiles Creation
The creation of users’ profiles is based on data that is continuously collected on the users’ consumption and general activities in the system. Vimmi collects the data from three main sources:
- The user activity in the app – including buttons that the user had pressed, trailer watched, channels watched, viewing duration, spending time on each of the app’s pages, and other parameters.
- Consumption data which is originated from Vimmi’s AMS backend system. This includes transaction data, such as: top movies that were ordered, top SVOD and channel packages that were subscribed.
- Content metadata that comes from Vimmi’s CMS, such as tags, content properties and attributes keywords in the content synopsis etc.
The data is being stored at Vimmi’s data warehouse for future processing.
Viewers’ Profile Consumption Criteria
Vimmi uses numerous pre-defined users’ profile categories for the immediate use of the service provider. User profiles can be defined by composing any type of data query, however the most used queries are based on the following criteria:
- Consumers of Specific Genre (such as sport, drama, comedy, action, reality, etc.) – for example: a profile of “Sport Fans” can be defined as the users who watch sport content for more than a total of 30 hours per month, and/or subscribers those who are subscribed to an extended sport channel package in the service.
- Specific Content Consumers – e.g. users who watched all episodes of a specific series in less than a month.
- Time of Day – users who consume content at a certain period of time in the day. E.g. morning viewers, or late-night viewers.
The creation of the profiles will be done with Vimmi PS. For each profile, the administrator will be asked to provide the profile’s name, profile’s description and the actual profile criteria. These profiles will be then stores in Vimmi’s AMS DB.
GUI for the creation and amendment of profiles will be provided in a future version of the AMS.
Creating Recommendation Item Lists for Profiles
For any created profile, Vimmi’s basic recommendation mechanism defines which type of content will be recommended. For example, consumers which are associated to the “Football Lovers” profile, will be get recommendations on content items (VOD and programs) that includes the “football”, or “soccer” tags.
For the sake of performance, the list of recommended items for each profile will be cached in the AMS, in order to reduce the number of queries that will be executed in the AMS DB. Cache invalidation will be performed every once in a while, according to system configuration.
Associating Users to Profiles
Once a profile has been defined, Vimmi’s recommendation mechanism will associate to it all the users who follow its criteria definition. The association will be triggered in the following scenarios:
- Whenever a new profile is created
- Whenever a user performs a login
- Whenever a user proactively asks for recommendations
- Every X hours, when X is customizable
In such a way, the AMS will know for each user, what are the most updated users’ profiles that he/she is associated with, and the retrieval of recommendation will be immediate.