|
|
|
Thirty years of rapid change
Immetrica, Inc. has been building audience research systems for three decades. We’ve served major television, radio, DBS and DVR players around the world. During this time, television has been changing dramatically:
Terrestrial broadcasting has been progressively losing audience to a burgeoning multiplicity of pay-TV-delivered channels, VOD and OTT (“over-the-top”, “third-screen” mobile and Web) services. In a phenomenon with little precedent, several major terrestrial networks have failed or were forced to go dramatically downmarket. DVR technology rose rapidly but then suddenly reached a plateau, ostensibly in favour of VOD and OTT The emergence of cord-cutters (especially among younger viewers) who do not use conventional television at all but only OTT The emergence of near-cord-cutters, people who haven’t “cut the cord” but watch only through DVR, broadband VOD and OTT, thought to remain pay-TV subscribers largely to access sports and premium-channel content not available to nonsubscribers, which is increasingly becoming available on standalone OTT as its producers seek greater monetization The increasing dominance of U.S. and, to a lesser extent, U.K. content of the multichannel programming in non-English-dominant countries, using price to offset irrelevance At the same time, RPD audience measurement became possible, offering realistic, sustainable competition to conventional TAM monopolies by offering much larger samples (thus permitting the measurement of smaller channels, and the separate reporting of cobranded feeds and SD v. HD feeds) at a much lower cost
Immetrica capabilities
We have substantial experience in the technical aspects of audience measurement systems, including a decade of direct participation in the design and development of such systems based on return path data (RPD) and the construction of entire such systems between the data collection and client delivery components.
Immetrica is the only company that offers such intermediate processing systems independently of the delivery and marketing of the measurement product, providing better quality especially in the RPD settings, transparency of the design, and control by the client, which permits it to give adequate client service and to change delivery-platform and marketing partners at will. Immetrica is also much more affordable than the large TAM providers.
Immetrica also conducts end-to-end audits of audience measurement systems, both in combination with design and development and as a standalone service.
We also offer a unique approach to reporting systems, used when the delivery software cannot produce the view or aggregation of the data that clients actually desire. Our rapid-development technology for such systems connects to Web-based, RDBMS and file-based data sources and let users see meaningful analysis earlier than conventional IT practices can deliver a nonfunctioning mockup. These are full-fledged reports, not dashboards. For one client, we produce more than 2000 reports monthly, each estimated to require 80 person-hours if composed manually. Our automated systems do this in two days.
For 17 years prior to founding Immetrica, the firm’s principal developed downstream processing components and conducted interpretative analysis for one of the largest programme producers and presenters in the U.S., acquiring a client’s view of audience measurement. The firm represents a progression from criticism of the work of syndicated TAM providers to an opportunity to do a better job.
Design and development of audience measurement systems
Essential elements
To claim accuracy, any audience-measurement service must meet certain industry-standards:
Define the intab, the crucial denominator of all rating and share calculations. The intab applies to the entire broadcast day. The intab should include STBs that were available for measurement during the broadcast day , even if they reported no viewing, and should exclude STBs that were unavailable (that have lost data) during some of that day even if they reported some viewing during it. At the moment the data are obtained from the STB for a given broadcast day, the dataset on that day must be complete. Tuning-without-viewing is a problem for conventional TAM as well, as some viewers leave both STB and monitor at full power while not paying attention, leaving the room or even going to sleep or to work. Different TAM systems try to limit it to differing extents and using differing approaches. But this problem is of much greater concern in RPD measurement because data on the monitor’s state (on/off/standby) are not available. Records of measurable events must be deduplicated prior to loading into delivery platforms. Channel management must identify content channels despite any change in their observable parameters (such as the viewer channel number), and be able to unite multiple instances of the same content channel appearing on the same or multiple included sources (such as different pay-TV systems or a given operator’s CATV and IPTV systems). Where a sample rather than a census is used, the sample should be continuously weighted (the sample should be balanced) to adjust its representativeness to that of the represented universe.
What not to do
Contrary to an understandable superficial impression, one cannot simply route raw audience-measurement data from the data collection system to the delivery system. The result would be utter rubbish:
The intab could not be computed but only roughly approximated at best (and often not even that). That is because cooperation (the television set’s participation in measurement), determined from measurement log timeframes, heartbeat events, or call records, would not be considered. Data collection pipelines that rely on telephony are especially severely affected, as they tend to lose substantial data to failed phone calls (noisy lines, inadequate infrastructure) and phone network outages. If the data collection system reports points in time (such as a channel change) rather than complete events (such as a viewing event to the channel, with start and end times), the last event of the day would necessarily be discarded because its duration could not be determined—or even the state of the set at the end of the broadcast day. The faster data are published after the broadcast day, the more severe the resulting errors. Tuning-without-viewing is of much greater concern in RPD measurement because data on the monitor’s state (on/off/standby) are not available, and many users leave their STBs on at full power. This requires viewing events to be capped, and the capping to be managed even when the capped event lasts across broadcast-day and data collection and publication cycle boundaries. Audience-measurement delivery platforms cannot handle duplicate records loaded in a different load. However, submitting raw collected data directly to them almost guarantees such duplicates. The least complexity in channel management could make an RPD service error-prone or impossible. A single pay-TV operator with a limited selection of channels, few daily activations/deactivations, one feed per channel, no mirrors, either permanently assigned viewer channel numbers or a permanent unique content-channel identifier, can theoretically route its data around the audience measurement pipeline. However, many pay-TV systems are not that simple, and any audience-measurement service that aggregates data from multiple pay-TV providers (as a conventional TAM monopolist does) by definition is not. Where a sample rather than a census is used (i.e., in all cases where the pay-TV delivery network does not contain an integrated broadband return path over which a large percentage of universe STBs can and do send audience-measurement data), the sample should be periodically (preferably daily) weighted (the sample should be balanced) to adjust its representativeness to that of the represented universe. Although this is possible to do without integrating the sample balancing into the data pipeline from collection to delivery, there is no benefit to doing so.
These are only the most serious consequences of routing data collection directly to the delivery platform. The results would look completely outlandish.
Intermediate processing system
Therefore, an intermediate processing system must exist in any conventional-TAM or RPD service, even if it is never acknowledged as a separate stage in the pipeline from collection to publication.
In RPD, the intermediate processing system is typically developed by the delivery-platform provider and contains components that are handled improperly (e.g., intab definition), opaquely (capping), or not at all (channel management, weighting). The delivery-platform providers, most of which are conventional-TAM operators, generally do not understand RPD-specific challenges, keep development to a minimum, and hide weaknesses behind “proprietary” algorithms. A pay-TV operator who permits this to happen will be unable to respond to clients’ complaints, because it does not understand how data are processed and partly is kept in the dark as to proprietary components. More importantly, the pay-TV operator would find it very difficult to change delivery-platform providers, as the published data would change radically and for reasons that it could not adequately explain. The delivery-platform provider would naturally leverage its power over the pay-TV operator for ever greater financial demands.
Immetrica solves these problems:
Immetrica develops the intermediate processing system with specific attention to the idiosyncrasies of the client’s technological and commercial specifics. The client becomes an owner of the system software Immetrica documents all design and implementation, maintains transparency, and works with client staff to understand the system so that the client could take on as much of the support burden as it wishes. Service to users of the audience-measurement service can be provided by Immetrica, the client, or both. Immetrica can operate the system or at any time duly agreed, turn it over to the client. Immetrica can work with any delivery-platform provider to produce the necessary files in the required formats. Immetrica audits the system and prepares it for accreditation or certification, such by as the Ernst & Young practice group that typically conducts accreditation audits for the U.S. Media Rating Council.
As part of the development, we:
Determine the intab using a correct and industry-accepted algorithm, adapted to the technical specifics of the data collection system. Determine cooperation, adding to the intab fully cooperating STBs even if they report no viewing for that broadcast day, and deleting from the intab STBs that returned viewing data but did not return data to cover most of the broadcast day. Manage the crediting of any late-received data per the rules of publication. Cap tuning-without-viewing as accurately as possible and sensitively to different programming types, channels, viewers; maintain capping across broadcast-day and data cycle boundaries. Ensure that duplicate records are not loaded into the delivery platform, and that where updates are needed, the old records are deleted in favour of the new ones. Perform channel management. Perform sample balancing, and adjust HD and simulcast SD channels’ ratings for any substantial nonrepresentativeness of the SD/HD distribution of the sample relative to the universe. Perform other contextualisation, especially if it cannot be performed separately from the audience-measurement pipeline, as sample management often cannot. Produce channel-coverage targets for premium channels, enabling users to see such channels’ ratings among their subscribing households rather than the entire universe; this is a unique capability not offered by conventional TAM and crucial to securing premium-channel operators as clients. Produce other demographic targets.
RPD’s unique advantages
Correctly implemented, RPD can deliver advantages that conventional TAM cannot:
Larger (often much larger) intab sizes, and thus greater stability. Thanks to those large intabs, the ability to identify viewing on less popular channels which in conventional TAM is obscured by a lower signal-to-noise ratio (greater volatility). Thanks to the large intabs, separate cobranded feeds and SD feeds from HD ones. Offer channel coverage ratings, which most conventional-TAM providers cannot.
Not surprisingly, operators of premium channels, those that in conventional TAM are lumped together, and minor channels are among RPD services’ earliest adopters and best customers.
|
