Monday, June 20, 2011

Google To FCC: Stop Letting The Voice Network Tail Wag The Internet Dog

Posted at TechCrunch: 20 Jun 2011 08:27 AM PDT

The Internet carries nearly 160 times more traffic than voice networks in North America, yet many of the regulations and inter-carrier traffic fees are based on the quickly receding era when voice networks ruled. Google calls this the "Tail Wagging The Dog" in a letter to the FCC (embedded below) urging them not to impose antiquated per-minute voice traffic fees on IP networks. This is becoming an issue as IP voice traffic approaches that of traditional circuit-switched voice traffic. Google's lawyers write in their letter:

Standalone voice traffic already is decreasing markedly relative to other forms of communications traffic; in fact, as depicted in the attached, the majority of voice traffic will be IP-based in just a few years. Accordingly, the FCC should not allow what amounts to the very small tail of legacy voice wireline services to wag the very large dog of all communications traffic exchange. In particular, per-minute voice traffic origination and termination charges are a persistent but unwelcome relic from the circuit-switched telephony era, and not best-suited for modern IP traffic and networks.

Google illustrates the changing nature of the network in a series of dramatic slides. Back in 1997, U.S. Internet traffic was only 3,300 terabytes per month, compared to 54,000 terabytes per month for the voice network. Three years later in 2000, voice traffic peaked at 66,000 terabytes per month, while Internet traffic had grown more than eightfold to 28,000 terabytes oer month.

By 2005, consumer IP traffic had reached 669,000 terabytes per month (with 2 terabytes of that being IP voice traffic), while voice traffic had shrunk to 48,000 terabytes per month.

In 2010, consumer IP traffic in North America completely dwarfed voice traffic with 5.7 terabytes per month versus 36,000 for the aging voice network. What's more, IP voice traffic (Skype, Google Voice, etc.) accounted for 21,000 terabytes per month, or nearly 60 percent of what was going over the old switched network.

All of these networks, both data and voice, are going to IP networks. By 2015, Google estimates that consumer IP traffic in North America will more than triple again to 19.4 million terabytes per month, whereas the voice network will shrink further to 26,000 terabytes per month. And IP voice traffic will be almost as big at 21,000 terabytes per month (See top slide). The entire letter with all the slides is embedded below.

Friday, May 20, 2011

As Online Audience Buying Turns Into a Sea of TLAs (Three-Letter Acronyms)

Move Over RTBs, the Hot Term of 2011 Is the AMP
By: Kathryn Koegel Published at AdAge: April 04, 2011


THE FORWARD 'I': The Advertising Option Icon

ADVERTISING OPTION ICON: A cute little "i" that can be put on ad
creative so consumers can find out how the data collected about them
is being used, and set various preferences. It's administered by
Evidon and a consortium of trade groups as part of an industry
self-regulation maneuver to keep Washington privacy advocates at bay
and protect the status quo of self-regulation.

AD BLOCKERS: Software enabled through a person's web browser that can
prevent ads from being displayed. Naturally, publishers aren't big
fans, as advertising is the primary revenue source underwriting online
content.

AD EXCHANGES: Wall Street-like commodity trading comes to web
advertising. Publishers designate inventory, buyers can access it. The
idea is that it creates a rational marketplace and automates the
tedious buying process, allowing publishers to set a "floor," or
minimum bid, for what types of ads they will accept, while buyers bid
for varying types of inventory available but rarely know in advance
where those ads will show up. There are pure exchanges where the
inventory is "blinded," and now publishers are in the game. The
Weather Channel has its own private exchange and a consortium of
Gannett, Hearst, The New York Times and the Tribune company run an
exchange through Quadrant One.

AD NETWORKS (and various permutations thereof): Audience Based: They
sell with the idea of aggregating users based on either demography or
intent (to purchase something).
Horizontal: Sells a wide base of inventory available, i.e., not specialized.
Vertical: Specialized. There are women's networks, sports networks,
networks of people in the market to buy a car, etc.
Mobile: Sells ads onto wireless devices including phones and now
tablets. There's more to it than just sales, as it takes work to get
one ad to appear on phones across multiple platforms (e.g., Android
and Apple). There are three big ones: Google (AdMob), Apple (formerly
Quattro) and Millennial (independent), plus some focused on rich media
or various verticals.
Performance: Clicks 'R Us. These networks have a wide range of
inventory available and are all about driving direct response at the
lowest possible price.
Video: The combo of sight, sound and motion is hot (and generates the
highest CPMs). Recently this space has been a land grab for more
generalized networks that are snapping up or merging with video
networks. They don't only aggregate video inventory against which to
run ads -- some also syndicate video content across a range of sites,
as there is not enough video inventory at the right price to meet
demand.
AD SERVERS: The technology that disseminates online ads and then
tracks and reports back on ad performance. DoubleClick/Google and
Microsoft Atlas are the leaders, along with "homegrown" servers --
which is when a site builds its own.

AD VERIFICATION: Software tools that advertisers use to determine if
impressions are displayed in the proper place and whether the ads are
privacy compliant. They are often used with exchange inventory when
the advertiser does not know exactly where the ads are placed or for
audience targeted buys to ensure "brand safety" -- that ads aren't on
pages with, say, boobs. The ad networks typically dislike verification
tools because networks say it's their job to find the right sites, and
they argue verification has technological limits. For example, they
say, if advertising for a woman's product shows up on a breast
cancer-related site, boobs may just be a perfect fit.

AGENCY-TRADING DESKS (ALSO MEDIA-BUYING PLATFORMS/DESKS OR DSPS): The
practice of an agency operating its own DSP so it can enrich its pool
of data from all the buys it executes. The bigger agencies are
building systems to look at data across media and frequency cap, i.e.
control the number of times a person is exposed to specific ad.

ANALYTICS: There are three different kinds:
1. Third-party panel-based companies, such as ComScore and Nielsen,
can tell you who actually looks at a site. The publishers always argue
that they are being undercounted due to the complexity of the web and
the difficulty of getting reliable metrics for smaller sites.
2. First-party software tools, such as Google Analytics, Omniture and
Webtrends, operate through tags on a site's pages and let publishers
get statistics on how many page views a site is generating and how
many unique users it has (a machine proxy for people/viewers). Their
numbers never match up to third-party tools due to issues such as
cookie deletion, and both groups argue about their data reliability.
3. Audience targeting is basically behavioral targeting with new
clothes and broader capabilities and applications. It's the practice
of using data to imply an audience, either by demography, life stages
or some sort of intent, such as people who have searched for info on a
new phone purchase. These analytics tools help marketers buy a
specific audience, not an audience implied by the context of a
specific site. (If I am looking to reach an "auto intender," I
advertise to them wherever they are online -- not just on auto sites.)


THERE'S A SCIENCE TO IT: AudienceScience bills itself as the largest
and most trusted audience aggregator in the world.
AUDIENCE MANAGEMENT PLATFORMS (AMPS): One of the rising terms of 2011.
The folks who provide the audience targeting now have platforms to
automate the process of buying audience-targeted inventory. Since many
of them also operate ad networks, it's a natural extension. Jeff
Hirsch, CEO of Audience Science, which has rebranded as an AMP, said
"AMP is the acronym representing a company that has DMP and DSP
capabilities."

BEACONS: A 1-by-1-pixel tag typically used by an advertiser or a
third-party ad server to track a unique user's activity over time. A
beacon helps to properly attribute an online action to ad exposure,
even if the action happened days after being exposed to the ad.
DoubleClick, which originated the concept, still uses the name
"spotlight tags."

BEHAVIORAL TARGETING: A now old-fashioned term that has been rebranded
as "audience targeting."

BETTER ADVERTISING (now Evidon): The folks bringing you the
Advertising Option Icon.

BLINDING: The practice of not designating where the inventory will be
placed. Exchanges or third-party networks will often "blind" an ad at
the publisher's request. Publishers do not want to create "channel
conflict" -- or a situation where someone besides their sales staff is
selling their inventory. (Never let your customer know someone else
sells the same thing cheaper.)

BLUNT TOOLS: Term referring to a broad-based "spray and pray"
approach, when a campaign doesn't use the available data to target
offers to specific interests of consumers. Amazon and Netflix are the
antithesis of blunt: they use their data to make buying
recommendations based on past purchase behavior that hopefully results
in higher conversions.

CLOUD COMPUTING: The technology that makes all this data connection
and movement work. Many ad exchanges run on a computing cloud created
by AppNexus which operates as an exchange of exchanges.

CLUTTER: Your typical web page, loaded with ads.

CONTEXTUAL TARGETING: There are two kinds -- natural context, where
you place a bank ad on a finance page, and contextual advertising,
which scans the text of a website for keywords and targets
advertisements based on those keywords, such as ads in Gmail that pop
up based on your email content.

COOKIES: There are several kinds.
First-party cookies: What your bank and Netflix use to know it's you
when you come to their sites.
Third-party cookies: What an ad server drops on your browser to
designate that you have shown interest in various product categories
or to place you in a demographic group discerned from your online
activity. These cookies can be used to target advertising.
Flash Cookies: Almost universally publicly derided in the industry
because they're difficult to get off your computer (but sometimes used
for the same reason). To delete them you have to go to Adobe's website
or manually do it through the Flash player setting on each site you
access.
Fresh cookies: Depending on the product category, fresher pieces of
data are better. For example, if I'm in market to buy a cellphone, I
may only shop around for a week or two -- so cookies dropped on me
during the past week or two are more likely to be reliable.
Cookie deletion: People concerned with privacy regularly use their
browsers to "dump their cache" of cookies. If you do it, be careful to
just dump the third-party ones or you will have to reenter your
information at places where you have registered. Cookies also dictate
frequency caps -- if you dump regularly, you are more likely to see
the same ads over and over and over again because the ad server won't
know you've already seen it many times.

CREATIVE OPTIMIZATION: Software systems that enable an advertiser to
default to the highest-performing creative or manage the frequency of
exposure, typically on a direct-response basis.

CPC (COST PER CLICK): All the amorphous branding talk aside,
cost-per-click -- a division of a campaign's cost by the number of ad
clicks it generated -- is still the reality for much online
direct-response buying reality.

DAA (DIGITAL ADVERTISING ALLIANCE): The alliance of a bunch of
acronymous trade orgs pushing for self-regulation: The 4A's, the
American Advertising Federation, Association of National Advertisers,
the Direct Marketing Association, the Interactive Advertising Bureau,
the Network Advertising Initiative and the Council of Better Business
Bureaus. The alliance represents more than 5,000 companies in the
space.

DATA AGGREGATORS: They pull together ad-serving data, conversion data
and third-party data, including those from offline sources (stripped
of personally identifiable info like names and addresses) like Acxiom,
Polk and Experian, to attach as many attributes as possible to online
cookies to refine targeting capabilities.

DMPS (DATA MANAGEMENT PLATFORMS): The hot term of 2011. These
self-service "dashboard" tools perform a range of services from
collecting, managing, segmenting, sharing and analyzing marketers'
advertising data -- and assuring that your data is your data.

DSP (DEMAND SIDE PLATFORM): Last year's hottest term. A computer-based
platform the buy-side (agencies and advertisers) uses to automate
media buying across multiple sources with unified targeting, data,
optimization and reporting. Data is treated like media in that it is
layered across the buy and becomes just another part of the cost. DSPs
do not own, purchase, represent or resell inventory from publishers.
As John Montgomery, chief operating officer of mOne puts it: "A DSP is
simply the plumbing that plugs into real-time inventory sources, such
as real-time exchanges, and participates in a public auction on behalf
of its clients." In contrast, ad networks own or represent inventory
from or on behalf of publishers.

DO NOT TRACK: Do not call killed telemarketing, and so the online
world fears what will happen if Washington institutes do-not-track
online privacy legislation. If there were one simple place where
consumers could opt out of any ad tracking, would they do it? Do they
understand what the implications are for their favorite sites?

ECPM: Effective CPM. Today, most buys are a mix of cost-per-thousand
impressions (CPM) and CPC. The eCPM answers the question: If I buy a
CPC campaign, what would I have paid if I bought it on a cost per
thousand (CPM basis)? It's used to compare whether a CPM or CPC buy
was actually more cost effective. To calculate: Campaign
cost/(impressions delivered/1000).

ENGAGEMENT RATES: What percent of total impressions for an ad were
hovered on, clicked or somehow interacted with? Usually applies to a
rich-media ad. Smart advertisers are taking it to the next step and
asking questions about what impact a "hover" has on ROI or an actual
sale.

EVIDON: (formerly BAA, Better Advertising Project) Selected by the DAA
to operate the advertising option icon self-regulatory program.

FREQUENCY CAPPING: Using cookies to manage the number of times a user
sees a specific ad creative. (Has never been done for most insurance,
teeth whitening or belly-fat ads.)


Noah Berger/Bloomberg
Randall Rothenberg
IAB: Industry trade group that has developed online ad standards and
guidelines, championed the cause of online publishers and the right to
target ads. Run by the respected and forthright Randall Rothenberg.

IMPRESSION: The act of someone seeing an online ad upon calling up a
web page. There is an approved and very technical definition of the
term, which has taken years of negotiation by the IAB. It's available
at iab.net.

INVENTORY: Common term for available ad impressions. For some reason,
the web world likes to talk as if it is moving units of underwear or
canned goods rather than pixels. Various types include:
Content farm: Inventory created because it's highly desirable to
advertisers and souped up through SEO so that people find it. Often
offers context, but can be marginal context.
Direct sold: Inventory bought directly from the publisher, typically
higher-quality stuff.
Contextual: Inventory that relates to a category of content or
advertising. Business pages on The New York Times are contextual
inventory for finance.
Premium/first-tier: As good as it gets, often contextual or highly
trafficked areas such as home pages and lead section pages.
Mid-tier: Inventory from a known site with decent content, but not a
top 100 site as ranked by ComScore.
Long tail: The vast inventory available from small, no-name sites and
blogs. There could be some valuable stuff in there that people really
look at -- or not. The challenge is segmenting it.
Remnant: Every large publisher's got unsold inventory, either simply
because you can't sell it or because of its context. Often gets thrown
into an exchange, where data is appended to turn it into an audience
buy.

OBA: Online behavioral advertising and also a compliance program led
by the IAB that sets standards for data usage in behavioral targeting.

ONE-BY-ONE (1X1) PIXEL TAGGING: What an ad trafficker needs to "drop"
on an ad (a pixel tag is an invisible image, basically) in order to
analyze a view through.

RTB (REAL-TIME BIDDING): The process of buying and selling impressions
instantaneously in which the highest bidder "wins" the right to place
a display ad while audience attributes are affixed to it. It was the
hot thing of 2010 but now that just about all the exchanges can do it,
they're on to the next acronym.

RETARGETING: An example: Say I shop for a car by looking at a couple
of auto sites. Those sites and third-parties drop cookies on me and
the computers and data management companies go into a fury trying to
find me on other sites as I surf the web, serving me auto ads until I
buy. Works brilliantly, but drives the privacy advocates nuts. I,
however, like it when I don't make an online clothing purchase due to
the price and the retailer "comes after me" and gives me a great offer
via an ad on another site. Thanks retargeted ad, you just closed the
deal!

SEMANTIC TARGETING: Targeting in which a computer system examines all
the words on a web page to identify the context of the page, rather
than relying on simple words or phrases. Simple contextual targeting
might see the word "golf" on the page. But semantic targeting can tell
you if it's an alligator-bites-man-in-Florida-on-golf-course story --
something Calloway wants no part of.

SENTIMENT ANALYSIS: Can also be used to determine if the "sentiment"
of the page is positive or negative, using that finding to possibly
suppress an ad (i.e., could help Calloway avoid placing an ad next to
an article about how golf is a lousy sport for overweight white guys
in media).

SOCIAL TARGETING: Uses data to find those obsessive social networkers
who influence the "social graph."

SUPPLY-SIDE PLATFORMS: A DSP but from the publisher or content
producer's perspective.

TLA: Three Letter Acronym. It's what the online ad industry is
becoming, jokes Omar Tawakol, CEO of BlueKai, a data aggregator and
now a DMP.

VIEW THROUGH: A look at the post-ad-impression behavior. If your "view
through" window is 30 days, a person's actions within 30 days of
seeing the ad can count toward the ad's effectiveness. For more, see
beacons.

VCS: The cash and promise of going public or getting bought by Google
or Facebook that fuels it all.

YIELD OPTIMIZATION: Employed by publishers to find out how much their
impressions are worth and how they can you manage flow of inventory to
make the most money? These platforms look at each impression available
on a web-publisher site and then match the impression with an
available ad from an ad network or exchanges.

Sunday, May 01, 2011

The Cloud Has Us All In A Fog

Posted at TechCrunch: 30 Apr 2011 12:05 PM PDT

Ever heard of Dropship? It's an open-source project that "enables
arbitrary, anonymous transfers of files between Dropbox accounts."
Dropbox hopes you haven't; they tried to squelch it this week, and
even accidentally reported that it was subject to a DMCA takedown
notice, with predictably futile results. I'm mostly sympathetic: I'm a
huge fan of their service, Dropship was a clear violation of their
terms, and for obvious reasons they don't want to turn into an
anonymous peer-to-peer file-sharing service. Unfortunately, they
accidentally built a system which enabled just that.

How about Sony's PlayStation Network? Of course you have. It was so
thoroughly hacked this week that Sony had to shut it down
indefinitely. Did you also know that Sony's PS3 firmware is
effectively wide open, because they made a hilariously stupid security
mistake? Did you know that that's probably how PSN got hacked, and
that it raised the spectre of the hacker(s) taking over every
connected PlayStation 3 in the world and turning them into by far the
biggest botnet in history? That probably wasn't what Sony had in mind,
but they accidentally built a system which enabled just that.

How about the new Google Docs Android app? Came out this week, and
it's pretty great. Among its many features is the ability to take a
picture of an image with text and have that text automatically OCRed
and turned into a document. Can't wait 'til they integrate Google
Translate into that, too, and recapitulate last year's hot app World
Lens. But I bet book publishers are pretty unhappy. Not long ago, if
you wanted to scan a book you had to actually build a scanner, or buy
a copy and turn every page. Now would-be book pirates can just
crowdsource 10 people to go to bookstores and take 20 pictures each,
et voila: 400 scanned pages in Google Docs. Easier book piracy
probably isn't what Google had in mind, but they accidentally built a
system which enables just that.

This was also the week that people who keep remotely controllable
Internet-enabled camera/microphone/GPSes on them at all times
expressed outraged surprise when they learned their privacy is at
risk. The panopticon probably isn't what the mobile industry had in
mind, but they accidentally built a system which enables just that.

What do these all have in common? The unexpected results of connecting
client devices to the cloud. (Yeah, I don't really like the term
either, but it's better than the alternatives.) People talk about
"moving to the cloud," as if we haven't already. The heavy lifting may
happen on the server farms (when they're up) but every connected
computer, phone, and game console already serves as a computing
cloud's eye, ear, and tentacle.

Emergent properties. Unintended consequences. Get used to 'em. My
favourite Douglas Adams books are the Dirk Gently novels, in which the
protagonist makes use of "the fundamental interconnectedness of all
things" to solve crimes in hilariously unexpected ways. Now we're
literally building that interconnectedness into (nearly) all things.
So we shouldn't be too surprised to find ourselves moving into a Dirk
Gently future, in which off-kilter left-field ricochet consequences
happen at an ever-increasing rate. You can bet that those cited above
are just the beginning — and that there's a lot of money to be made in
seeing them before they happen.

Friday, April 29, 2011

Who will pay the bands to make the music in the first place?

Published on 27 Apr 2011


The frontman of a group described as Scotland's best band ever has hit
out at the generation who no longer pay for music.

Stuart Murdoch, the Ayrshire-born lead singer of Belle and Sebastian,
said sales have dropped so low they no longer finance the recording of
new material.

His attack was backed by two other top Scots musicians – Alex Kapranos
from Franz Ferdinand and Adele Bethel from Sons of Daughters.

The broadside came as Spotify– a controversial online music library
that offers free music subsided by advertising – cut back on what it
gives users free.

People who do not pay a fee will be allowed to listen to a song only
five times every month, in a gesture to pacify furious record labels.

Spotify has previously been attacked for failing to give musicians
enough cash for their songs.

Using Twitter, the Belle and Sebastian song-writer wrote: "On Spotify
graciously deciding to charge its free listeners – are you going to
pay the bands that make the music in the first place?"

Murdoch warned he could not record new music if sales didn't pick up,
adding: "We'd like to make more songs and used to do that with the
money we made from selling songs."

Belle and Sebastian were named Scotland's number band in a poll
conducted by The List Magazine in 2005.

Singer Kapranos, backed Murdoch, saying: "Spotify exists for moral
hypocrites – those who are too cowardly to steal music, yet too cheap
to pay for it – the kind of person who'd never consider nicking a bag
of crisps from a newsagent or siphoning neighbours' petrol, but who
still expects to lift the music that brings them joy for nothing.

"Either be honest and be a thief in all areas of your life or decide
you don't steal.

"Most people don't want to pay for their music, therefore don't pay
for new music to be made, which is fair enough. But they can't expect
to hear anything other that tunes recorded on a laptop in a bedroom."

Adele Bethel, lead singer with Sons and Daughters, said: "I agree with
Stuart Murdoch on this and am glad he's spoken out about it. The
percentage given to artists is so minuscule.

"If you don't pay for music, it will die out."

Spotify said it hopes to become a way of musicians to earn more money
for their songs and that by offering free music, it encourages kids to
stay away from illegal pirate music.

Jim Butcher, spokesman for Spotify, said: "We're still young and as we
continue to grow, the money we pay back to the industry will grow with
us. "And remember we're monetising a youth demographic, many of whom
might otherwise be downloading illegally and not paying a penny back
to the industry."

Spotify is a legal way of listening to free music. It pays record
labels a fee every time someone listens to a song. Last year it paid
out almost £40million in royalties.

Spotify's chief content officer Ken Parks said: "We've shown that the
model is doing extremely well, but as things stand we need to tweak
the service to ensure everyone has access to legal music in the long
term."

Monday, April 11, 2011

Finding the 'Sweet Spot' for Ads in Your Favorite TV Show

From 'Modern Family' to 'NCIS,' Each Show's Ads Have A Most-Watched Moment
By: Brian Steinberg Published: April 11, 2011 at AdAge


What if each and every show on TV, be it "Modern Family," "Survivor"
or "Biggest Loser," had a "sweet spot" for advertisers -- a particular
moment where commercials running in the program were more effective
and received better by the audiences watching them?

Analyzing second-by-second viewing data provided by TiVo, Interpublic
Group of Cos.' Magna Global says it believes the highest-rated
commercial pods often occur adjacent to particular moments in sitcoms,
dramas and reality programs, bolstering the idea that running a
commercial helter-skelter across a TV network's schedule can actually
reduce the power of the spot.

It's no secret in the ad business that the first and last ads in a
commercial break are often the most watched -- after all, they air
closest to the TV show that drew viewers to the boob tube in the first
place. Yet the Magna Global study suggests attention to commercials
parallels attention paid to the programs they interrupt, where viewers
eager to see the start, conclusion or other specific parts of a show
will give more consideration to the ads placed alongside those
segments.

"This type of measure isn't currency yet, which is kind of why we've
been pushing pod measurement as a way to get more details like this,"
said Brian Hughes, VP-director of audience analysis at Magna Global.
"It helps us buy smarter."

Among the findings of Magna Global's study:

Attention paid to the ads accompanying initial segments of many CBS
procedurals -- including "CSI: Miami," "NCIS" and "Criminal Minds" --
is more substantial than that accorded ad breaks alongside other parts
of the programs. Mr. Hughes suggested CBS's programming is heavily
consumed by homes with digital video recorders, where viewers may want
to sample the opening segment of the show. That's often where a crime
is committed (and, oftentimes, comes with gore or violence) and
viewers are likely trying to see if they like the set-up or if they've
seen the episode in the past, said Mr. Hughes.

The trend can't be extended to all procedural dramas. Magna Global
found viewers of NBC's "Law & Order: SVU" paid the most attention
during the last ad break in the show.

Heavy attention is paid to different moments in different sitcoms.
Viewers of ABC's "Modern Family" and NBC's "30 Rock" paid the most
attention during the last ad breaks in the programs (which often come
before quick, funny segments that run alongside credits), while
viewers of CBS's "Big Bang Theory" paid the most attention to the
first ad break in the half-hour comedy.

"It has to do with the way the shows are sort of pulled" or broken up
by advertising, said Mr. Hughes. "It depends on whether it's a family
comedy or an ensemble cast."

Viewers pay the most attention to the last ad break of many serial
dramas. The highest-rated ad breaks in "Grey's Anatomy," "Desperate
Housewives" on ABC, "Gossip Girl" and "90210" on CW and "Parenthood"
on NBC were always the last one, Magna Global found. Interestingly,
these are the ad breaks that precede segments in which networks
tantalize viewers by offering scenes from the next episode.

"People are really sort of attuned to that moment where they see what
the set up [for next week] or the cliffhanger is going to be," said
Mr. Hughes.

The Magna study has its limitations. TiVo reaches only a portion of
the U.S. populace, and households with DVRs tend to view TV
differently than those without.

Magna believes DVR users can be more important for advertisers to
understand and reach than viewers watching at specific day-and-date
appointments, Mr. Hughes said. DVR users tend to focus more intently
on the show they've recorded and may not multitask. "It's a very
important audience to understand, even though this isn't necessarily
representative of the total," he said.