NONE: ONLINE-ADS>> DEBATE: 4/14/98 - measuring impressions, digest # 02
ONLINE-ADS>> DEBATE: 4/14/98 - measuring impressions, digest # 02
richard_at_tenagra.com
Tue, 14 Apr 1998 12:09:38 -0500 (CDT)
How should our industry measure an impression?
4/14/98, digest # 02
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
This is special "debate" mailing from the Online Advertising
Discussion List. Anyone on the list is free to send in a response
to be considered for inclusion, however the core debate will be
among 16 industry experts on this subject.
You will receive this mailing in addition to your normal
Online Ads posts/digests/reports.
web site supporting list: http://www.o-a.com/
archive: http://www.o-a.com/archive/
moderator: Richard Hoy
posting address: online-ads_at_o-a.com
subscribe/unsubscribe instructions are at the end of this message
========================================================================
------------------------------------------------------------------------
This week's Online Advertising Discussion List sponsors:
ICONOCAST and ValueClick
"The quality of ICONOCAST, the brevity and frequency is absolutely
spot on!" - Marc Phillips, Director, www.aptstrategies.com.au
E-mail majordomo_at_iconocast.com with "subscribe iconocast" in body
---
Got unsold ad Inventory??? We've got advertisers!!!
ValueClick is your hands-off solution
http://www.valueclick.com/host
------------------------------------------------------------------------
========================================================================
Contents
More Position Statements:
1.) Keith Pieper - MatchLogic
2.) Kathleen Bagley - Accipiter, Inc.
3.) Tom Wentworth - Narrative Communications
4.) Brad Aronson - i-frontier
5.) Neil Monnens - WebRep
Panelists' Responses:
6.) Dave Zinman - AdKnowledge, Inc.
Online Ads List Responses:
7.) Maarten Van Laere - Ad measurement in Europe
8.) Nancy Ging - Counter-points to I/PRO and AdKnowledge
9.) Patric Miller - Paid only for the impressions that are sold
10.) David Prager - I/PRO's position and privacy, ethics
11.) Ulrich Voss - German solution to the problem
------------------------------------------------------------------------
------------------------------------------------------------------------
<MODERATOR's NOTE:>
I messed up and didn't send MatchLogic's info/position
statement yesterday. They turned it in on time, but I
accidentally overlooked it.
It is the first one today. thx, richard
</MODERATOR's NOTE:>
Keith Pieper
Idea Architect
MatchLogic
kpieper_at_matchlogic.com
Keith is involved in strategic product planning and market
development for MatchLogic. He writes a column published
every Thursday on ClickZ.com, covering a broad range of
internet marketing topics, has been published in DM News and
his work has been featured on News.com. Prior to MatchLogic,
Keith has worked with clients like Sun Microsystems and DLJ
Direct, and was involved in growing and expanding the
AmeriTrade, AccuTrade, eBroker and Ceres Securities internet
brands online prior to their consolidation.
MatchLogic, founded in 1996, offers integrated solutions
that allow advertisers and their agencies to quickly and
easily orchestrate complex, integrated Internet advertising
efforts. MatchLogic's advanced systems and expert staff
enables advertisers and agencies to achieve rapid and
sustainable performance improvement in digital advertising
and direct marketing. Using a sophisticated, distributed
network architecture and TrueCount(SM) technology,
MatchLogic provides the most comprehensive centralized ad
management services available.
MatchLogic's position statement:
MatchLogic believes that an ad impression should be measured
when a user's browser resolves and displays an advertisement
on a user's screen.
-----------------------------POST NUMBER 2
Kathleen Bagley
Strategic Relationships
Accipiter, Inc.
kbagley_at_accipiter.com
Bottomline, we support the IAB/CASIE standards and believe
that sites should use the lowest common denominator
available to all. An impression should be counted when the
ad delivery agent/engine/responder (ad delivery responder)
successfully responds to an ad request. The industry needs
to propose a counting standard that satisfies the following
conditions:
1. Easy to audit by 3rd Parties
Sites can implement an API or "call-out" at this level that
can hook in to any auditing service's technology.
2. Accurate and Reasonable to implement
It makes sense to record an impression when the ad delivery
responder has successfully responded to an ad request. This
method allows sites to make only one simple change at the ad
delivery responder level as opposed to modifying multiple
"client" components. We support counting at this level as
it is the best way all sites and ad delivery tools can
participate on an even playing field.
3. Provide comparable data across sites that is acceptable
to advertisers and agencies
If the metric is "successful responses to ad requests," data
will be comparable across sites regardless of ad delivery
product. Although this metric is straightforward and clear,
it remains unclear whether advertisers and agencies will
readily accept this proposal. As an industry, we must be
certain that advertisers and agencies play an active role in
approving any proposed counting standard. Ultimately, these
standards are necessary in order to satisfy the demands of
the advertisers and agencies and will facilitate the growth
of advertising and commerce on the web.
-----------------------------POST NUMBER 3
Tom Wentworth
Product Manager
Narrative Communications
Tom_Wentworth_at_Narrative.com
This is a very interesting question, especially when you
consider measuring content types other than .GIFs, such as
Enliven, Shockwave, and Java. Everyone knows that measuring
impressions from an animated .GIF is difficult because of
proxy servers and caching. Thankfully, other media types
don't necessarily share this limitiation.
With Enliven, we measure three numbers: deliveries,
impressions, and partial impressions.
- A delivery is defined as the number of times the Enliven
Java player was requested. This number is similiar to what
is being measured with .GIF banners today.
- We define an impression as the number of times the banner
appears on a page, cached or not. Each time an Enliven
banner appears on a page, it sends data back to the Enliven
server that indicates an impression was received. Because
of caching, we find that the impressions number is usually
much larger than the deliveries number.
-A partial impression occurs when there is a delivery
without an impression. This can happen when someone leaves
the page before the banner appears on the screen.
The challenge is deciding which number to use when analyzing
the results of a campaign. Click through rates drop
substantially when counting cached impressions. We have
seen campaigns that have had better than 50% click through
when based on deliveries and 10% based on impressions just
to give you an idea of how widespead caching. As "cache
busting" is implemented across a wider variety of content
types we will see a dramatic drop in click through rates.
Is this bad for the industry? Not in my opinion ... as long
as a there is an IAB standard so we can compare campaign
results equally across all content types.
-----------------------------POST NUMBER 4
Brad Aronson
President
i-frontier
brad_at_i-frontier.com
As a definition for an impression, I'd choose when 100
percent of the ad is transferred as well as cached ad
impressions. One problem is that the 100 percent download
can't be measured accurately by some server software (so
folks would need to switch). However, this is what I want to
purchase. I expect to pay for the ad that is downloaded
(spiders and bots eliminated from the count when possible).
I don't want to pay:
1) When a page download is aborted by the user after only a
portion of the ad is downloaded.
2) When a page is downloaded but the ad is not.
3) When people have turned off their graphics. (There is
some value to the alt-text, but I'm not paying for a text ad
-- unless I buy an email ad.)
The challenge is that we need everyone to agree to an
industry wide definition. It's difficult to compare the
different numbers we currently receive from sites:
impressions, ad requests, ad views, etc. What our agency and
our clients want is comparability; comparability will
increase advertiser trust and spending levels.
-----------------------------POST NUMBER 5
Neil Monnens
President
WebRep
nmonnens_at_webrep.net
Of the two extremes, the advertiser only wants to pay for a
fully downloaded and viewed ad. The publisher wants to
charge for the page requested.
A possible middle ground with a little twist.
The publisher establishes a recommended file size and format
(8k .gif as an example). If the page is viewed long enough
for an 8k .gif ad to be viewed, the ad is counted as an ad
impression. If the advertiser provided a 10k ad and only
90% loaded in the 8k time frame, then the advertiser loses
10% of the impressions. If an advertiser chooses to use a
third party to serve the ad, the ad impression should also
count when an 8k .gif ad would have loaded. This also holds
true if the advertiser provides a Shockwave, RealAudio or
other non-.gif ad.
The publisher can choose to have a higher or lower file size
limit as well as choose they file types they offer.
-----------------------------POST NUMBER 6
Date: Tue, 14 Apr 1998 00:47:12 -0700
From: David Zinman <dzinman_at_adknowledge.com>
Yesterday, I made the case that the goal of measuring ad
impressions on the web should be to get as close to the true
number of times that an ad shows up on a viewer's screen as
possible. I was glad to see support for this position from
Tom Hespos and Jim Meskauskas, two savvy media buyers. I
promised in my last post that I would describe exactly what
method for counting impressions should be used, and that is
the subject of this post.
To get as close to counting the actual number of times an ad
shows up on a computer screen, we should try and count the
VERY LAST THING a server does when it delivers an ad to a
user. An impression should be recorded whenever an ad
graphic is "sent out on the wire" by a server. This number
gets much closer to the actual number of impressions than
the IAB's current recommendation. In addition, many
commercial ad servers and outsourced services support this
method. I was pleased to see that Tom Shields recommended
this method (he described it as being in the "ad download"
family) in his post yesterday.
When the IAB made their recommendation to count an ad
insertion (when the ad tag is inserted into an html page),
they made the case that ad impressions should not be counted
because there are caching issues. Nearly all of these same
issues apply to the counting of page requests, however. And
there are many ways to defeat caching when counting an
impression.
Since counting ad impressions instead of ad insertions
would, on average, reduce the overall inventory of most
sites, it is unreasonable to expect the IAB to adopt this
tighter standard without strong input from marketers and
agencies. The IAB WILL take the view of marketers into
account, however. After all, it is the budget of the
marketer which can make online publishing profitable, and
content free for surfers. We all must keep the best
interests of the marketer in mind when developing these
standards. If we do not, they will cease to support this
medium, and that it not good for anyone.
Tomorrow: some technical details on how to count an ad
impression.
-----------------------------POST NUMBER 7
Date: Mon, 13 Apr 1998 21:30:46 +0200
From: "Maarten" <maarten_at_vademecum.be>
As a Belgian small sized company, we have developed an
audience measurement system. In Europe ( yes we do exist )
we are lagging behind but apparently we have the same
problems as you do. However our company decided to
implement the CASIE standards allowing for comparability if
everyone doing measurement uses the same definitions. Those
standards should be adopted by everyone. A commission is
going to be set-up in Europe to define what has already been
defined in the US throught the CASIE standards :-)
Another problem is: has the full ad been shown? There are
several possibilities to do so but I think that the
following is the easiest to implement: a Java pushed ad. A
simple command towards the server would indicate that the
client has successfully loaded the complete add (only a few
lines - fast download) and thus we could report on this is.
A complete study has been done by our company on this item.
If we compare what we can do with traditional media, we are
much further but a lot of challenges stay ahead in front of
us.
Maarten Van Laere
VADEMECUM gcv
VOICE:+ 32 16 29 02 99
FAX:+ 32 16 49 04 37
INTERNET:maartenn_at_vademecum.be
URL:http://www.vademecum.be
-----------------------------POST NUMBER 8
Date: Mon, 13 Apr 1998 13:43:05 -0700
From: Nancy Ging <turtle_at_whidbey.net>
A couple of responses to messages in the first debate digest.
Dave Barlin of I/PRO said:
>What
> advertisers really want to know with their impressions
> measurement - and what they've never been able to get in any
> medium - is "how many times was my ad actually seen by the
> end user?" The only way that we'll be able to get that is
> by moving to client based measurement.
When you talk about client-based measurement, remember that
you're talking about using my machine to do your work. I
have a fundamental problem with that idea. My feelings about
it are something akin to *taxation without representation*
-- i.e., pretty negative. No matter how important your
numbers are to you (and I sympathize with how important the
numbers are), I don't think that gives anyone the ethical
right to impose their methodology without clear information
and choices on the machine owner's part.
David Zinman of AdKnowledge said:
> Shouldn't the Web be MORE accurate than
> TV?
How could it possibly be? TV is a one-way medium, where
output is totally controlled by the broadcasters. Viewers
are either watching TV or they aren't. But the Web is
interactive. As a viewer I have a lot of control over what
appears on my screen. That's why I (and so many others) LIKE
the web better than TV. I like all the choices.
IMO, anyone who attempts to impose on the viewers' freedom
of choice (whether choice of how to use their machine or
choice of viewing practices) can't succeed. I think it
reflects a lack of understanding of the medium.
Do you remember the optical illusion that either looks like
a vase or looks like two faces? When you can't see one or
the other at first, and then suddenly you do, there's sudden
shift in perception -- in how you understand the picture. I
think understanding the Web involves the same kind of shift.
I think the fundamental concepts are viewer control and
unimpeded freedom to communicate globally. Any advertising
measurement methodology needs to take this into account and
work *with* the medium, not against it.
Respectfully, (and thanks for the forum, Richard!)
Nancy Ging
turtle_at_camano.net
Turtle Island Enterprises
http://turtle.camano.net/
-----------------------------POST NUMBER 9
Date: Mon, 13 Apr 1998 14:41:30 -0700
From: Patric Miller <pmiller_at_kwebradio.com>
The timeliness of this question could not be better for my
company, as we are struggling to launch a solution for those
looking for regional/local online ad solutions. As a
relative newcomer to this media, but a longtime buyer and
supplier of electronic, print and outdoor media, I am
surprised by the online industry's inability to find
consensus on the question of "how should our industry
measure an impression."
I feel the solution is simple. As a web server/broker
selling impressions to advertisers, if the web ad was
served, loaded properly and was reported to the accounting
software, it is called an impression, and is billable. In
regards to the cache question, I do not understand why it is
even an issue. In electronic media, there is no guarantee
that the commercial you have aired was not heard by a
significant portion of the audience that may have already
heard/seen the spot. With outdoor advertising, the sale is
based on the number of estimated "drive-by's" as it relates
to total potential viewers. Only in print media, do we
have truly measurable figure based on circulation. But
there are NO guarantees in any of these media that the
person actually heard/saw/read the ad. Even when an ad IS
seen, the industry standard for # of impressions before the
potential customer reacts is between 6-10. What I am
getting at here, is that those impressions, cached or not,
are valuable! As long as the price is agreed upon, and
reflects the fact that some of the customers may be cached
(again this can be a good thing when it comes to
site/product/subject loyalty) there should be no complaints.
It also seems to me that the site should be paid only for
the impressions that are sold, eliminating a lot of "minimum
payment" and "counter tampering adjustment" language in the
contracts. If the impressions are marketed, delivered and
paid for, no matter what the number of hits, the site should
be paid. Click-throughs should warrant a higher rate, as
this is a customer who didn't just "drive by" but stopped to
"drive in." After that, as in any retail model, the rest is
up to the advertiser to keep them there and get their
business.
With ad serving and accounting software becoming more
sophisticated and cost effective, the whole question should
become moot.
Patric Miller
President- Kweb Radio.com
Access Northwest
pmiller_at_kwebradio.com
-----------------------------POST NUMBER 10
Date: Mon, 13 Apr 1998 21:37:27 -0400 (EDT)
From: David Prager <dsprager_at_mailbox.syr.edu>
Dave Barlin of I/PRO said:
> In summary, I/PRO believes we should use ad requests in the
> short term to achieve comparability, but should move to
> client based measurement asap.
At this point, the debate moves away from the concept of
what is or is not to be considered an impression, into a
question of ethics, privacy and online rights.
Mr. Barlin argues (and accurately so) that client based
measurement would provide perfect data as to what users are
looking at what ads where and when. Without a doubt, that
is true. However, imagine the ramifications of such a
system - suddenly, all the advertisers know where you've
been and what you've looked at. So much for privacy. Or
anonymity. A system like that would give essentially anyone
(including businesses, governments, and evil hackers) the
knowledge of who you are and what you do online.
At times like this, even advertisers have to weigh profit
against social ramifications. A system of client based
measurement would change the entire structure of the
Internet because it would eliminate the privacy and
anonymity aspects that are essential to the free flow of
information. How many of us would voluntarily agree to have
all of our online actions monitored, regardless of who is
doing monitoring or why?
There is no argument that a client-based system of
measurement would provide advertisers with perfect
information about the behavior of the consumer and how the
consumer responds to advertising. However, such a system
would come at such costs to the rights of the individual
consumer that it might do more to deter Internet usage than
it could ever do to aid Internet commerce.
Ultimately, it is a question of profit versus ethics. Even
though this list typically provides neither the time nor the
place for such a debate, it is important that we keep such
issues in mind. Our actions aren't always limited to the
scope of a single company's bottom line - what we do can
affect other people, and that is something we need to
remember.
Dave Prager
General Manager
Orange Source - source.syr.edu
dave_at_source.syr.edu
-----------------------------POST NUMBER 11
Date: Tue, 14 Apr 1998 17:31:39 +0000
From: "Boersenspielteam" <boersenspiel_at_vocalweb.de>
Hi all you US guys :-)
may I add, how add impressions are counted in Germany?
Yeah, and this is standarized (SP?), we Germans are great at
this ;-)
Every page with ads contains a 1*1 pixel gif, which is sent
by the server in an uncacheable mode (pragma no-cache,
etc.). And what we count is exactly this gif, not the page
or the banner.
Pros:
* You get an idea of cached pages, pageviews jump 20-40
percent
* You don't have to make pages or banners uncacheable, the
small gif is enough (saves bandwidth)
* technically simple (compared to javascript or java
solution), uses standards (HTTP)
Cons:
* You need a modified web server
* You don't get people with images off (but they don't view
the banner, so is this really a con?)
* Some java(script) solutions promise to be more accurate,
but I doubt, that this is really true with so many people
turning off java(script), bowsing with old browsers and
sitting behind firewalls in corporate networks.
* You can't tell, if the ad is completly downloaded
Of course click-through-ratios drop, but who buys a ratio?
You buy a certain number of click-throughs. And of course I
think, cached impressions should be counted. You count the
number of readers of newspapers, not the number of buyers
(circulation).
This is a really simple solution and gives complete
comparability between sites, that use this technique.
I wonder why the rest of the world completly ignores this
ready-to-install solution ... Just discussions going on and
on...
regards
Ulrich
Ulrich Voss \ \ / /__ / ___|__ _| |
VoCal web publishing \ \ / / _ \| | / _` | |
voss_at_vocalweb.de \ V / (_) | |__| (_| | |
http://www.vocalweb.de \_/ \___/ \____\__,_|_|
Tel: (++49) 203-306-1560 web publishing
========================================================================
------------------------------------------------------------------------
This week's Online Advertising Discussion List sponsors:
ICONOCAST and ValueClick
"The quality of ICONOCAST, the brevity and frequency is absolutely
spot on!" - Marc Phillips, Director, www.aptstrategies.com.au
E-mail majordomo_at_iconocast.com with "subscribe iconocast" in body
---
Got unsold ad Inventory??? We've got advertisers!!!
ValueClick is your hands-off solution
http://www.valueclick.com/host
------------------------------------------------------------------------
========================================================================
SUBSCRIBE INSTRUCTIONS
To subscribe to the Online Advertising Discussion List,
Fill out the form at: http://www.o-a.com/
UNSUBSCRIBE INSTRUCTIONS
You may leave the Online Advertising Discussion List at
any time by sending an email message with the word UNSUBSCRIBE
(just the word) in the body or content of your message to:
online-ads-digest-request_at_o-a.com
if you get the digest version, and
online-ads-request_at_o-a.com
if you get the non-digest version.
Online Advertising Discussion List
http://www.o-a.com/
a publication of The Tenagra Corporation
http://www.tenagra.com/
