Considering a New Way to Pay for Ad Serving
By Jim Meskauskas
Do you remember the first time you used a third-party ad server for
an online advertising campaign? How excited were you that you no
longer had to track down reports from ten, twenty, perhaps forty
different sites and then input them all by hand?

Okay, sure, at first, you were still doing that, because ad serving
counts never matched site-reported delivery. In order to reconcile
invoices, you had to check the site's version of things.

But for the most part, third-party ad serving was a godsend. For the
first time, agencies handling multi-site buys for the clients had a
single data source to which they could go and pull reporting for all
the sites at once. The output was more or less easy to read and the
data easy to extract.

As time went on, third-party ad servers became more and more robust
with data collection and mining tools. It isn't just simple
impressions and clicks anymore; we can learn post-click activity and
post-impressions. Visits, registrations, purchases, amount spent...
you name it and the third-party ad server can give it to you.

With the advent of the third-party ad server, agencies firstly
became more efficient and secondly became smarter. Personnel whose
responsibility was to handle client reporting fell on their knees
and wept with joy at the amount of time saved by using a single data
source, while planners and analysts and strategists rejoiced at
having more data they could turn into information for their clients.

And all this could be had for (and is still had for) a small fee; a
CPM, to be precise about it. For anything from 10 cents to $5 I
could serve one-thousand impressions across a vast array of sites.

In the early days, this system seemed to make sense. The argument
was that a CPM pricing structure was scalable, compensated for use
bandwidth, and worked pretty much just like media pricing.

The difficulty was - and still is - that you could never really plan
accurately just how much you were going to need to budget for ad
serving. Site delivery, by and large, is notoriously inconsistent,
meaning that your cost estimates are going to be off. And these
days, many sites with the extra inventory want to look good to the
advertiser, or perhaps yield a lower effective CPM, so they
intentionally over-deliver. Sometimes, a placement is event-
sensitive - maybe you road blocked the weather page and it turned
out there was a huge storm that day, making the area very popular.
With this, again you have more over-delivery. To paraphrase Yoda,
always in motion is the ad serving fee.

I've tried different things to prepare for the unknown of a final ad
serving fee, but nothing works too well. Estimate some percentage of
over-delivery, like 25 percent. Or set aside a contingency budget.

If you overestimated the costs of ad serving, you're typically okay;
it looks like a savings to the client. But if you underestimate,
watch out.

I've been in advertising some ten years, and I have yet to meet a
client that doesn't mind an unfixed cost variable. Typically, when
you tell a client something is going to cost 'X,' then that
something needs to cost 'X.' It is a terrible thing to go to the
client and say, "the good news is, our campaign over-delivered by
100 percent. The bad news is we need another $25,000 in ad serving
monies."

The time has come to change the way we pay for ad serving. Instead
of tiered CPMs based on impression load, we need tiered flat fees.

I think, to serve less than ten million impressions, it should cost
$2,500. Ten to twenty million, it should cost $5,000. Something
along these lines would work best for everyone. For the media
planners and buyers, they would have a fixed cost variable for their
plan. For the ad serving companies, they would make more money on
the low-end impression level bracket that more than makes up for
whatever they might be losing on the high-end of the impression
level bracket by transitioning from CPM to flat fee. It is kind of
a "dollar cost averaging," if you will, across different buys at
different impression levels.

Another good thing about a compensation system like this, is that
agencies will no longer be paying a penalty for being good
negotiators. What I mean by this is, the lower my media CPM is, the
larger percentage ad serving becomes of my media expenditures. I
would rather have my media dollars working for me in media rather
than in the facilitation of media.

This is not an issue of earth shattering importance at this time;
I'm not saying this is a big deal, but it is certainly something to
think about. Ad serving companies are still going to make their
money, and as bandwidth costs next to nothing, that no longer holds
sway as a good reason why advertisers should pay on a CPM basis.
Let's be daring and go with a flat fee.


Online Spin for Thursday, August 12, 2004:
http://www.mediapost.com/dtls_dsp_Spin.cfm?fnl=040812

We welcome and appreciate forwarding of our newsletters in their
entirety or in part with proper attribution.
(c) 2004 MediaPost Communications, 16 W. 19th St., New York, NY
10011