The Tyranny of Third-Party Servers

Let’s say you are a widget manufacturer. You produce X units of widgets and sell them to Retailer Y. Before you load the shipment in the truck, you count the widgets, and sure enough, your shipment is exactly what the customer ordered.

On the other side, Retailer Y receives the shipment and has an outside firm the same firm that placed the order with you counting the widgets. And that firm has a special interest in making sure that Retailer Y gets as many widgets as possible, because that’s the firm’s job.

Well, the widgets are offloaded from the truck, counted, and, by golly, the firm comes up with a count that is 20 percent, maybe even 40 percent, less than what your count was. The firm demands you send it that missing 40 percent and an additional 20 percent to make up for the inconvenience.

Well, mistakes do happen though this one seems to be a whopper and even though you are pretty darned sure you delivered the number of widgets you promised, in the interest of good customer relations, you ship out the widgets that were allegedly not sent.

The following month, another order comes in and gets handled this time with extra care to ensure no screwups and despite your best efforts and the use of some pretty darned reliable technology employed just for this kind of situation, the counting firm calls back and said you short-shipped by 30 to 40 percent again, and would you please reship those products, with the customary bonus of another 20 percent to make up for Retailer Y’s pain and suffering.

And this happens month after month after month after month.

For some reason, your system, despite being highly advanced and designed by some of the finest engineers in the business, never seems to be able to deliver more than 60 to 70 percent of the order. Or is it the other way around? Is it that the counting firm’s system never seems to be able to count more than 70 percent of what it actually received?

One would have to wonder, wouldn’t one?

Well, for those of us who are site publishers, this tale is all too familiar. You know where I’m going with this, don’t you?

You may employ some of the best ad serving technology available DoubleClick DART, AdForce, Engage AdManager, etc. and no matter what you use, somehow, some way, the third-party ad servers (you know who they are) always undercount. Always!

Have you ever heard of or experienced a situation where an agency notified you that its third-party count came out higher than yours? I didn’t think so… I can assure you it will never, ever happen.


Well, perhaps I’m paranoid, but in my heart I believe third-party ad servers are designed to undercount. What they count is real, I’m sure, but what they accept as countable falls into a narrow range. So narrow that many valid impressions are knowingly and intentionally not counted.

I also believe that there is an inherent conflict of interest. These servers are designed to be used by agencies. Undercounts benefit agencies because then they can deliver more impressions to the client than they had to pay for.

There are other problems we’ve experienced with third-party servers. It’s not unusual for agencies to use them to bypass the file-size guidelines on your site and to serve far larger ads than you would normally permit. You lose all control over the ad content being served up on your site.

In short, I have yet to see any advantage to the publisher in permitting the use of third-party servers. They create disputes over numbers, delay bill payment, and foster ill will among parties that should be cooperating, not quibbling.

At the beginning of this year, we banned the use of third-party served ads on ClickZ for precisely those reasons. As time went on, we softened our stand and permitted their use for ad serving, with the provision that the billing is based on our count, not theirs. We don’t want to spin cycles quibbling about ad-impression counts.

It’s not a position I would advocate for many of you because it will cost you business if you do so. Plenty of your competitors will be only too glad to take the third-party served business, despite all the headaches. We’re lucky to be in a niche area with a limited inventory of ad impressions.

What I wonder is this: What would happen to the third-party ad serving technology if both publisher and agency shared the expenses? Would they make a concerted effort to ensure a fair and accurate count was the result? My guess is that the answer would be yes.

In the meantime, what are we as publishers to do?

My recommendation would be to minimize the portion of revenues generated by IAB standard banners as much as possible. I just ran the numbers, and banners only represent 14 percent of ClickZ’s overall online sales year to date. How do you compare?

Banners are a commodity, hence diminishing in value on a daily basis. The value they contribute to the advertiser is subject to reasonable debate, and this medium could stand for more experimentation and use of alternative ad models. By moving to alternative models (you should read some articles I’ve written about alternatives worth considering), you free yourself from the tyranny of the third-party server.

One might ask: “Andy, are you a fair guy? Will you let me write a rebuttal to this column? I think third-party servers are the best thing since sliced bread!”

To which I reply: We have the ClickZ Forum where we can debate this openly. Not only that but, over the past few years, ClickZ columnists Tom Hespos and Jim Meskauskas have been passionate spokesmen for the cause of third-party servers.

Meanwhile, why don’t we publishers take some positive action within our businesses to free ourselves of this burden.

Related reading