Not long ago, I had a fascinating conversation with one of the savvier lead buyer in the online education space. She works for a mid-size school and takes a much more aggressive approach to weeding out undesirable behavior than most others. The discussion ended up being about transparency, and whether we might reach a point in time where the right price gets paid. The short I answer is not yet, but before we get there, let's talk about transparency in the minds of performance marketers, i.e. transparency on spend, the sub id variable.
Part 1 is for those without a lot of day expertise in the performance marketing space, and it explains the reasons and benefits for sub id (used interchangeably with "subid") tracking through a hypothetical example that could occur in the performance marketing space
Sub ID Tracking 101
We always follow the money, so we begin with our
advertiser BillYouLongTime.com. They have come out with the latest craze for
health products, Woo-Me Tea. They offer users a free trial of the
product with enrollment in the auto-ship program. Users sign up for the
trial, enter their credit card number, pay a small processing fee, and
if they don't cancel, they will receive monthly credit card charges
that correspond to regular shipments of the product. The advertiser
works with CPANetwork.com to gain distribution of their product. The
challenge for Woo-Me tea is that in order to get affiliates to run
their product, they have to take some risk with payout. They might
charge the user $6.95 processing, but out of the almost seven dollars,
little of it is profit. Even if they made two dollars profit, no
affiliate would run the offer for two dollars. A percentage of users
who sign up for Woo-Me, will continue with the program, and those that
continue will stay on for a certain period of time. Woo-Me crunches all
that data to come up with an expected value and carves a piece of that
to pay up front for a new sign-up to their trial program. That is why
they will $20, $30, $40 or more for a new acquisition. The important
thing to know is that each user who signs up for the trial, doesn't
necessarily mean guaranteed revenue for Woo-Me. And, it works the same
way for many lead generation offers, like online education. A person
who requests free information about a school might earn the network
$30, but that earned the online school nothing, unless some percent of
those leads turn into enrolled students.
With Woo-Me now signed on to CPANetwork, the network must empower their affiliates. CPANetwork has a link from Woo-Me that resembles http://www.woo-me-tea.com/v1. They add a redirect layer in order to capture information. CPANetwork's affiliates login to their account, see the new Woo-Me offer, browse the available creatives, and select one to run. The affiliate gets a link, and it looks something like at the bare minimum http://www.cpantrack.com/affiliateid=1234&creativeid=5678. When a consumer clicks on the link, it records the data with CPANetwork and sends the user to the Woo-Me page. If a conversion happens, Woo-Me shows a pixel on the thank you page that lets CPANetwork know that a conversion happened and it reads the pixel to capture the affiliate id and creative id so that they know who gets credit. There are two challenges with this, one for the affiliate and one for Woo-Me. We will start with the advertiser. Woo-Me sees a bunch of inbound clicks to their site, but to them all visitors look equal. Imagine if CPANetwork had two publishers - Aye and Bee. Aye's traffic has visitors that when they sign up to Woo-Me stay on for a long-time. Bee's on the other hand, have little interest and a high percentage cancel during the trial phase. Woo-Me makes money on Aye, but loses money on Bee. Combined it looks about average and within their threshold.
All of the sudden Woo-Me might start to lose money overall and threaten to stop the campaign, or they might actually have stopped it and told CPANetwork that they won't work with them again. That's where the notion of subids come into play. Woo-Me now gives a link to CPANetwork that looks like http://www.woo-me-tea.com/v1&subid=. They now understand that different publishers perform differently, and they want to track those separately on their end. They ask CPANetwork to pass in the affiliate id into the subid field, which they can do easily. Armed with that information Woo-Me can tell CPANetwork that publisher 123 (Aye for example) does well but Bee (125) does not and to turn off id 125.
For publishers / affiliates, subid tracking has an equal benefit. The easiest example comes via paid search. Imagine yourself as a search publisher promoting Woo-Me on one of the engines. You bid on a variety of keywords. Without subid tracking, you look at your day's spend, compare it to your day's revenue and see if you made a profit. The challenge comes when you want to scale. How do you know which keywords you make money on and which keywords you lose money. Ideally, you would want to have the insight so that you could buy more traffic of the kind that makes you money and lower your bids / turn off traffic on the words that lose you money. With subid tracking, you now can. Instead of one tracking link, you can now have millions of unique links. Many networks, understanding the importance to publishers, even offer more than just one subid. They will have a link that includes subid1, subid2, and so on. Not all publishers will need more than one, but it comes in handy especially if the publisher has publishers of their own. Or, sticking with the search example, they might use one subid to reference which search engine and another to reference the keyword.
While those using subid tracking might not think of it as such, sub id tracking is a vehicle for transparency. For all parties that rely on it, (advertisers, publishers, networks) the sub id performs the same function, increasing transparency to make better decisions. In an ideal world, transparency would lead to increased efficiency, both for the advertisers, the networks, and the publishers - in unison. Not surprisingly, though, it has some unintended consequences that both sides say they want to fix but tend not to.
The Trouble with Transparency - We say we want transparency, but do we?
Given that sub ids play such a large role in tracking and that they
create transparency, that should mean transparency is a good thing,
i.e. that people would welcome more transparency. Just look at our social
lives - everything shared, overflowing with data points for those with
whom we interact. Not yet in the performance marketing world. At a point, too much
transparency seems to create friction rather than the other way around.
I
remember when LeadPoint first came out. Their model of
transparency radically altered the way publishers sold traffic. Prior
to LeadPoint, almost all leads were treated equal. One publisher
received the same amount as another. Yet, those familiar with the
mortgage market know that leads, especially higher dollar leads, can trade
within a huge range - almost one thousand percent. Having each
publisher receive the blended average of the whole meant those with
lower quality earned more than they should, and those with good quality
subsidized the lower quality guys. Only the large players with the
robust sales force knew the real value of each of their leads.
LeadPoint changed that, allowing any publisher who sold to them to earn
the true value of their lead, as a percentage of what LeadPoint made,
showing the lead seller exactly what transpired with their lead. All of
the sudden, those with great quality earned what they should, and it
allowed them to compete more effectively in the higher priced inventory
where they previously could not due to a blended payout. For those with
lower quality though, such transparency did exactly the wrong thing.
They now made some significant percentage less. Problems solved, right?
The golden rule of
performance marketing is that advertisers want to pay as little as
possible and publishers want to make as much as possible. Except for
the small handful of truly high quality performance sites, nobody wants
to make what they are worth, and while advertisers say they want to pay
the right amount, they have a mental barrier in doing so. Transparency
should solve this, but it doesn't. Transparency only highlights the
current market inefficiency. It's like sticker shock for some.
Going
back to lead generation as a way to explain this, imagine a site that
converted at 100%. Every lead turned into a sale. Let's say that the
buyer currently saw an average lead to sale conversion rate of five
percent, and they paid 20 dollars per lead, or 1000 per sale. They
might pay a high quality site 60, maybe 80 per lead, but even if they
knew that this site converted at 100%, they couldn't pay the 1000. They
would still think they could get it for less. And, so long as sellers
will offer them for cheaper, they will try even if the end result means
a higher overall cost. Thus, they settle on the blended average,
continually buying different sources, saying they want only the best
ones but hoping the next cheapest one will be that undiscovered source
of high quality. It doesn't differ much from a parking lot, where
people would save time by finding the first space, but we so like the
variable payout and frequency (ala slot machines), we drive around
convinced that this time might be the time. What results is a certain
amount of money dedicated to waste - money spent in the belief that
things should cost a certain amount, but ultimately paying more than it
might because what it costs seems higher than you'd like and you think
more should exist than actually does.
The advertiser doesn't
have all the culpability. The publisher and the networks do as well. Back to the golden rule,
publishers want to make as much as possible money, and generally they do not actually want to make what they are
worth. The sub id which helps with transparency has a fundamental flaw.
It tends to only help one part of the equation with transparency. It
helps the advertiser see the advertiser side and the publisher see
their side. It doesn't do what LeadPoint's exchange did and let each
side see the other on a transaction level. Unlike the exchange which
takes a flat fee per transaction like Visa, the network's act like any
network would, maximizing revenue and profit for it by manipulating the
two sides of the equation. Think of a crab cake. The best crab cakes
use very little filler. The network continually tweaks with how much
filler it can insert while still charging the customer a premium rate.
They know filler exists, but they know that they will make more money
by balancing the filler than charging the right price. They will sell
more cakes using the same amount of crab for a higher total. Their job
is to balance the filler per cake; if one cake is too much filler, it
hurts them, and if one is all crab, they've provided too much for the
price. The blended average is the blessing and the curse of the
performance marketing world and of transparency. But like transparency
itself, it takes both sides to change, and neither seems ready.
