The ABCs of PMPs: Part III

Read Part I here and Part II here.

So you want to set up a PMP. The conversation typically goes like this:

1. The Advertiser reaches out to a supply partner and says it wants to set up a PMP. Advertisers find supply partners in a variety of different ways—some use comScore, some use an SSP’s private deals marketplace, some have a longstanding relationship with an existing publisher. Oftentimes a DSP will be asked to reach out to an inventory source on behalf of buyers. One way or another, the conversation starts.

2. Next step is sorting out a price. Prices will vary widely from publisher to publisher. I’ve heard of some niche medical publications doing a $200 CPM (that’s two hundred—not a typo), I’ve seen some smaller, lower-tier publishers be happy with a $1 CPM. Most publishers have different rates for endemic and non-endemic advertisers, and for PMPs it’s no different. 

Advertisers, too, have a price they have in mind. Remember, on the group most PMPs are preferred deals and not guaranteed, so ultimately the price doesn’t matter all that much—what matters is if the advertiser is motivated to spend at that rate. I’ve had certain buyers tell me that they would set up a PMP at the rate I was asking for, but they’ll never have inventory available because the CPM was too low. Win rate/fill rate will generally go up as the price goes down. 

3. Once the price is agreed on, the publisher sets up the PMP in the deals section of their SSP, putting in the available placements along with frequency, audience targeting, and sites among other things. No I/Os exist for PMPs. 

4. With that done, you send out the Deal ID to the client, who approves the deal, then targets the deal ID in their DSP.  Then comes the troubleshooting, which can often take a week, even for motivated buyers and sellers. 

The honest truth is that the deal ID the buyer and seller are working on isn’t the only thing they’re working on, and thus the feedback loop can get a little long and it takes days before whatever issues there might be are worked out. Guaranteed deals in my experience are the easiest ones to setup and manage.

The troubleshooting is where things can get fouled up. Most DSPs have some sort of real-time troubleshooting tool that says why a certain campaign isn’t spending—reasons include losing bids to other campaigns, mismatching inventory sizes or formats, or a total lack of available inventory.

The use of such tools are almost essential—otherwise buyers and sellers are reliant on guesswork, which is nothing that you want in a business.

How Publishers Can Guard Guaranteed Revenue 

What can publishers do to protect their business among the non-guaranteed revenue zeitgeist?

Restrict who can buy your inventory through RTB channels. Some high-value, endemic publishers have long-standing media sales relationships with certain advertisers. We're talking long-term, multi-year relationships, where the advertisers see value in the listing and publishers see that these ads fit their audience well. 

These are almost exclusively media sales, I/O-based deals that can renew every year, provided that they're driving performance. Relationships such as these will often make up a significant portion of yearly revenue, and thus they must be protected.

Real-time bidding has the potential to damage relationships like these. It's no secret that premium publishers supplement their media sales revenue by selling their inventory on the open exchanges. And it's also no secret that advertisers are constantly working to make their media spend more efficient. 

So it makes sense that those long-standing media sales partners will attempt to buy this media that they used to only get with a signed I/O using programmatic channels, be it PMPs, or on the open exchanges.

The issue with this is that they're going from guaranteed revenue (in the form of an I/O) to non-guaranteed revenue in open exchange and PMP spending. And that puts premium publishers head to head against some less premium guys that can win when comparing CPMs. The Financial Times ought to charge advertisers like ScottTrade more, because Financial Times readers mean more to ScottTrade than most other advertisers.

Publishers should protect this revenue and these relationships, and require guaranteed PMPs, or work off an I/O. For long-term, endemic advertisers (or premium placements) I would suggest having a company-wide commitment to programmatic Gguaranteed Deals.

This one is going to require some backbone. If you re-read my section on PMP definitions, advertisers do not want to set up guaranteed deals. Guaranteed deals limits their ability to optimize and more or less defeats the purpose of programmatic buying from the advertiser perspective.

But for the publisher, units such as video or well-placed, high-impact inventory that helps create demand should command a premium. If endemic advertisers want access to these units, then they should pay guaranteed rates for them—particularly if they always have paid guaranteed rates.

Keep a firm hold on your data. Some publishers, attempting to drive additional revenue, are selling their audience data through services like Krux, Lotame, and BlueKai.

The thinking goes that if there's a target account that won't buy on an I/O basis and won't set up a PMP, it may still work with a publisher if it can get turned on to the publisher’s products by selling audience data.

It's true that audience selling will open up new revenue streams, but publishers that do this— especially premium publishers—might be hurting their display business. After all, if audience is the thing certain advertisers focus on, why would they ever need to buy media or run audience extension campaigns?

Likewise, be strict about who is purchasing your inventory and what technologies they use on your inventory. If someone is piggybacking Lotame, Krux, and BlueKai pixels on every impression they’re buying on your site, they’re going to capture a lot of data. Are you comfortable with them using that data build a panel and the kind of user that visits your site?

Note that for motivated buyers, it’s difficult-to-impossible to stop data leakage and theft. However, publishers can make it harder to accomplish.

Have advertising products that can’t be bought or are difficult to buy machine to machine. It’s not possible to buy synced units (for instance, 728x90 and 300x250 running together) or full page takeovers through RTB. The technology isn’t there yet. In addition, WelcomeAds (custom-sized banners that appear when a user first visits a page) are hard to buy programmatically.

Sponsored posts, which are bought/paid for by brands (think Walmart sponsoring a post on BuzzFeed) are also outside of the traditional ad tech ecosystem. Having products such as these will help protect more traditional digital inventory.

Develop non-advertising revenue sources. I’m talking lead generation, conferences, subscriptions, highly segmented email lists, etc. Most publishers are already doing this, but if they’re not, they should be. 

Making money from non-traditional revenue sources allows media sales organizations to have the confidence to hold the line on CPMs. Hosting conferences, building segmented email lists and running targeted campaigns, premium subscriptions/content and lead generation efforts, paid posts to well-followed social media pages all offer great revenue generation potential. 

Keeping Up the Bottom Line

PMPs are all the rage today and for good reason—they let advertisers and publishers cut out the middlemen that siphon off ad dollars, and ideally let both sides get better inventory for more favorable prices. But the industry-wide push to work exclusively through PMP deals is a significant threat to a publisher's bottom line.

For high-impact units, or with endemic advertisers where long-term value has been demonstrated, publishers should insist on guaranteed money deals, be they guaranteed programmatic or tag-based buys.



Will Rand is a long-time digital ad veteran who has worked on the sides of buyers, sellers and middlemen.

Rocket Fuel