Using DFP to pit ad networks against each other

[jpenns]

I've seen people talking about setting up DFP with multiple ad networks like Lijit and having DFP basically auction your ad space off to the highest bidder. This sounds great to me. The problem is I can't figure out how to do it. I'm not familiar with DFP at all, but even as I've spent hours familiarizing myself with it, I haven't been able to figure out this specific task. And there's not a lot of info out there about how to do it.

I saw an older thread on this forum where people were discussing it, which is what drew me here. Does anybody here do this, and would you be able to post a guide or something?

Thanks so much in advance!

[diberry]

I was hoping someone with more DFP experience would chime in, but here's what I think I've learned from my experiments so far re: frequency caps and CPM. Please be aware this is all very experimental.

I gave each line item the CPM I expect to get for it when it runs. Then I gave it a frequency cap of 3 per user per day. Then I noticed some of my placements weren't loading ads, and realized I had some line items targeting more than one ad unit, so, for example, if one line item targets two units, I give it 6 per user per day.

I had read about using a 3x24 cap elsewhere around here, so my reasoning was this: most of my visitors only look at one pages. Many look at three. Few look at more than that. And from daisy-chaining in the past, I know that my highest paying networks can usually manage at least 3 impressions before passing back to the next network. Therefore, it's a good guess that this frequency will let me have the highest paying network monetize the vast majority of impressions... and as far as I can tell, that's what is happening.

Your setup may vary. It all depends on the ad networks you use, and how likely you are to get Adsense clicks. I.E., if you are currently just using Adsense and your goal is just to make it compete harder by adding another network to the mix, then you may benefit from inputting a higher CPM than your other network(s) can actually deliver. For my site, Adsense underperforms compared to several other networks (so far... I am re-testing this, actually, but so far it doesn't look any better than it did 6 years ago), so I just let it backfill whatever.

Remember, this information may be worth what you paid for it. ;)

[scottb]

Optimizing yield with DFP, price priority and multiple networks requires a commitment of time that isn't worthwhile for small sites but probably worthwhile for big ones.

If you replace 100,000 monthly, low quality AdSense impressions paying a 50 cent CPM with another network that pays $1, you earn $50 a month more. Keep in mind that those 100,000 impressions might be out of a total of 1 million or more.

You might spend 15 minutes a week adjusting the settings of each network to account for swings in the average CPMs. The more networks you use, the more time it takes to manage them in DFP.

[anefarious1]

Scottb that really puts it all in perspective for me. For most people, it simply isn't worth the trouble. It seems to me that Google, if they wanted to, could incorporate the inner workings of DFP into Adsense and publishers could greatly benefit from the competition without even knowing it. In a perfect world I guess...

[poppill]

Thanks for your thoughts diberry. I'm about to dive back into my DFP setup, I'll keep you updated on what I do.

Whats your routine with the ad server like?

Currently I check ad reports about once a week, for every network in there. Compare them to what dfp shows (dfp is always off, who knows if i set it up wrong or what), then i try to guesstimate the right cpm's for all the networks in there.

I don't actually have frequency caps right now, maybe I should include that. My setup is mostly one big network that i have set at a higher cpm, which passes back to another network, which only fills at half that cpm. I've now added another network, with a cpm about midway between both of them.

Monkey business!

[child please]

If you are using DFP, you should not be using pass back tags. That's one of the main points of DFP - to eliminate the need for pass backs. At the very least, if you are using them together, the traffic to your pass back tags should be at a minimum. If there is a lot of volume, you are not optimizing correctly.

Also if you are not f-capping and your CPMs are different, DFP will always deliver to the network with the higher CPM. As I wrote earlier, it isn't that smart to be able to dynamically predict third party CPMs.

Ultimately, in order for DFP to work the best you have to constantly monitor third party CPMs and update accordingly, at least a few times per month and preferably weekly. Then based on those numbers, go back and adjust the values in DFP.

[diberry]

Child Please, I'm finding you do need to have passbacks set up *within the network control panel*. That is, say, you set up Lijit to passback to Burst within Lijit's settings. That way, if DFP gives an impression to Lijit but for some reason Lijit doesn't want it, at least it goes to Burst and the impression is still monetized. Is that what you're saying is unnecessary, or are you talking about within the DFP setup?

But ideally, you should minimize passbacks as much as possible by using your frequency caps.

As for my routine that poppill asked about: with daisy chaining, I have always checked my earnings for the month-to-date every single day to see what's going on because it only takes a few minutes and I like to really be on top of things. CPC fluctuates a lot, so you may need to check daily. I use mostly CPM networks, which don't fluctuate that much, so probably as often as Child Please recommends would be fine. Anyhow, you'll know if you aren't checking often enough, because you'll see you made less money than you should have and discover it's because a CPM unexpectedly went up or down on you.

When I check my earnings, the reports also display for each ad unit. If I notice it's changed significantly, I just log into DFP, do a search for "Burst" line items (or whichever network) and change the CPMs. I'm rarely logging into DFP, actually, because once you find the right setup, things don't change that much.

[child please]

diberry, yes I think we are on the same page with the passback. Sorry to clarify, I didn't mean that you shouldn't have passbacks set up, I just meant you shouldn't be seeing significant traffic to it. It's fine to have them in the odd instance your network can't monetize, but if you are using DFP the way it is intended, you won't need passbacks as you should only be targeting impressions that each network can monetize.

Even if you are using a CPM floor with your third party network (Lijit), passbacks should be kept to a minimal. Say for example, you've entered in your Lijit console that the floor for every ad should be 0.50 CPM. And then you've gone in and entered 0.60 eCPM in your DFP line items for Lijit (as you're assuming the ads that will be delivered by Lijit should exceed the floor, even if slightly). You also activate a passback code back to AdSense or another network in your Lijit console.

One month goes by and you see tons of impressions going to the pass back code. This means either your targeting is not efficient and/or your CPM floor is too high (and by extension your eCPM in DFP). In other words, Lijit isn't able to monetize a lot of impressions at that 0.50 guarantee and is passing back. The next step would be to either fix the targeting/capping or lower the floor and eCPM so less impressions are passed back.

Also your statement about if you have your DFP setup correct, you don't have to log in frequently is totally correct assuming your CPMs from Lijit, etc. are pretty static. I work with a company lets call them Network A - they were delivering $1 CPM to my US/CA traffic in March so I had the eCPM set to 1.00 in DFP. But now in April (start of Q2), it has dropped to about 0.63 so I have gone into DFP and adjusted. If I left it at 1.00, that would mean I am giving that network a lot of impressions that it is only monetizing at approximately 0.63 even though Adsense for example could have beat it and monetized it at 0.7, 0.8, 0.9, etc. But the ad went to Network A because of my incorrectly entered high CPM which AdSense could not beat.

[diberry]

Yep, we are on exactly the same page. Frequency caps and CPM settings in DFP are all about minimizing passbacks so you get the highest CPMs available for each pageload.

I'm also finding that once in a while, a particular ad unit from a particular network will have such low inventory that it's actually better to pause them for a week than even have a frequency cap of 1. A week later, I turn them back on and see if things are looking up or not.

If you've just been using Adsense and that's working for you, it's probably not worth switching to DFP, as others have noted. But if you work with a lot of networks to maximize income and are concerned about providing the best UX you can, DFP gives you all the advantages of Adsense (async, minimal effect on pageloads, 100% fill rate) without the disadvantages. "Disadvantages to Adsense?" you ask, stunned. Well, yeah, sometimes Adsense isn't the highest paying network in your niche. And on some sites, its keyword targeting backfires horrifically - i.e., you have a page griping about Senator Mushbrains, and Adsense gives you ads for his campaign. I've seen things like this happen many times to clients and fellow webmasters, and the only advice I can give them is: avoid keyword-targeting networks.

[LuckyD]

For someone who is yet to dig deeper into DFP, thank you! I'm currently juggling AdSende and Medianet at 1.2MM pageviews. Will give DFP a shot and report back. In the meantime, do you guys have any preliminary results from your test?

[Runfun]

I'm also curious about test results of others. With some help of this topic I finally manages to setup DFP and now I'm using it just for one position. I still have some questions:

- I see DFP clicks on Google Adsense but in my Adsense dashboard I do not see Adsense clicks on the banners I created especially for DFP. Does anyone have experience with this?

- I will now compete several blocks of Adsense and OpenX with an affiliate for one place/position, is this recommended?

- I see two banners with the text ready but still not delivering, does anyone have experience with this?

- Last question is about setting rotating creatives, I have a setup with evenly. Is this the best sollution to make the different networks compete with each other?

[ember]

When I used dfp, there was no other ad network that paid better than Adsense, so I didn't see the point of making networks compete. Adsense always won so why go through the frustration of setting up dfp. I'm sure I was missing something....

[diberry]

Runfun, when you set up your line item, it defaults to allowing Adsense to serve as backup. Those clicks won't show as clicks on your Adsense ads. Could that be what you're seeing?

As for set up evenly: that's probably not the best setup. What I do is have them compete on price, and I input good guesstimates of the current CPM for each ad in each network, and then limit them to a frequency cap of 3-5 per visitor. Rotating them evenly could work somewhat, but the way I do it ensures you get the best paying ads loading first for each visitor.

Supposedly, anyway. I'm finding networks just ignore ad calls more and more, and I'm not sure why. It's all pushing me back toward affiliate marketing and selling my own products.

Ember, here's the deal. There's SEO, and then there's actual, real marketing. In real marketing, you do stuff like sponsored content, guest posting, selling links, because those are the online equivalents of real marketing practices from before the internet. In SEO, of course, those are naughty bad spammy things, according to the only company that gets a say. Almost everyone in this forum has built their sites around SEO, and in those cases, Adsense will always perform best (what I see as "made for Adsense" is MUCH broader than how SEOs define it). I built my sites to promote via social media, so for me, other networks - which focus on and encourage real marketing techniques even if Google shuns them - always outperform Adsense. That's why I rarely ever come here anymore. Real marketing was always going to outlive Google's power-tripping BS rules, and I jumped on that bandwagon years ago.

[Runfun]

Yes you are missing something because (for example) Openx pays more in some situations.

[Runfun]

@diberry, I've no idea how I can compete the line items with each other. I can choose:

- Evenly
- Optimized
- Weighted
- Sequential

[diberry]

Runfun, hmm. This is from my tutorial on the first page of this thread:

Now skip down to the Line Item section. I give it the same name as the order. Pick the "Inventory Sizes" from the drop down - you can have more than one, for example 300x250 and300x600. Now go down to Type and select Price Priority. Start Time: immediately. End Time: Unlimited. You DO NOT need to input any CPM at all - I've read that DFP will always pick the highest ad it can serve anyway. Inputting CPM will help DFP create projections of income - one of its many powerful features that I didn't bother trying to set up - but since CPMs vary greatly from day to day, it may not help that much anyway.

So you *should* be seeing "price priority" at some point in your selections. I can't even guess why you wouldn't be seeing that option (this is what sucks with DFP - very confusing interface, lots of extra steps, from my perspective at least, that muddle the process.)

As for what you said about OpenX paying more... now I'm really confused because I thought it was another ad server like DFP, and the ad servers don't pay you, the ad networks do. Is OpenX an ad network now?

That said, it also occurs to me that we may be at cross-purposes on what I was trying to achieve with DFP. I'm not trying to get Adsense to pay more than any other network. I'm just competing them against my other, higher paying networks so they will at least pay what those networks pay. In my case, other networks generally pay more than Adsense because of my niches and other variables, so I'm just trying to boost Adsense, not get it to outperform the others. That's never going to happen for me.

[Runfun]

Thanks for the help, and I have created new line items as previously created line items couldn't be changed anymore in price priority. I test it provisionally in one position but with both Adsense and OpenX (a total of 6 different blocks / dimensions).

OpenX is not only an ad-server but also an ad network. For using the ad-server I need a lot more views than I've now. Within OpenX I see buyers such as Yahoo, Just Click Media, Sociomantic, AppNexus, Criteo, Media Math, Turn and so on.

What was the improvement for you with the CPM after the switch to DFP?

[Runfun]

The result so far is an imploding of impressions and impressions for OpenX totally dried.

[child please]

OpenX is both an ad server, an ad exchange and a traditional network. I don't use their ad server technology but do use them for ads and they are pretty good!

Someone asked what's the point in using DFP when no other ad network pays better. You're missing a huge point. If AdSense pays $1.00 eCPM on your ads without competition from another network, if you use DFP and have AdSense compete dynamically, they may now bid $1.25 on that EXACT SAME impression due to that competition! Think about it...you want as many advertisers battling for your impressions - this is what AdSense hypes. Now why not take it to another level and have multiple NETWORKS battling for those impressions, each with multiple advertisers? :)

[child please]

BTW, Google themselves told me that AdSense will bid higher on an impression through DFP than through only AdSense itself. Then I tried it and it was true.

[diberry]

child_please, are you saying you use OpenX to provide some ads (line items) in DFP? Just asking because I'm always on the lookout for another good ad network.

[Runfun]

@Childplease, Thanks for the links in the private message :-)

@Diberry, I use OpenX ads at line items and that works fine with me

The problem I have now is that with price priority and a CPM of 0.00 no more impressions where made. No idea how so I switched back to a amount of CPM and not price priority but network. Hoped it would pit against each other but it doesn't work yet.

[diberry]

Runfun, I have price priority AND set CPMs. That's what makes them compete.

[Runfun]

Okay I thought you said to setup without a CPM (0.00). I'll change it tomorrow because now it's bedtime.

[diberry]

Whoops, you are right. I initially thought that was the case, but then someone corrected me, but by then it was too late to edit the post. I am planning to write a tutorial on my own website if I ever find the time. Every day, I have my to do list, and everyday, 6-8 emergencies come out of nowhere because there are so many moving parts in everything I do, and something's always going wrong.

[child please]

Yes, OpenX provides ads in a "traditional" network sense.

You need to set Price Priority and an accurate CPM that reflects what OpenX is delivering. I usually take the eCPM of the third party network and raise it a little bit. So say OpenX was delivering 0.67 eCPM, I will put 0.75 as price priority into my DFP Line Item. In theory, now AdSense will need to beat 0.75 in order to show an ad. If they can't, it goes to OpenX.

Not only do you get the most value out of each impression (why send it to AdSense for $0.50 if OpenX will deliver $0.67) but it also makes AdSense bid as high as possible in order to win.

Also, the more networks you have and the better your targeting, you can get even more value out of each impression.

[diberry]

Which targeting options do you tweak, CP? So far I haven't delved into that for fear I don't really know what I'm doing and could make things worse, LOL.

[child please]

Two main ones:

Geo (perhaps most important)
Device (desktop vs. mobile)

And not specifically targeting, but make sure you are using accurate frequency caps

[Runfun]

Well today wasn't that bad for me. CPM of Adsense was almost doubled than normal but I've no idea where I can find that amount in my Adsense dashboard.

I've competing Adsense, line items containing both Adsense and OpenX but OpenX dropped in earnings despite I'm using a higher CPM. I'll figure this out next days but I think it's strange because normally OpenX pays a higher CPM than Adsense.

[child please]

Sounds like you may need to raise your OpenX rate/value CPM in each of the line items. Probably what is happening is that AdSense is winning most of the 'good' impressions, leaving the less desirable ones for OpenX, which isn't paying as good of a rate on them.

Also, what do you mean by having a line item containing AdSense? Not sure I understand that. AdSense will automatically be competing for your inventory if you select that option under inventory and then network settings (or in each ad unit individually). It is way less efficient to have that disabled and then have AdSense entered as a line item - it isn't competing with dynamic allocation then, just at a static CPM.

Basically you want it like this, say you have a 728x90 ad unit:

Order: OpenX
Line Item: 728x90
CPM: whatever

Order: Network #2
Line Item: 728x90
CPM: whatever

Then under inventory, ensure AdSense is competing in that ad unit.

[Runfun]

I'm using several dimensions at one position but that's just coincidence for that position. So I've these line-items

OpenX 970x250 CPM 0.80
OpenX 300x250 CPM 0.70
Adsense 970x250 CPM 0.30
Adsense 728x90 CPM 0.30
Adsense 336x280 CPM 0.30
Adsense 300x250 CPM 0.30

These CPM's are an example but it's still delivering more to the Adsense line-items. OpenX was paying even more before I didn't used DFP.