Facebook Developers

headieherbs · 2008-03-31 10:30:25

My QR went from +50 to -43 in about 7 days and I'm sick of it. I know that this happens to people all the time with Cubics, but do the rates ever go back up again, or do they just continue to screw you after tantilizing you with a .44 CPM for a month or so?

Please let me know so I can decide what to do, right now I'm looking for an alternative that pays as much!

thanks everyone!

peace

claudfs · 2008-03-31 11:05:53

In my experience my cubics rating has never gone up again after dropping.

Try SocialMedia socialmedia.com

They pay quite well.

sweetsteve · 2008-03-31 13:37:18

My ratings were going up every day. They made it up to 93, then yesterday I was mucking with code, messed some stuff up, and got dropped to -17. Today it's creeping back up, now at 33.

Michael · 2008-03-31 13:53:53

Since September when I started with Cubics they've dropped their base CPMs but they're still pretty reasonable. I think there have been 3 periods (including this one) where they're obviously running ads they can't pay out big on so they start using their "quality rating" to adjust the eCPM. Personally, I run a combination of Cubics, AdSense (currently doing very well) and SocialMedia.

gungadan · 2008-03-31 13:54:42

headieherbs wrote:
My QR went from +50 to -43 in about 7 days and I'm sick of it. I know that this happens to people all the time with Cubics, but do the rates ever go back up again, or do they just continue to screw you after tantilizing you with a .44 CPM for a month or so?

Please let me know so I can decide what to do, right now I'm looking for an alternative that pays as much!

thanks everyone!

peace

Ours have gone (in reverse order):

50 <- 23/3
27
40
57
52
44
51
50 <- 30/3

So yes, the rating can go up as well as down. What I don't understand is why we get 50% fewer impressions with them than through Socialmedia (1 ad from each displayed).

JohnFoe · 2008-03-31 14:48:16

Because they load very slowly during the prime parts of the day (and even in the slow parts of the day).

It does suck, but I'm able to tolerate it for now. If you are looking for a substitute for the eCPM model that cubics has you should try lookery. I'm not saying lookery is GREAT, but the one time I did use them they paid me without me knowing which is a good thing since I didn't have to use up my time begging for payment.

headieherbs · 2008-03-31 15:02:31

yeah I tried the lookery code yesterday and no ads showed up? it sounds like their rates are tolerable, does anyone here use lookery, if so what kind of rates do you usually get? (without the guarantee)

gatesvp · 2008-04-01 13:57:40

headieherbs wrote:
My QR went from +50 to -43 in about 7 days and I'm sick of it. I know that this happens to people all the time with Cubics, but do the rates ever go back up again, or do they just continue to screw you after tantilizing you with a .44 CPM for a month or so?

Hello headieherbs, there seems to be some confusion about how this works, hopefully, I can clear this up a bit.

When you start with Cubics, we give you a rating of +50 which is used to indicate "average". We continue to give you that rating until we get a significant number of views so that we can actually "peg" your traffic quality. Sometimes is takes hours, sometimes it takes days or weeks to get an accurate quality rating.

The reasoning is pretty simple, we track clicks (amongst other performance metrics). Some publishers are getting 0.01% CTR (that's 1 click per 10,000 impressions). If you send us 3,000 impressions / day for 5 days (15k impressions) and you have two clicks, what's your CTR? Is it 0.015%? Is there really any number here that's accurate? The best I can tell you is that your traffic probably isn't very good.

But maybe it's a new app, maybe you're still growing, maybe you put your ads in a poor location, who knows? At 15k impressions your CTR is just not significant. So you just get an "average" quality of 50 until we get enough views to infer some significant number. In your case, this significant number was -43. That means that the "quality" of your traffic isn't very good.

The truth is, if we went from 50 to -43 in 7 days, we probably lost money for the first 7 days.

So what influences quality rating? Lots of things: viewer's geographical location, # of clicks, # of conversions, specific ads that are being clicked, visibility of the ads...

Some of these things are up to us to control: ad inventory, value of the ads. As Michael mentions, sometimes you ride the tide. For example, there was a run before Christmas where several advertisers were raising bids across the network. This boost raised nearly everyone's Quality for a week. In the same vein, if Facebook decided to ban certain high-performing ads, we could see everyone's quality drop. That's all just inherent with the business, you can't control these things, so you'll have to serenely accept them. This will be true for all ad networks.

Then there are the things that you can control: ad visibility. Look at Sweetsteve's example

sweetsteve wrote:
They made it up to 93, then yesterday I was mucking with code, messed some stuff up, and got dropped to -17. Today it's creeping back up, now at 33.

Doesn't it seem significant that his changes would actually affect the quality rating?

There are tons of things that you can do to increase your traffic quality. Run several A/B tests: ad locations, ad formats. If you put ads where nobody sees them, then nobody clicks on them, if nobody clicks on them then advertisers don't pay us and we can't pay you. If you put ads at the bottom of a page and nobody scrolls to the bottom, then that ad is basically worthless and your quality rating will reflect this.

Also, what's your traffic makeup? If 70% of your traffic is coming from Turkey, then it's worth a lot less than having 70% of your traffic coming from North America. North American advertisers pay 10 cents to $5 for clicks, Turkish advertisers pay 2 cents for clicks. If your traffic contain a lot of Turkish clicks, then your quality is going down (and the reasoning should be obvious).

gungadan wrote:
What I don't understand is why we get 50% fewer impressions with them than through Socialmedia (1 ad from each displayed).

I would check the source of your traffic. We don't count views that we can't monetize. Cubics does have a relatively large advertiser base, but there are some places we just don't cover. If you're getting significant traffic from Vietnam and Singapore (for example), then we may not count some of those views so that we don't drop your eCPM.

If you're finding a significant number of "dropped" views, add geo-targeting to your app and run some comparisons. Based on our experience Cubics has top payouts for "English" traffic (US, UK, CA, AU), b/c we actually have ad sales teams in those countries. You may find that you need to turn up Cubics in some countries (where we pay better) and rebalance the mix in countries where we don't pay out as much. (of course, caveat emptor if you're getting lots of 10 cents clicks in Singapore from other networks)

Not all views are created equal, not all clicks are created equal. If your CPMs are really low, then it's likely that the value of each impression is quite low. I can provide custom suggestions for your apps, but no one can guarantee results.

If you're interested, please drop me a line with a link to your app (use the e-mail link to the left).

Last edited by gatesvp (2008-04-01 14:06:32)

jasek · 2008-04-01 15:07:07

Gates, why pay on a CPM basis then? Wouldn't it be more appropriate to use a CPC model if the amount of revenue accrued is based on the number of clicks?

sweetsteve · 2008-04-02 08:00:24

By the end of the day after screwing things up and dropping to -17 (which btw ended up at -13 by the time they adjusted it), my rating made it all the way back to 85.

I will qualify this by saying that most of my cubics ads are running on Bebo, which is mostly a UK audience.

Let me explain what my mistake was so others don't make the same one.

I had a request to run some other ads for an individual app on my application, and I wanted to rotate them with cubics. Rather than do a simple random thing, I decided to get fancy and use my openads ad server.

I set up the two accounts to run with even weighting, put the cubics code in the one campaign, and the other apps banners in the other campaign.

Everything worked fine, and I let it run all day. But by the end of the day, my ratings had dropped to -17.

I thought about it, and decided what had happened was that by feeding the cubics ads into an iframe from my ad server, they weren't getting all the fb_sig parameters, so they lost all their targeting information.

I made one other stupid mistake. My app is flash, and people stay on the same page for 10 minutes or more. OpenAds has a nice feature to be able to automatically refresh the ads at intervals which I set to 60 seconds. I figured it would get better click results by changing the ads every now and then and that it would make cubics happy. Big mistake. My adviews went up, but rating dropped like a rock.

That would be one request I'd make of cubics, to put in an ad refresher so that the ads don't become stale on applications that have long viewing times. It would increase our ad views, help with click-through rates, increase everyone's revenues.

Last edited by sweetsteve (2008-04-02 08:02:14)

gatesvp · 2008-04-02 10:07:25

jasek wrote:
Gates, why pay on a CPM basis then? Wouldn't it be more appropriate to use a CPC model if the amount of revenue accrued is based on the number of clicks?

Because it's not just clicks:

gatesvp wrote:
we track clicks (amongst other performance metrics)

What I'm giving you is the simplified version of what ends up being a very complex system.

We also have some CPM advertisers that actually do pay CPM. But we want to normalize these CPM advertisers amongst the Publishers that we know perform well. Otherwise the CPM advertisers just drop you when you send them a million views and zero conversions.

There are also CPA advertisers that pay out based on conversions. We need to back-convert reported conversions into CPM payout, but we also need to credit high-converting publishers (higher pay and increase in the converting traffic). Imagine that 5 publishers each send 20 clicks (100 total) and they result in 10 conversions divided as follows: 1,1,2,2,4. Obviously that last guy deserves more money per click than the first two guys. Of course, our CPC advertisers can also track conversions, so similar logic will apply.

It's all the same issue: All views are not equal, all clicks are not equal.

Again, thanks sweetsteve for a solid example. Instead of showing one ad every 10 minutes he showed 1 ad every minute. But it's quite likely that the extra 9 ads were never really "seen" b/c users were really engaged with sweetsteve's cool flash app.

Ads have to be seen and they have to translate into sales or nobody is going to advertise. Cubics pays by eCPM b/c it gives you a reasonable evaluation method against which to compare your own traffic. It also provides a reasonable method for aggregating all of this complex data. Obviously, comparing your eCPM to some other app is pretty pointless. But comparing your own eCPMs on a day-to-day, week-to-week basis does provide you with some metric for evaluation.

nstory · 2008-04-02 11:23:14

As a publisher (mostly on Bebo right now), I found your posts, gatesvp, about the workings of Cubics the most enlightening to date. I've long been suspicious of running your ads on my applications, because the lack of transparency the "quality rating" gives me leaves me feeling uneased. After reading your posts I see how you determine the CPM rates given the variety of systems you make use of: CPM, CPC, CPA.
Still, I'm left feeling that the CPM adjusted by quality rating system is not what I want. I would feel more comfortable, and would be more likely to make use of your network, if you would give me more numbers regarding exactly how my ad spots are performing. Why not expand the publisher dashboard to include a breakdown:
Tell me the number of CPM, CPC, and CPA impressions my app is delivering me.
Tell me how many clicks I'm delivering to, at least, the CPC and CPA ads.
Give me a break-down:
$XXX earned through CPC clicks
$XXX earned through CPA conversions
etc.
In my applications I'm careful about where and when I display advertisements. When one of my users clicks on an advertisement, that means I'm giving him up for, typically, a few cents on average. For each user that I give up, I'm not receiving from him the viral actions he might commit, and the cross-promotions he might follow. Essentially, I'm sacrificing some growth of my applications' user bases in exchange for advertising revenue.
Without more detailed reporting from my ad network, I'm left to guessing as to what my users are doing.
How can I A/B test if I'm not given the statistics to analyze the outcome? The "quality rating", which changes according to a, I assume, ever-changing formula that only you people know, is not a good metric for me, as a publisher. Even the eCPM you report doesn't give me enough information.
When you tell me that applications with low-impression counts are automatically given the average quality rating, I'm left thinking that developers which optimize their applications to deliver quality traffic are effectively subsidizing the hordes of low impression applications. If you would simply deliver good reporting of the calculations behind the scenes, and calculate a publisher's revenues based on the actual advertising dollars he is bringing in, it would go a long way towards creating a trusting relationship with your publishers, and probably put an end to the vast majority of the "Cubics sucks", etc. threads here.

Thank you for reading my rant !

Last edited by nstory (2008-04-02 11:24:28)

sweetsteve · 2008-04-02 11:36:26

Well said Nathaniel

AG1 · 2008-04-02 12:30:44

I tried both Cubics & Socialmedia :
> Cubics = $0.25-$0.35 eCPM
> Socialmedia = $1.25-$1.35 eCPM
...

gatesvp · 2008-04-02 14:26:20

Hey Nathaniel;

Thank you for the post, I think there are two things going on here:

Issue #1 is the revenue calculation:

nstory wrote:
...calculate a publisher's revenues based on the actual advertising dollars he is bringing in ...

How else do you think we calculate it?

My post above is just a brief overview of the underlying complexity involved in arriving at this number fairly. This process is a black box and it's likely to stay that way. (Do you know how Google's AdSense calculates payouts?) But fundamentally it's about sharing revenue. Your app brings in X and we pay you some number less than X. That's the basic formula but it's not terribly ingenious. No one's giving away the "magic formula" and no one ever will. And it's changing every month, it has to keep up.

And this leads into issue #2:

nstory wrote:
...Give me a break-down:
$XXX earned through CPC clicks
$XXX earned through CPA conversions...

What are you planning to do with this data?
- Can you control the mix of CPC vs CPA vs CPM?
- What's your plan for optimizing CPA conversions?
- What's your plan for ensuring that users are only clicking on high-paying ads?
- Are you really going to "optimize for CPM"?

Yeah it's "fun" to have these numbers, but you can't really control them. As an ad network, it's already our job to optimize for these factors as best as we can. It's already in our best interest to make you as much money per view as possible (bigger X -> more money for everyone). We have data about all of the views and clicks in the system as well as data about the current portfolios of advertisers. How do you plan to optimize your ad mix with less data than we have?

It's all come back to the same advice: make your ads visible, run A/B placement tests, track individual ad performance, attempt to optimize underperformers, gather lots of data.

Personally, I'm more interested in giving you data that you can use.

nstory wrote:
...Without more detailed reporting from my ad network...

And this where we can do something. What other useful stats could Cubics provide? I've asked this question at a couple of conferences and the standard answer is "We don't really care as long as the eCPMs go up." Subsequently, there haven't really been a lot of changes.

I have had exactly two reasonable requests for better reporting and it's stuff that's being coded right now:
- Country breakdowns: traffic & performance by country
- Better channel reporting tools: grouping by channel instead of just filtering, channel grouping (into those groups), data exports so that you can use a spreadsheet for analysis

Now, if you have some actionable data that you want to see, then please let us know. (and no, click data is not actionable, but country data is)
So please drop me a line directly or post up here, I'd love to hear them. The ones with the most support will be rolled out first.

AG1 · 2008-04-02 14:30:31

@gatesvp : then how do you explain then difference between your eCPM and Socialmedia's ?
I'm not a Socialmedia shareholder and I really had this difference.... From $0.30 to $1.30, the gap is huge!

gatesvp · 2008-04-02 14:52:56

AG1 wrote:
then how do you explain then difference between your eCPM and Socialmedia's ?

How do you explain that several top publishers in this space are running Cubics?

Obviously this is a competitive space and one Publisher's experience is clearly not enough data by which to declare a winner.

AG1 · 2008-04-02 14:56:49

I don't declare any winner... I'm just trying to understand how such a difference is possible.
Perhaps could you imagine that any publisher is asking himself what is the best deal for him...

nstory · 2008-04-02 16:32:37

I think when I was talking about optimizing as a publisher I wasn't really going down the right track. It's true, statistics regarding what ad types are bringing in the cash aren't something I could really take action on. When it comes to connecting with advertisers, managing ad inventory, and all that.. well. I have neither the skills nor the inclination to get involved. That's why I'm glad that there are ad networks out there like you guys!

It's true that I like having numbers to look at. I really like numbers. They are "fun". I probably reload my various stat pages five times an hour, just to see the numbers move around. I understand, though, that it's not your responsibility to amuse me. What feels important to me, and I really mean to use the word "feel" 'cause this is only semi-rational, is the scent of transparency. I understand that your internal metrics and systems are nothing to be published, both to keep your company secrets safe from competitors, and to prevent fraudsters from gaming you, but simply by giving me the illusion of seeing where the money is coming from, I feel better.

Consider how many posts there have been on this forum from people, just like the parent, asking "Why did my quality rating drop?", and "Why is my impression count so low?", and all the like. If publishers could see a table showing the numbers adding up, we would feel that our ad network is working honestly on our behalves. Feelings matter. I stay away from companies with poor websites, because they "feel" shady, and I'm attracted to companies that give me good vibes.

If I could see a table like the one I've outlined below, I would feel better about placing your ads on my applications. Not because the data would be actionable, but because it would increase transparency. And, really, do you like answering all the posts, sometimes angry posts, from the publishers here that are trying to understand quality rating this, and that? Just get rid of it, or at least limit it to the true CPM impressions. Some more transparency could create a lot more good will. And good will leads to more revenue for you guys, a better reputation for Cubics, more advertisers, and ultimately more revenue for me and other publishers!

Code:

+-----+-----------+------+-----------+--------+
|     |Impressions|Clicks|Conversions|Earnings|
+-----+-----------+------+-----------+--------+
|CPM  | 10,000    | *    | *         |  $4.00 |
|CPC  |  5,000    | 50   | *         |  $2.50 |
|CPA  | 10,000    | 100  | 2         |  $4.00 |
|Total| 25,000    | 150  | 2         | $10.50 |
+-----+-----------+------+-----------+--------+
Un-monetizable impressions (not counted above): 1,000

Thank you again.

headieherbs · 2008-04-06 23:57:33

elequently put. dude. I completely agree and couldn't have said it any better!

edit: looks like my QR and eCPM are starting to slowly creep back up, so that's good

peace!

Last edited by headieherbs (2008-04-07 00:16:22)

markdoub · 2008-04-08 08:38:01

I have problems with cubics to.
My eCPM was dropping. I asked gates why, he aswered me and then my eCPM started to go up
I didn't do anything to the code of the app or the ad...
Then about a week ago Cubics ad Views started to drop. Last days I have 500views/day in cubics but I have 2000 DAU in Facebook stats.
This is very funny. ))
Gates can you explain this to?
If you want more data, I can give you.

markdoub · 2008-04-14 09:34:38

20 days and no answer...
these are cubic's stats:
2008-04-14 115 49 $0.40 $0.00 $0.40 $0.05
2008-04-13 426 50 $0.40 $0.00 $0.40 $0.17
2008-04-12 449 50 $0.40 $0.00 $0.40 $0.18
2008-04-11 575 50 $0.40 $0.00 $0.40 $0.23
2008-04-10 487 50 $0.40 $0.00 $0.40 $0.20
2008-04-09 1,692 50 $0.42 $0.00 $0.42 $0.71
2008-04-08 713 50 $0.40 $0.00 $0.40 $0.29
2008-04-07 511 50 $0.40 $0.00 $0.40 $0.21
while fb shows for last week 2100 DAU.
Maybe I should chage ad network...

Facebook Developers

Forums

#1 2008-03-31 10:30:25

Do Cubics quality ratings ever go back up after dropping?

#2 2008-03-31 11:05:53

Re: Do Cubics quality ratings ever go back up after dropping?

#3 2008-03-31 13:37:18

Re: Do Cubics quality ratings ever go back up after dropping?

#4 2008-03-31 13:53:53

Re: Do Cubics quality ratings ever go back up after dropping?

#5 2008-03-31 13:54:42

Re: Do Cubics quality ratings ever go back up after dropping?

headieherbs wrote:

#6 2008-03-31 14:48:16

Re: Do Cubics quality ratings ever go back up after dropping?

#7 2008-03-31 15:02:31

Re: Do Cubics quality ratings ever go back up after dropping?

#8 2008-04-01 13:57:40

Re: Do Cubics quality ratings ever go back up after dropping?

headieherbs wrote:

sweetsteve wrote:

gungadan wrote:

#9 2008-04-01 15:07:07

Re: Do Cubics quality ratings ever go back up after dropping?

#10 2008-04-02 08:00:24

Re: Do Cubics quality ratings ever go back up after dropping?

#11 2008-04-02 10:07:25

Re: Do Cubics quality ratings ever go back up after dropping?

jasek wrote:

gatesvp wrote:

#12 2008-04-02 11:23:14

Re: Do Cubics quality ratings ever go back up after dropping?

#13 2008-04-02 11:36:26

Re: Do Cubics quality ratings ever go back up after dropping?

#14 2008-04-02 12:30:44

Re: Do Cubics quality ratings ever go back up after dropping?

#15 2008-04-02 14:26:20

Re: Do Cubics quality ratings ever go back up after dropping?

nstory wrote:

nstory wrote:

nstory wrote:

#16 2008-04-02 14:30:31

Re: Do Cubics quality ratings ever go back up after dropping?

#17 2008-04-02 14:52:56

Re: Do Cubics quality ratings ever go back up after dropping?

AG1 wrote:

#18 2008-04-02 14:56:49

Re: Do Cubics quality ratings ever go back up after dropping?

#19 2008-04-02 16:32:37

Re: Do Cubics quality ratings ever go back up after dropping?

Code:

#20 2008-04-06 23:57:33

Re: Do Cubics quality ratings ever go back up after dropping?

#21 2008-04-08 08:38:01

Re: Do Cubics quality ratings ever go back up after dropping?

#22 2008-04-14 09:34:38

Re: Do Cubics quality ratings ever go back up after dropping?

Board footer