Understanding bidding : CPC vs CPM vs Smart CPM – what method to use? (13)
Understanding bidding : CPC vs CPM vs Smart CPM – what method to use?
My goal for 2017 is to fire up a ton of up-to-date articles/guides, that would cover the most important and most "asked about" topics that relate to running campaigns. I'm sure you will agree, that bidding is one of such topics for sure.
Even thou we have a great series (about bidding) written by Caurmen already on the forum, there seems to be a lot of confusion in terms of understanding how to approach it - judging by the frequency of appearance when it comes to bidding related questions 
The goal of this article will be to explain how various bidding types work, what's the difference between them and how to approach each of them. As I mentioned already, we have several other threads dealing with bidding here on STM, I will post the links to them at the end of this one
CPC, Smart CPC, CPM and Smart CPM … WHAT DO THESE MEAN ACTUALLY?
I guess you all know what these abbreviations stand for, but anyways :
Cost Per Click (CPC) – You are billed per click, your bid is also the final price that each click will cost you.
Smart CPC – This term is not used much, but many “quality score” based networks use some kind of smart CPC algo – think networks like Google Adwords – you set a bid, they charge you less for some clicks, more for others … but the final CPC will always be pretty much equal to your bid or lower.
These bidding algos are way more complex thou, so I won't focus on them in this article.
Cost Per Mille (CPM) – Mille means 1000 impressions in this context (M is also Roman numeral for 1000). With standard CPM, the price you set is also the price you pay, without any adjustments.
Smart CPM – Same as CPM, with one exception – the algo will adjust the final price that they charge you, to be a bit above the bid of the closest competitor “below you”.
THE MAIN DIFFERENCE BETWEEN CPM or CPC:
There is one thing you need to understand here, any traffic network wants to make the most revenue they can, that's the only thing they care about, really.
Some of them will pride themselves on doing whatever they can to enhance user experience… but in reality, if it was possible, they would literally show more ads down the throats of their surfers … with a shovel
So, whatever bidding system they use, they do it to maximize their revenue. If they choose CPC model over CPM, it's not because they want to sell cheap clicks to you, it's because they believe that their ad serving algo can make more money for them, compared to relying on the auction model alone.
The theory here is easy : when using the CPM model, they get some $ amount per 1000 impressions, no matter how well the banners perform, their revenue will remain the same. But what if they could pick the banners with the highest CTR and show just those… that could mean more revenue, right?
This is basically the only difference between the CPM and CPC model, but it totally changes they way you need to work with networks who use one system compared to the other. Here are the main issues you will have to deal with:
ISSUE #1 – Casting Period
All CPC networks use some kind of a casting, when you upload new banners. Remember when I said they want to find the banners with the highest CTR in order to sell the most clicks? That's what they do in the casting, each banner gets a more or less fixed amount of impressions to set their initial CTR.
Once the casting is over, only banners with the highest CTR will be served. This complicates the testing process, especially when you want to test a lot of banners. Some algos are so stupid, that they will only push the single banner with the highest CTR, once the casting is over. And we all know that the highest CTR doesn't equal to highest ROI.
The only way of working with such an algo is to create campaigns with very limited amount of banners, even 1 banner per campaign at times, finding the best converting ones and then playing with bids to reach the sweet-spot.
There is one more problem connected with the casting periods : these often mean “loosing” money for the network, because they have to test ALL banners, even those with lousy CTR. So they slow this process down, in order to minimize losses. This can mean quite some delay when launching a lot of new campaigns or uploading lots of banners.
CPM is way easier to work with in this context : you create a campaign, set the bid and start testing. The algo will evenly distribute the impressions across all the banners in the campaign. Some networks will even let you set % per banner, so you can pick what banner to show more. With the right bid, you can get steady volume straight from the start, without having to wait for any castings to finish.
ISSUE #2 – Traffic Volumes
Keeping your volume levels high with the CPM model is pretty easy, all you need to do is bid high enough and keep your impressions cap at a reasonable level. In order to fight banner blindness, you should always run several banners in every campaign and let the algo rotate them. When your banners loose traction, simply upload new ones and keep on rolling There are obviously more factors to consider here, but the core principle is simple.
With CPC bidding, it's a bit more tricky. CPC algos can only make more money, compared to CPM, when they favor high CTR banners with solid bids. So you need to both bid high and have high CTR banners – combination of these two will give you a high Score (AD Rate). This alone is not a problem, but the strong preference of such banners is.
In order to get solid volume from a CPC network, you need to be the one with the highest score. Being the top served campaign equals to faster banner burnout, which equals to fast CTR decline, which equals to a falling Score … these banners tend to be short lived.
We all know that the first few impressions generated by users always yield the highest CTRs, and since the CPC algos strongly prefer the banners with the highest CTR, they also serve them as first the most. This inflates their CTR in a way, and makes it almost impossible for the other banners to get solid CTRs too. In other words, if you can't reach high CTR during the casting, there is almost no chance for it to improve later on.
So whenever a competitor takes the 1st place away from you, the CTR of your banner will drop significantly and even a higher might not help you. That's why you need to monitor CPC campaigns more and react quickly. Whenever you see a sudden drop in traffic, quickly raise the bid to reclaim your position … or kiss the banners goodbye, since they will loose CTRs in the coming hours
ISSUE #3 – Handling Competition
As I already mentioned, with CPC networks, there is usually one winner who takes bulk of the traffic, while the distribution is more even in case of CPM networks. In case you're a banner ninja, it will be easier for you to slay the competition in a CPC network. On the other hand, if making sick banners isn't one of your strengths, you have better chances to get solid volumes with CPM.
Fighting for volume with competition in a CPM network is pretty much just about the bids, while in a CPC network, bid is just one part of the equation. Without a strong CTR banner, you don't stand a change, unless you are willing to bid VERY high, which you don't want to do in most cases.
ISSUE #4 – Frequency Cap
With CPC networks, you can't usually set a frequency cap, it's all on the algo to decide. If it thinks it's a good idea to show the same banner to the same surfer 10 times in a row … it will just do that It's a known fact that the 1st few impressions produce better results in terms of conversions, so many media buyers set their caps to 1/24h for example.
With CPC, it's not possible, so again, it's forcing traffic on you that you might not want. With a CPM network, this is easy to control and use to your advantage. Set whatever CAP you want and the network will deliver just such impressions to you.
WHICH ONE IS BETTER?
Whenever possible, I choose CPM over CPC, I simply find it easier and more comfortable to work with. It's easier to test a lot of banners with the CPM model, banners usually last longer too, I can get a fair amount of traffic without fighting for the TOP spot …
The only time I work with CPC is when the network doesn't offer the CPM option I know that the future will probably bring more AI into the world of media buying, but for now, I prefer to rely on my skills and not some algo that will pick the creatives for me. Just sell me the traffic and let me do my work
My clear winner is CPM bidding model and particularly the SMART CPM variation, so let me explain what that SMART part means actually :
When using the classic flat CPM model, you will pay exactly what you bid. If I set my bid to $1 per 1000 impressions for example, the network will happily charge me that amount, even thou my closest competitor is only bidding $0.30. Good for them, but not for me
That's why I prefer the SMART model, which would charge me about $0.31 by using the previous example. Smart bidding model uses your BID as the maximum, while the real amount charged will be just slightly higher than what your closest competitor is bidding.
This works very well when you want to aim for #2 spot for example. Simply bid extremely high, check what you are billed, then bid a bit below it. Or maybe you want to keep #1 spot? Bid significantly above the current top bid, the SMART algo will make sure you're not overpaying much and whenever a competitor raises the bid, the SMART algo will simply start to charge you more… Smart CPM is cool, always use it!
As I mentioned at the beginning, we have more threads/articles that deal with bidding. Check them out :
Caurmens classic series on bidding first :
1. https://stmforum.com/forum/showthrea...-The-Landscape
2. https://stmforum.com/forum/showthrea...ing-Techniques
3. https://stmforum.com/forum/showthrea...he-Competition
And one more from me, comparing bidding very high VS very low : https://stmforum.com/forum/showthrea...oach-is-better
Thanks for reading, comments welcome
Just a short clarification, based on a question I received via a Direct Message - all the above text relates to BANNER traffic!
Good info man, thank you.
good article! thanks!
Originally Posted by vincent9
Would you also suggest going for SmartCPM on Pop traffic?
Originally Posted by plamen
Originally Posted by chris_climbs
Originally Posted by matuloo
My creatives CTR is about 0.03% =(
Spent about $20.60 and got 41 clicks
Running on Premium network Exoclick
Normally I just bid 0.08 CPC though!
Is there anything I'm doing wrong?
Originally Posted by pavel_apostolov87
Whatever the case, the problem isn't the smart CPM system itself. Something was wrong with either your banners or the campaign setting ... or poor spots. Hard to say without knowing more details.
I am running in belgium, desktop
the minimum CPC according to exo is 0.05 cpc, I bid 0.08 cpc
These are my stats from exo platform
my daily spend
general stats from exo
My banners were taken from adplexity, results from last 7 days, filtered to "most traffic sent"
These are my stats for my banners
Exo campaign settings
Network selection: premium
partner networks: disabled
zone targeting: all
I didn't realize my banner CTR was so bad. I thought if I had several on rotation, then that would be the "average" ctr that I was getting... 0.02%
How do you choose the spots? I assume it is by "Zone"?
How I optimize the campaigns, I will usually optimizes by Banners and Domains straight, sometimes I will optimize by Category
What other details do you need to decipher what's wrong?
Couple things here :
- Belgium is not a tier-2 GEO, its prime target EU country ... but that doesnt matter much
- I see a banner in your mix that mentions Netherlands, which is not good. On top of that, all your banners are in Dutch, while large part of Belgians speaks french.
- not sure about the banners either, since you posted a screen from just the CPC campaigns. CTR with cpc campaigns on exo is always low, since you are basically getting just the traffic nobody wants to buy at CPM.
Please start a new thread for this in the adult section and link it here, this will be a longer discussion so I dont want this thread to be hijacked by it
Great insights, nice detailed breakdown.
Thanks for letting us ''spy'' on your epxertise! #matuloo
Home > >