Media Buying, Ask Me Anything



Neither, everything I learned about SE0 I was taught from Subway. Make sure your lettuce is fresh and your tomatoes are distributed evenly and you will see success.

Will read this again. You just opened up my mind to a whole new world in SE0. Shit man... I always thought it was the olives.

Another question unrelated to SE0, before the thread gets derailed.

1) What are the average CPMs people (aff and not branders) are bidding on for US/CA/UK/AUS traffic.. by demographics.. Do people make a profit on a regular lead gen on a $2 CPM, without upsells and such?

2) Where do people go for making a ton of banners, quickly.
 
Will read this again. You just opened up my mind to a whole new world in SE0. Shit man... I always thought it was the olives.

Another question unrelated to SE0, before the thread gets derailed.

1) What are the average CPMs people (aff and not branders) are bidding on for US/CA/UK/AUS traffic.. by demographics.. Do people make a profit on a regular lead gen on a $2 CPM, without upsells and such?

2) Where do people go for making a ton of banners, quickly.

1) I have no clue, obviously they change with the traffic quality. CPMS don't matter much in media buying at the end of the day, what matters is your CPA. If your CPA is lower than the CPA you're being paid you're making money. Higher CPMs have backed out for me and lower CPMs have tanked....

2) I had a designer on staff who I put on retainer. The designer always picked up my work first and got it done ASAP. Besides that, paint....
 
If you had the ability to combine one thing from subway and one thing from your favorite mexican place, what would they be and why?

Foot long Jalapeno Chicken on Honey Oat, with lettuce, tomato, cucumber, green peppers and Jalapeno sauce. Nuff said (cookies on the side too please)
 
1) I have no clue, obviously they change with the traffic quality. CPMS don't matter much in media buying at the end of the day, what matters is your CPA. If your CPA is lower than the CPA you're being paid you're making money. Higher CPMs have backed out for me and lower CPMs have tanked....

2) I had a designer on staff who I put on retainer. The designer always picked up my work first and got it done ASAP. Besides that, paint....

1) I know that CPMs don't matter. I was asking the general CPMs that "affiliates" are bidding on. I own bunch of sites, and I have always got great CPM rates from Networks.. I can't for life get any affiliate offers on those banner spots, since the rebill and farticle days. That is what I was asking..
 
It really depends how many imp's your getting and the offer right. If you have a high payout offer, you need to send it more traffic. Let's say you have a $1 cpm for a general news site. If you have a $3 dating offer, you could rotate multiple offers (lets say 3) and multiple creatives (lets say 5), with 10,000 imps per, so you'd need 50,000 imps a day to test that and get real data. Let's say the payout is $30 though with the same stats you'd need 500,000 imps a day to get relevant data in my eyes on that. Make sense? It's all relative.

Can you elaborate on this in more detail?

Why the 100,000 imps per creative for $30 payout vs 10,000 imps per creative for $3 payout?

I realise that with the same number of conversions, it then makes the eCPM the same (conversions x 3/10000 x 1000 = conversions x 30/100000 x 1000). This means for the $3 payout, you need more than 3 conversions in that 10,000 imps (which cost $10) to break even. And for the $30 payout, you need more than 3 conversions in that 100,000 imps (which cost $100).

Also, how do you arrive at that figure of impressions for the data's relevancy? 3 conversions from a test seems too small to have any statistical confidence with and could easily be anomalies.

Edit: Should be "you need more than 3 conversions in that 100,000 imps" not 33.
 
Last edited:
Great responses so far guys. Dropping knowledge.

Let's say you have been profitably spending 250k/month on Google Search for 5+ years. Your ads lead to your own in-house lead-generating pages monetized by a combination of strong LM/Coreg partners. You have slowly expanded and tested every logical angle within your space and are finally starting to hit a plateau on scale. Now you decide you want to scale with the help of Display/MediaBuys.

What is the best way to leverage your 6-years worth of paid-search data in your niche to efficiently scale to Display. Let's assume you have access to Google Display, Yahoo/MSN/RIGHT, and most DSPs, each with 6-figure test budgets.

Also have a WhatRunsWhere + AdBeat account at my disposal if you have specific process suggestions for going from Search -> Display.
 
blogspotter: It really depends, affs can't compete with brand cpms, which is why we direct site buy, or buy via network (on the larger brand croweded networks, a lot sell you remanant traffic)

top: No specific color, different colors work for different niches there's a ton of psychological findings about colors and their emotional linkages, I'd check those out

nethbert: well, I said that as a rule of thumb for what I'd use to TEST creatives, not pick ones that win. By giving 3x CPA I get an idea if a banner is converting. I never said you'd get 3 conversions per banner. Some will have none, others will have one and some may have 5. It gives you an idea which are busts and which are winners (and which shows promise), while giving you enough data (in my mind) to make a decently informed decision to cut an ad while giving it a fair chance.


dt808: Well you'd have a good idea of your demographics and a great rep with google. You could launch a content network campaign to start to find direct sites that you can approach to buy on. You can test banner creatives quickly and cheaply by paying CPC. You'd also have an idea of what type of wording you should start working with.

For WRW: plug in your competitors see what they're doing for display, check their demos and demos of sites they're advertising on. Test a few out, find what's working and scale by inferring relationships between the type of site or demo and whats working.

gohard: no we do not at this time.
 
nethbert: well, I said that as a rule of thumb for what I'd use to TEST creatives, not pick ones that win. By giving 3x CPA I get an idea if a banner is converting. I never said you'd get 3 conversions per banner. Some will have none, others will have one and some may have 5. It gives you an idea which are busts and which are winners (and which shows promise), while giving you enough data (in my mind) to make a decently informed decision to cut an ad while giving it a fair chance.

Just trying to understand. I know it's about testing, and I wasn't suggesting that there'd actually result in a specific number of conversions.

I'm wondering how you determine what is enough to feel confident that the conversion rate is actually representative to extrapolate from the sample to a larger buy. Is it something you just settled on from experience?
 
It's just something from experience and gut. Some people want 5x, some people want 2x CPA. I like 3x, I feel it gives me enough data while not spending so much of my budget that I'd spend mid to high xx,xxx just testing to "make sure"
 
It's just something from experience and gut. Some people want 5x, some people want 2x CPA. I like 3x, I feel it gives me enough data while not spending so much of my budget that I'd spend mid to high xx,xxx just testing to "make sure"

And my final (hopefully :) ) question to make sure I understand this process. In this context you say 3x CPA, is that saying roughly a spend of 3x $3 payout? Meaning $9, which means 9,000 impressions (bumped up to a round number of 10,000)?
 
I do not, the way I used adbuyer was to vet right media networks. You can select manual targeting and use it to generally test networks, when you break down networks and see which are performing, then cut out adbuyer and go direct. Make sense?

Indeed. +rep to you. Off to run some traffic :)
 
What's an ad server? When would you need one? And how much do they cost?

Along the same lines as this. Did you have an adserver from the get go and have you seen a big discrepancy in billing #'s between your own adservers #'s and the #'s of the network?