1st Campaign on Facebook (32)

Finally time for my first campaign. Hopefully you guys can offer some insight as I make my way through all this...

Traffic Source: Facebook
Ad Type: Right Hand Side - Clicks to Website
Linking Type: Direct Linking
Tracking Tool: Prosper202 self-hosted on BeyondHosting VPS

Offer Vertical: Health Services / Health & Wellness
Payout: $7.60
Network: MundoMedia
Offer Notes: Can only be run Monday-Friday, no weekends

Targetting: 1 entire state + 2 major cities in the USA. English (all). Relationships (all).
Potential Reach: ~12.5 million
Suggested Bid: $.30 - $.65
My Bid:$.30
My Daily Budget: $20
# of Ads in Campaign: 10

I dug around Social Ad Ninja and found an offer that seems to have been around for quite some time. There are several ads that have run within the past 24 hours and have run for 10+ days. I learned that Mundo had the offer, signed up with them, and got put on the offer.

I decided on an angle and took a close look at the types of images being used for that angle or similar angles. With that idea in mind, I grabbed 42 images that were similar. From there, I opened up photoshop to resize, brighten, sharpen, and crop each image.

Tracking was set-up in Prosper202. I used Caurmen's Simple Guide to Prosper to help me along.

My first hiccup occurred when I was trying to create my first batch of ads in Facebook. I entered my tracking URL as the URL I wanted to promote. There were two issues with this:

1. I didn't want my tracking URL appearing on my ads, and
2. Facebook wasn't allowing me to promote my tracking URL

1. How many impressions without conversions should I allow for each ad variation before pausing it?
2. What's the rule of thumb for pausing an ad that has a few conversions, but isn't profitable?
3. If I pause 1 ad, should I immediately replace it was a new image variation?
4. With a $7.60 payout, how many clicks/conversions do I need to reach statistical significance?

Subscribed Go for it!

Good luck..

I'd bid higher than the minimum though, I'd bid at the high until you see some of your ads getting high CTR, and then it will naturally get a lower cpc as Facebook adjusts it..

12,5 million people in a target group is pointless. Too big with zero effect. Such big target groups should only be used with oCPM model so Facebook's algorithm can pick a working targeting pattern for you. Split your campaign based on these targeting options: geo, sex and age. Size of target group up to 100-200k people.

Else you will never be able to track results and evaluate what's working with such a big target group and small budget.

Originally Posted by dynamicsoul

Good luck..

I'd bid higher than the minimum though, I'd bid at the high until you see some of your ads getting high CTR, and then it will naturally get a lower cpc as Facebook adjusts it..

Set a different keyword in your URL for each ad - that way you'll be able to see per-ad data by looking at the keyword view in Prosper.

Search for the string "If you're using a traffic source that uses banners," in the Prosper 202 tutorial - that section talks about setting per-ad keywords.

Thanks for the info guys!

Not getting any impressions after 1.5 hours. Might just delete the campaign, start a new one, raise bids, fix the targetting as suggested above, and fix the tracking.

Any other suggestions are welcomed!

I don't run facebook, but when I don't get much clicks I raise my bid (double it) then wait and see results.

I'm getting to a point though where I'm just gonna bid ultra fucking high to get clicks ultra fast to see if the angle im testing is even worthwhile. So rather than waste time bidding low, then doubling my bid, I bid sky high, collect data fast. Then optimize.

Maybe this can work too? (Again I know jack about FB)

Ok....so I learned the whole tracking setup and facebook setup process last night. It's like second nature already. Got the second campaign up and approved in under an hour.

Targeting: 1 state, Men, 30-35, College Grad, English
Potential Reach: 300k
Bid:$.52

Hopefully that gets me on the board. My targeting options were based off of data I pulled from Social Ad Ninja.

Still kind of confused about the individual ad tracking. When I go to FB ads, I choose create an ad for clicks to website. I enter the URL I'm going to promote. From there, I create my ads and demo targeting. Below that, I can set the campaign name, campaign run time, and max bid. I don't see anywhere where I can set an individual URL for each ad. What do you guys suggest?

Originally Posted by magnetic

12,5 million people in a target group is pointless. Too big with zero effect. Such big target groups should only be used with oCPM model so Facebook's algorithm can pick a working targeting pattern for you. Split your campaign based on these targeting options: geo, sex and age. Size of target group up to 100-200k people.

Else you will never be able to track results and evaluate what's working with such a big target group and small budget.

I don't know about universal formula, but this is how we do it. We run few hundred campaigns per day, monthly budget is over 100k.

Every campaign is based on these parameters:
1, One country per campaign, never more.
2, One age group per campaign, never more.
3, One sex per ad campaign, never more.
4, One specific like set per ad campaign, never more.
5, Up to three creative per campaign.

For example:
TS 012541 US 35 M (35-44, oCPM casino_test_NF) where TS 012541 stands for our internal campaign ID (for further evaluation), US is country, 35 is starting age and M stands for Males. Model is oCPM and the target audience has some casino game among their interests.

Every campaign has a daily budget 10-20 USD and in case it's performing well, we increase budget and eventually after some time we refresh it as it naturally decays. If there is little to zero amount of conversions after first two days, we either kill the campaign completely or try to refresh it with slightly modified settings. Great metric is also ad frequency, if it's above 5 and conversions are low, we kill the campaign. And so on, we use plenty of different metrics. Also, we track ROI and we have various prediction models that are estimating days until some campaign will be paid off, even it's still running.

Some time ago we used to handle this via Power Editor + our internal analytic tool, but now we are using 3rd party tool called SocialMoov and our own prediction and analytic tool. There are various tools from various PMDs focused on different areas, however you need to have certain volume of ads and revenue to take a full advantage of them.

We have also some advanced features for targeting since we have close relationship with guys in Facebook. For example we are able to target campaigns on people who actually used their credit card in the last two weeks and paid something via Facebook (virtual goods, gift cards, etc.). Custom audiences and look-alike audiences are also very powerful, learn how to take advantage of them.

And yeah, those steps I described above might look a bit complicated for a small scale - and they actually are. But maybe you will find something useful and will be able to use it in your own campaigns.

Originally Posted by magnetic

And yeah, those steps I described above might look a bit complicated for a small scale - and they actually are. But maybe you will find something useful and will be able to use it in your own campaigns.

great post by magnetic, that system works..

also, don't go by what you see on social ad ninja in terms of profiles. A man aged 30 from New york might have seen the advert, but there are a bunch of other targetiing options that could have been made which SAN will not show..

It can give you a good idea on types of images for age ranges and geos, but that's about it

Originally Posted by dynamicsoul

great post by magnetic, that system works..

also, don't go by what you see on social ad ninja in terms of profiles. A man aged 30 from New york might have seen the advert, but there are a bunch of other targetiing options that could have been made which SAN will not show..

It can give you a good idea on types of images for age ranges and geos, but that's about it

Ok, so the campaign has been running for about 4 hours now. Facebook is showing NO DATA whatsoever, but Prosper and the Affiliate Network have both logged about 20 clicks! What the heck is going on. Does FB take a day to update the data?

Ok. Been running for almost a full day now.

Here are my stats from FB:



This was for 15 RHS ads.

As you can see, there aren't many impressions. I have a feeling this has to do with my bidding.

Here is/was my thought process:

• FB suggested bid was $.30-$.60
• I bid $.3 and received no impressions for 2 hours
• Noticed the suggested bid changed to $.45 - $1.20, increased bid to $.50 and still didn't get any impressions
• Noticed the suggested bid changed to $.80 - $1.60, increased bid to $1.00. Impressions started flowing and received 2 clicks
• Impressions stalled around 11pm and haven't started again
• As of this morning, bid suggestions are back down to $.30 - $.60. I'm not getting any impressions, even though my bid is still at $1.00

yeah, don't dick around with it..

Set up a camp with 3 of the same ad/image in each camp (have each image into campaigns on their own).. set the bid high, one of them after a few thousand imps will start to get the traffic.. delete the others after the best one has maybe 10 or 15 clicks.. around this time it starts to level out the cpc... but if you don't have high CTR ads.. you won't get low CPCs..

See, here's one I launched this morning. All images in this campaign are the same.. same headlines etc.. Just one will get the clicks.. and others won't.. I set the bid at the high range and get the low cpc cos it gets clicked often..



It will probably level out to 0.15 or 0.2 CTR as the day progresses...

Oh wow, so you're doing 3 identical ads per campaign? Very interesting. So, I should literally have 15 campaigns, 1 for each image. Going to set that up now.

EDIT***

I just tried adding 3 identical images and it wouldn't let me. Are you using Power Editor?

Didn't get much done yesterday. Had to visit the tax man and submit my pound of flesh.

After talking to a few people about my issues, it looks like I'll be creating 15 campaigns - each with 3 identical ads in them. This is to "trick" the platform into giving each image variation a good amount of impressions. Targeting is going to be very broad at first - slowly optimizing and breaking demos down until I've targetting a few converting demos in 100-300k potential reach range.

The offer has to be paused over the weekends, so I'm going to try to get these up quickly so I can get some data today.

I was able to run my campaign for about 8 hours on Friday. It was comprised of 10 campaigns, with 3 identical ads each. At the end of the day, I only had about 200-300 impressions per campaign. I had set my daily budget to the max ($250) and bid $2 (suggested was $.68). So, I don't really know why I wasn't getting any impressions. My only guess is that I wasn't getting any impressions due to the fact that I had too many ads running with a low max daily adspend. $250 / 30 ads = $8.33 per ad per day. I'd imagine FB would give the impressions to an account that has a much higher daily spend potential per ad.

With that theory in mind, I'm going to try a much more narrow demographic with max day spend and a high bid. Let's see where that gets me!

Odd that you didn't get many impressions at a $2 CPC bid - could be that the demo is just very pricey. Take suggested bids as indicative but overly useless. I just ignore them, honestly. If you know a demo well you know the bids well too. If it's a new demo just start at something moderate, e.g. $1, wait a few hours, bump to $1.50, wait, etc. Bid high budget low - i.e. feel for the bid floor with a campaign set to a $10 budget so you have little to lose in the process.

FYI the 3 images per campaign approach is to get ads to 'stick'. How well an ad initially performs can have a big impact on subsequent performance - an early spike of clicks leads to a high CTR > more delivery > more clicks > CPC drops > ad may drop in CTR or remain OK and 'stick' - yay! Other ads will have less favourable starts. We eliminate these and let FB focus on the ad it has blessed with divine luck.

Also, try cloning a campaign and running it in oCPM bidding mode @ default bids. Only one ad needed in that campaign.

Good luck, will be following!

Thanks! Suggested bids are well below a dollar during the day, and go up around 5pm-9pm to about $1.50. I figured a $2 big and a $100 daily budget would be plenty to get eyes on the ads, but I was wrong.

I totally understand what you're saying about the 3 images per campaign and pausing the non-performers. That was the goal - I just wasn't getting the impressions.

Originally Posted by zeno

Odd that you didn't get many impressions at a $2 CPC bid - could be that the demo is just very pricey. Take suggested bids as indicative but overly useless. I just ignore them, honestly. If you know a demo well you know the bids well too. If it's a new demo just start at something moderate, e.g. $1, wait a few hours, bump to $1.50, wait, etc. Bid high budget low - i.e. feel for the bid floor with a campaign set to a $10 budget so you have little to lose in the process.

FYI the 3 images per campaign approach is to get ads to 'stick'. How well an ad initially performs can have a big impact on subsequent performance - an early spike of clicks leads to a high CTR > more delivery > more clicks > CPC drops > ad may drop in CTR or remain OK and 'stick' - yay! Other ads will have less favourable starts. We eliminate these and let FB focus on the ad it has blessed with divine luck.

Also, try cloning a campaign and running it in oCPM bidding mode @ default bids. Only one ad needed in that campaign.

Good luck, will be following!

FINALLY GOT SOME DATA!

Ok, so I went overboard with my spend today to get some data flowing in. I went super broad with targeting. I did this to get some clicks coming in and to make sure that the reason I wasn't getting impressions was due to low bidding and not a soft ban.

I ran 10 campaigns. Each campaign had 3 identical ads in it. Once one of the ads in each campaign gained a majority of the impressions, I paused the other two.

2/24/14: Spent $88.12 and Made $15.20 on 2 conversions.

Here's a screenshot of the campaign performance:



CTR was predictably horrible. Again, this was a test to gain insight on which sex, age, and geo demographics I should be targeting.

Here's a screenshot of my network stats for the day:



Again, poor performance was expected due to my broad targeting.

Here's the good stuff:



In the image above, you can see the top 10 performing ads out of the 30 ads I ran for the day. Using the 80/20 rule, I'm focusing on the top 20% to determine which sex and age demographic I should zone-in on for the next iteration of the campaign. Using that logic, Females aged 45-64 are the clear winner.

I also hit up my AM for a list of top performing geos for the offer. On his list, the top 9 performing campaigns were targeting the entire United States. This is obviously way too broad for my budget, so I looked to the best performing city, which ended up being Chicago.

So, here is my new demo that I will start targeting tomorrow:

Females, 45-64, English (All), Chicago Illinois

Setting up a dummy ad on FB showed me that this demo has a potential reach of ~400k.

I'm still trying to figure out how to track each add individually. Once I'm able to do that, I can finally run this demo for a few days while slowly cutting non-performing ads and pumping the ones that are bringing in the conversions.

What do you guys think?

Glad you're making progress. When you say Chicago was the highest performing city what do you mean - is this based on your AMs info or from your responder demographics on FB? If the latter then how many impressions/clicks did you have for each city?

For individual ad tracking you need to customise the URL of every ad. This is likely easier to do via the power editor so I recommend using that. For example:

Advert 1 URL: hxxp://mydomain.com/offer.php?c1=US&c2=45-64&c3=Ch&c5=ad1img1
Advert 2 URL: hxxp://mydomain.com/offer.php?c1=US&c2=45-64&c3=Ch&c5=ad1img2
Advert 3 URL: hxxp://mydomain.com/offer.php?c1=US&c2=45-64&c3=Ch&c5=ad1img3

Then for an advert with completely different ad copy, perhaps:
Advert 1 URL: hxxp://mydomain.com/offer.php?c1=US&c2=45-64&c3=Ch&c5=ad2img1

This string of text after the ? is called a querystring. We use them to pass text values between pages and can then use them inside the page via PHP, javascript, etc. For your page you said you had a meta-refresh redirect? You can do better. A PHP redirect script is much faster and accomplishes the same thing. What you want this page to do is read all the querystring values and then pass them on to your prosper202 link - literally appending them to the end. Prosper202 expects querystring data in the form of c1=something&c2=somethingelse&t202kw=mykeyword. The latter is the special keyword which comes up under the keyword heading in P202. All the others are c-variables. In the STM P202 mod you can easily look at all of these in group overview and drill down into your data.

Here is an example tutorial PHP script to help you get this sorted. Test it out before using it for live ads - edit to suit, upload it, then visit the URL and add things like ?c1=im_testing_yay&c2=blerg&c3=derp, check it redirects to the tracking link/offer, then look at your P202 stats to confirm everything is being passed properly. Juicyness below:

<?php
/* First we need to get the parameters passed with the URL after the ?. 
We do this using $_GET['name'] - here 'name' is whatever we had in the URL, i.e. subid1, subid2, etc. 
Then we are going to set them to be equal to a variable - here I just use the same thing, $subid1, $subid2, etc. 
Variables start with $  */

   $c1 = $_GET['c1']; //Subid 1
   $c2 = $_GET['c2']; //Subid 2
   $c3 = $_GET['c3']; //Subid 3
   $c4 = $_GET['c4']; //Subid 2
   $c5 = $_GET['c5']; //Subid 3
     
header('Location: http://myp202.com/tracking202/redirect/tracker.php?t202id=1234&t202kw='.$c5.'&c1='.$c1.'&c2='.$c2.'&c3='.$c3.'&c4='.$c4);  //P202 tracking link with tracking data appended
?>

You're really blowing my mind with your answers for the past few days bud!

Originally Posted by zeno

Glad you're making progress. When you say Chicago was the highest performing city what do you mean - is this based on your AMs info or from your responder demographics on FB? If the latter then how many impressions/clicks did you have for each city?

Originally Posted by zeno

For individual ad tracking you need to customise the URL of every ad. This is likely easier to do via the power editor so I recommend using that. For example:

Advert 1 URL: hxxp://mydomain.com/offer.php?c1=US&c2=45-64&c3=Ch&c5=ad1img1
Advert 2 URL: hxxp://mydomain.com/offer.php?c1=US&c2=45-64&c3=Ch&c5=ad1img2
Advert 3 URL: hxxp://mydomain.com/offer.php?c1=US&c2=45-64&c3=Ch&c5=ad1img3

Then for an advert with completely different ad copy, perhaps:
Advert 1 URL: hxxp://mydomain.com/offer.php?c1=US&c2=45-64&c3=Ch&c5=ad2img1

This string of text after the ? is called a querystring. We use them to pass text values between pages and can then use them inside the page via PHP, javascript, etc. For your page you said you had a meta-refresh redirect? You can do better. A PHP redirect script is much faster and accomplishes the same thing. What you want this page to do is read all the querystring values and then pass them on to your prosper202 link - literally appending them to the end. Prosper202 expects querystring data in the form of c1=something&c2=somethingelse&t202kw=mykeyword. The latter is the special keyword which comes up under the keyword heading in P202. All the others are c-variables. In the STM P202 mod you can easily look at all of these in group overview and drill down into your data.

Here is an example tutorial PHP script to help you get this sorted. Test it out before using it for live ads - edit to suit, upload it, then visit the URL and add things like ?c1=im_testing_yay&c2=blerg&c3=derp, check it redirects to the tracking link/offer, then look at your P202 stats to confirm everything is being passed properly. Juicyness below:

Code:

<?php
/* First we need to get the parameters passed with the URL after the ?. 
We do this using $_GET['name'] - here 'name' is whatever we had in the URL, i.e. subid1, subid2, etc. 
Then we are going to set them to be equal to a variable - here I just use the same thing, $subid1, $subid2, etc. 
Variables start with $  */

   $c1 = $_GET['c1']; //Subid 1
   $c2 = $_GET['c2']; //Subid 2
   $c3 = $_GET['c3']; //Subid 3
   $c4 = $_GET['c4']; //Subid 2
   $c5 = $_GET['c5']; //Subid 3
     
header('Location: http://myp202.com/tracking202/redirect/tracker.php?t202id=1234&t202kw='.$c5.'&c1='.$c1.'&c2='.$c2.'&c3='.$c3.'&c4='.$c4);  //P202 tracking link with tracking data appended
?>

<!doctype html>
<html lang="en">

<head>

	<meta charset="utf-8" />
	<title>some title here</title>
	<meta name="description" content="" />
	
	<meta http-equiv="refresh" content="0;url=http://PROSPERTRACKINGURL.com/tracking202/redirect/dl.php?t202id=9119&c1=Female&c2=55-64&c3=Chicago&c4=English&t202kw=">
	
</head>


<body>

<p>One moment...</p>

</body>

</html>

Originally Posted by stackcash

P.S. Any PHP training books/sites you recommend? I tried codecademy a few months ago and couldn't really translate what I was learning into real world use.

I learned everything I know from messing around and solving individual problems as they popped up, so I can't really recommend any particular learning sources.

OK, this should be pretty simple to get working if your vanity URL host allows PHP.

I assume you're just sending FB to http://www.yourvanityurl.com , with the meta-refresh you're using as the index of that URL?

Firstly, in your Facebook ads, change the URL to be

http://www.yourvanityurl.com/?adname=ADNAME

Customise the ADNAME for each ad - so, the first one should be adname=ad1 , the second one should be adname=ad2 , and so on.

Now, change your index page to be PHP rather than HTML.

Now, just edit your meta refresh code to be

<!doctype html>
<html lang="en">

<head>

	<meta charset="utf-8" />
	<title>some title here</title>
	<meta name="description" content="" />
	
	<meta http-equiv="refresh" content="0;url=http://PROSPERTRACKINGURL.com/tracking202/redirect/dl.php?t202id=9119&c1=Female&c2=55-64&c3=Chicago&c4=English&t202kw=<?php echo $_GET['adname'];?>">
	
</head>


<body>

<p>One moment...</p>

</body>

</html>

Thanks for all the help guys. Going to try this as soon as I get to my home computer.

Originally Posted by stackcash

FINALLY GOT SOME DATA!

Ok, so I went overboard with my spend today to get some data flowing in. I went super broad with targeting. I did this to get some clicks coming in and to make sure that the reason I wasn't getting impressions was due to low bidding and not a soft ban.

I ran 10 campaigns. Each campaign had 3 identical ads in it. Once one of the ads in each campaign gained a majority of the impressions, I paused the other two.

2/24/14: Spent $88.12 and Made $15.20 on 2 conversions.

Here's a screenshot of the campaign performance:



CTR was predictably horrible. Again, this was a test to gain insight on which sex, age, and geo demographics I should be targeting.

Here's a screenshot of my network stats for the day:



Again, poor performance was expected due to my broad targeting.

Here's the good stuff:



In the image above, you can see the top 10 performing ads out of the 30 ads I ran for the day. Using the 80/20 rule, I'm focusing on the top 20% to determine which sex and age demographic I should zone-in on for the next iteration of the campaign. Using that logic, Females aged 45-64 are the clear winner.

I also hit up my AM for a list of top performing geos for the offer. On his list, the top 9 performing campaigns were targeting the entire United States. This is obviously way too broad for my budget, so I looked to the best performing city, which ended up being Chicago.

So, here is my new demo that I will start targeting tomorrow:

Females, 45-64, English (All), Chicago Illinois

Setting up a dummy ad on FB showed me that this demo has a potential reach of ~400k.

I'm still trying to figure out how to track each add individually. Once I'm able to do that, I can finally run this demo for a few days while slowly cutting non-performing ads and pumping the ones that are bringing in the conversions.

What do you guys think?

Originally Posted by dynamicsoul

This post and outcome is your gold..
Now just find some ads that fit that 400k demo for best CTR.. think bold colors, try borders, no borders.. open up Google images and type in some relevant phrases, zoom out your screen so they are really small and run your eyes left to right.. any jump out at you? Maybe a place to start...

When you hit some profitable CTR ads.. try some other states same age range ..

Looking forward to seeing the new follow-along!