Sorry everybody for the lapse in posts for the last month.  To makeup for it, I have a couple case studies coming…

I was looking back over some of my previous case studies in preparation for a new campaign I wanted to test out on Plenty of Fish and decided to start a new case study.  It will be extremely similar to the previous case study where I used POF’s Conversion Tracking Pixel to find out which users converted.  Except this time I’m going to figure out who doesn’t convert and make sure those users aren’t included in my next round of testing.

I’ve decided to use True.com from Revenue Ads.  It’s an offer that has performed very well for me in the past on multiple traffic sources.

Since this offer accepts 18-24 year old traffic, I’m going to target single, divorced, widowed and separated males who are 18-24.  I’m not going to split test which ad copy converts the best.  I’m trying to weed out people on POF that do not convert.  So I’m going to go with the ad copy that has worked the best for me in my previous two case studies.  The headline will be “Want a Girlfriend?”  The body will be, “Single women in {state:your state} are looking for guys. Sign up and meet them tonight!”  I’m also only targeting users who have a Login Count between 1 and 50.  The images I’m going to use are ones that have proven to garner a high CTR for me.

Last weekend saw the final phase of my case study using POF’s conversion tracking to help turn a campaign profitable. Unfortunately, Phase 1 provided some skewed results because I tried to split $500 in testing between men and women and I only was able to spend $250 testing each gender. Simply put, that wasn’t enough and I realized it after analyzing the results. But, I took what data I had and went with it. Since I learned that $250 wasn’t enough testing, I decided to test the results of only the male gender because they provided the most statistically significant results. The campaigns didn’t lose as much money as the first round and none of the campaigns were profitable, but I’ll explain why it was my fault and what you can try to turn a similar campaign profitable.

Targeting Criteria: Body Type – Average

Analysis: On my previous case study an average body type showed a conversion rate of 5.56% which was higher than average for the entire campaign in the testing phase. On this final phase these numbers held true and the conversion ratio even increased a bit. This campaign had the second most amount of traffic. The CTR was the highest of any of the other campaigns. If you were going to start a campaign, I would definitely include this targeting criteria in your campaign.

Targeting Criteria: Drinking Habits – Socially

Analysis: When I did my original test campaign, this criteria converted right at the same percentage as the overall campaign.  In this second go round, it didn’t fare so well and performed well under what I expected to.  This target criteria by far had the most amount of traffic available.

Targeting Criteria: Education Level – Bachelors Degree

Analysis: This target criteria performed the best in the initial testing stage converting at 7.61%, but it only had 7 conversions.  I had high hopes for this campaign and a little bit of worry because I wasn’t sure if the conversion ratio was a fluke.  And as you can the conversion ratio didn’t perform as well as I had hoped.  After the testing, it converted just under the average ratio for the campaign.  I’m also pretty sure the target demographic for this criteria is pretty small, hence the CTR dying out on all my images really quickly and most images not even being clicked on after a few thousand impressions.

Targeting Criteria: Income – $35,001 – $50,000

Analysis: This is another target criteria that was pretty risky.  In my initial tests it only had 7 conversions, but it converted at 5.22%, well above the campaign average.  After sending a fair amount of traffic to it, it converted even better than in my initial tests.  The only thing holding this campaign back from being profitable was the CTR.  I struggled with CTR in this campaign for a reason unknown to me.  I used what I thought were some of my best images at this campaign without any luck.  So if you know you have some images with really good CTR, this is something you might want to build into your next campaign.

Targeting Criteria: Search Type – Long-term

Analysis: This criteria converted right at the campaign average and I wanted a fifth criteria to test so I threw it in there.  Unfortunately, the campaign didn’t convert nearly as well as it did in my initial tests, which could be contributed to it only having 6 conversions.  These users seemed to be pretty click happy as it had the second best CTR of all the campaigns, they just didn’t convert nearly well enough.

Final Thoughts

The day after the campaign I received an email from Convert2Media that informed me the payout for Singlesnet 25+ was raised to 4.50. So you can take all the revenue figures here and increase them by 12.5%. It still doesn’t make any campaign profitable, but it definitely helps out quite a bit.

It would really help if Plenty of Fish could include the estimated number of people in our target demographics to give us an idea of how long we can expect our images to last before the users succumb to banner blindness.

One of my biggest problems with these final campaigns was choosing the correct images.  When I was running these campaigns I had numerous other campaigns I was working on so it was tough to keep an eye on these campaigns, even while using Mr. Green‘s POF Tool to upload new images throughout the campaign’s livelihood.  If you can keep your CTR up on your campaigns you will be way closer to profitability than I was.  I also have to fault myself for not split testing images.  I rarely ever do this and is one thing I need to severely address.  I’ve done this a few other times and I do know that images to play a role in the conversion rates.

There are a number of things you can take away from this case study.  By looking at the numbers above you can clearly see two targeting criteria that convert better than others, so try building those into your campaigns.  Maybe you can even combine the criteria and see how they perform together, but that will limit the amount of traffic you can get it.  Or you can do it the other way and exclude poor performing criteria from your campaigns.

If you use any of the tips I mentioned here and don’t mind sharing in the comments below how it turned out for you, please do!

I spent $550 and tried to split the amount between the male and female campaigns for Singlesnet.  I’ve included a copy of ad copy split testing results and the conversion breakdowns for both, males and females. For the campaign targeting males I spent a total of $300.89 for 26 conversions at $4.00, resulting in a loss of $196.89. When it came to females I spent $254.43 for 21 conversions at $4.00, which also resulted in a loss to the tune of $170.43. These kinds of losses can be expected when you do really broad targeting to a really broad dating offer. My goal here was to try and find certain target criteria that converted better than others and I did just that, which I’ll talk about shortly.

Here you will see the ad copy split testing results and conversion breakdown for criteria with more than 5 conversions for the male gender.

As you can see from these results, I will be using the, “Want a Girlfriend?” ad copy, paired with a number of images that did well for me on this campaign. I will also be targeting men who have a bachelor’s degree, average body type, income from $35,001 – $50,000, drink socially and are looking for a long term relationship.  I will build out separate campaigns for each of these 5 criteria.  The rest of the targeting will be exactly like it was in Phase 1, except that I will only be targeting 25-29 year olds and leaving out the 30-32 year olds. They didn’t convert nearly as well as the 25-29 year olds.

And below you will find the ad copy split testing results and conversion breakdown for criteria with more than 5 conversions for the female gender.

Again, we see that the ad copy that performed the best was, “Want a Boyfriend?”. There were no obvious criteria that performed better than the others, so I’m going to leave the female campaign alone and focus solely on the male campaigns since it had more useful data.

I will be building the aforementioned campaigns tonight. Once they blow through the $500 in the coming week, I will make a follow up post with the results.

Please note that I didn’t pause any ad copy on purpose. I only checked the stats once a day and any images with .09 %CTR or lower was paused. So that is probably what contributed to the large discrepancy in the number of clicks each ad copy received.

Plenty of Fish has recently implemented conversion tracking into their ad platform.  This will show you the exact characteristics of your converting users.  You can use these reports to take a non-profitable campaign and turn it into something that will help you regain the money you lost testing and ultimately turn a profit.

This case study will take a real general dating offer and advertise it to the masses.  I will take the data from their conversion tracking report and break it into ultra-targeted campaigns.  I decided to go with the Singlesnet 25+ offer on C2M.  I’ll be using some pretty generic ad copy that has been proven to work and will give them away below.  The images I’m going to be using were gathered by me personally and I have a feeling that they will do quite well on POF so I won’t be outing those, sorry fellas.  The frequency cap will be set to 3 and my bids will start out at $0.46 using Mr. Green’s 1 Penny Tip.

Targeting Criteria

1. Single, widowed, divorced or separated Males that live in the United States and are between the ages of 25 and 32.
2. Single, widowed, divorced or separated Females that live in the United States and are between the ages of 25 and 32.

Male Ad Copy

1. Single Available Women!
   Are you looking for more single women in {state:your state}? Sign up now and find them tonight!
2. Single Women!
   Are you looking for a single woman in {state:your state}? Sign up now and find her tonight!
3. Want a Girlfriend?
   Sign up now and find a girl in {state:your state} tonight!

Female Ad Copy

1. Single Available Men!
   Are you looking for more single men in {state:your state}? Sign up now and find them tonight!
2. Single Men!
   Are you looking for a single man in {state:your state}? Sign up now and find him tonight!
3. Want a Boyfriend?
   Sign up now and find a man in {state:your state} tonight!

If you advertise on POF and aren’t aware, Mr. Green has released a POF Ad Uploader that is effin awesome!  It will save you tremendous amounts of time uploading ads and split testing campaigns.  It comes with free lifetime updates and is well worth the money.  Here’s a link to my walk through video.

As promised in Plenty of Fish Case Study: Session Depth and Plenty of Fish Case Study: Session Depth (repost) I would do a case study where I would test the conversion ratio against the POF user’s Session Depth.  After letting this campaign run for a week I was very disappointed with the results of this case study and I’m sure you will be too.

At first I wasn’t getting enough data in the campaigns with the higher session depth counts, so I was constantly raising my bids in hopes of more traffic.  Even after a full 7 days of running this campaign I didn’t have enough test data and was tired of running the campaign.

After the initial test run it seems my better images have lost their touch and aren’t getting nearly the CTR they used to.  Some of them were less than .10, but for the integrity of the case study I left them running.

The conversion ratios appear to increase as session depth increases but I wouldn’t even come close to saying that the data below is significant after the first thirty impressions.  Probably not even after the first twenty.  And I think the 7.82% increase in conversions between the first and second sets of data may be nullified after sending the same amount of traffic to the second one.  It would probably be best to do another case study for the first twenty impressions, but if you want that data, you’ll have to do the testing yourself.

The only thing I found useful in the case study is that the highest CTR came from the users that had first logged in