Newb POF Question: Pruning Ad Copies

[Alejandro]

I'm targeting a pretty healthy-sized demographic on POF with a dating offer recommended to me by my AM. I know this offer has been hot for at least 3 weeks. When my AM first mentioned it to me he said it had a network wide conversion rate of 12%.

My 1st round results look very different.

Here are my numbers
56 ad copies (2 txts x 28 images)
CPM = .52-.56 (this blew me away, I thought I'd be able to make do by undercutting the recommended CPM, but I guess not. I got like 10 impressions bidding <.52)
Impressions = ~60k over 7 days (Just over 1K/ad)
CTR = .199 (2 of my ad copies are doing >.7% CTR)

so far so good...
Clicks= 119
Conversions = ...2

A far cry from the 12% i heard about.

My AM suggested some tips for my ad copy for the next round of optimization, but here are some questions-

1. Should I be optimizing around CTR or CVR? One of my 2 conversions came from my highest CTR ad, and 1 came from a low CTR one - nothing to conclude there. However, if I prune ads based on CTR, i have some clear winners. 8/1080 beats 2/1080 by a 97% confidence interval?
2. I'm on a modest testing budget, how do you recommend I proceed into my next stage of testing? Collect more data from my pruned ads until I can whittle it to about 10 and then test different ad copies?
3. Should I keep trying to run this offer? AM says they may have started scrubbing? Cut my losses & move on to different offer?

I'm probably going to get flamed here, but hopefully a few of you will recognize that I'm trying to actually hustle here and not just troll forums all day for advice that I'll never use.

Thanks in advance for your input.

 

[WTFlux]

1k impressions/ad creative is not statistically significant.

If you have a modest test budget start with 5-10 creatives and optimize from there.
Make sure each creative gets approx. 10k impressions before making any assumptions.

You're on the right track though, good luck.

 

[MistorToker]

1. Both. Get some actual data - whatever splittest accelerator tool you used is for clicks-->actions, not impressions-->clicks. You've calculated your confidence interval incorrectly.

2. If you aren't already, make a landing page now. Since you're cash poor, try to be more efficacious and actually market w/ bridge pages that mimic the offer LP. The lazy guys who direct link get by with energetic daily spend and throw-and-stick testing. You can't quite make your money work for you yet, so you have to do the legwork in getting as close to the point of action as possible. If social advertising isn't going to be the final chapter of your days in affiliate marketing, and you need to make a quick buck, cloak adult on POF because they don't retro-disapprove (do NOT recommend this, POF guys are good people and you might not want to burn that bridge).

3. Someone is making your offer work and you are not being scrubbed - no advertiser is going to evaluate Alejandro's or Joe Schmo's aff id after 119 clicks. I seriously doubt a network-wide scrub coincided with your first campaign. In which case, you are simply shit at writing copy, but this is never a problem. Do not copy ads though because that is insincere and half of the fucks on POF don't know what they're doing (dynamic kw insertion idiots know who they are). Some general tips that everyone already knows are to capitalize every word of your headline and copy, question marks, first person, etc

 

[graffitimonkie]

Dating ads on POF have been beat to death.

And also watch your quality score on those offers, a lot of the copies with really loose copy might convert higher but in the end will hurt your quality score and cause you to lose an offer.

 

[Alejandro]

1. Both. Get some actual data - whatever splittest accelerator tool you used is for clicks-->actions, not impressions-->clicks. You've calculated your confidence interval incorrectly.

How should I be calculating confidence intervals?

According to 3 different tools, some of my ads are >95% confidence better CTR

The tool I used for the numbers in my OP is Split Test A/B Test Marketing Calculator - landing page-ppc-email split tests | SplitTestCalculator.com

If you put in 1080 in the first row for both and 2 & 8 in the second row & click calculate it says 97%

Next I tried Split Test (A/B Test) Calculator

This tool says that 2&8 are only 90% confidence, but 1&8 is 98%

A/B Split Test Calculator

This tool says 8 & 2 is ~90% confident and 8 & 1 is 96.4% confident.



Am I doing something wrong, or should I be using different tools?

 

[Alejandro]

2. If you aren't already, make a landing page now. Since you're cash poor, try to be more efficacious and actually market w/ bridge pages that mimic the offer LP.

Perhaps I'm not understanding something here, but how does a "bridge" page that "mimics the offer LP" help increase conversions? Won't the visitor think "huh? Why did that button bring me to another page that looks pretty much the same?"

Sorry, perhaps I just didn't understand your wording. I understand the purpose of review style landers (flogs, etc.) or comparisons/rankings style landers, but why mimic the offer LP?

 

[Dankface]

I'm targeting a pretty healthy-sized demographic on POF with a dating offer recommended to me by my AM. I know this offer has been hot for at least 3 weeks. When my AM first mentioned it to me he said it had a network wide conversion rate of 12%.

My 1st round results look very different.

Here are my numbers
56 ad copies (2 txts x 28 images)
CPM = .52-.56 (this blew me away, I thought I'd be able to make do by undercutting the recommended CPM, but I guess not. I got like 10 impressions bidding <.52)
Impressions = ~60k over 7 days (Just over 1K/ad)
CTR = .199 (2 of my ad copies are doing >.7% CTR)

so far so good...
Clicks= 119
Conversions = ...2

A far cry from the 12% i heard about.

My AM suggested some tips for my ad copy for the next round of optimization, but here are some questions-

1. Should I be optimizing around CTR or CVR? One of my 2 conversions came from my highest CTR ad, and 1 came from a low CTR one - nothing to conclude there. However, if I prune ads based on CTR, i have some clear winners. 8/1080 beats 2/1080 by a 97% confidence interval?
2. I'm on a modest testing budget, how do you recommend I proceed into my next stage of testing? Collect more data from my pruned ads until I can whittle it to about 10 and then test different ad copies?
3. Should I keep trying to run this offer? AM says they may have started scrubbing? Cut my losses & move on to different offer?

I'm probably going to get flamed here, but hopefully a few of you will recognize that I'm trying to actually hustle here and not just troll forums all day for advice that I'll never use.

Thanks in advance for your input.

Alejandro, Alejandro, Alejandro. Just because your AM claims that an offer is converting at xx% doesn't mean you should believe them. Would a used car salesman tell you that his car is going to break down in a few months? Hell no. So next time don't be surprised if you're let down by your network.

 

[BenPOF]

Perhaps I'm not understanding something here, but how does a "bridge" page that "mimics the offer LP" help increase conversions? Won't the visitor think "huh? Why did that button bring me to another page that looks pretty much the same?"

Sorry, perhaps I just didn't understand your wording. I understand the purpose of review style landers (flogs, etc.) or comparisons/rankings style landers, but why mimic the offer LP?

Correct me if I'm wrong but most direct links I've seen are sign up pages so the "mimics the offer LP", means use the same font/color/feel (for continuity reasons) but make it a sales page.

Think about it, if you're in a mall and some guy at a kiosk says HEY! BUY THIS PRODUCT. You're aren't going to buy unless he sells you on it first. Same thing with a click > sign up, you're missing the sales aspect.

 

[Garrett]

I know this offer has been hot for at least 3 weeks. When my AM first mentioned it to me he said it had a network wide conversion rate of 12%.

Things have to be advertised to the correct audience, #1. It's entirely likely that you need to just rethink your strategy though. For example, if you capture that traffic and coerce them down a funnel of 5 or 6+ offers, then you can bid wayyyy more than the next guy and kick his ass outta there hardcore. I've had a tough time straight out arbitraging traffic--&-->offers. Except when trends roll around at least :xmas-smiley-016:

My 1st round results look very different.

Here are my numbers
56 ad copies (2 txts x 28 images)
CPM = .52-.56 (this blew me away, I thought I'd be able to make do by undercutting the recommended CPM, but I guess not. I got like 10 impressions bidding <.52)
Impressions = ~60k over 7 days (Just over 1K/ad)
CTR = .199 (2 of my ad copies are doing >.7% CTR)


so far so good...
Clicks= 119
Conversions = ...2

A far cry from the 12% i heard about.

ok, so you spent $30 in traffic and got 2 conversions. How much are the conversions worth to you?

My AM suggested some tips for my ad copy for the next round of optimization, but here are some questions-

1. Should I be optimizing around CTR or CVR? One of my 2 conversions came from my highest CTR ad, and 1 came from a low CTR one - nothing to conclude there. However, if I prune ads based on CTR, i have some clear winners. 8/1080 beats 2/1080 by a 97% confidence interval?

You should be optimizing by earnings per ad impression. fuck CTR on your landing page and all that shit, it's easy to get tripped up on that shit in the beginning. track the funnel as a whole for starters. How much your ads make every impression. Sure... maybe an ad has a killer ctr on pof or whatever, but maybe another one makes you better money with a seemingly shitty ctr. DIvide your earnings -by- ad impressions to get the truer picture.

• I'm on a modest testing budget, how do you recommend I proceed into my next stage of testing? Collect more data from my pruned ads until I can whittle it to about 10 and then test different ad copies?
• Should I keep trying to run this offer? AM says they may have started scrubbing? Cut my losses & move on to different offer?

Look for the traffic first. Not the offer. BFF Forevers. New opportunities all the time.