Home > Paid Traffic Sources > Facebook & Instagram

---

oCPM campaign - split testing images - when to cut? (6)

---

Hi guys,

Running this for 48 hours now.

The problem here is that it's for a CPS offer which converts after 7 days. The payout is $100+

Example ad group: Targeting 1 interest (240'000 peeps), 10 ads with same texts, but 10 different images (newsfeed).

This is what we have so far:


I know it's too early to say cut this or that, because there's no conversion data yet, BUT $1.41 clicks are pretty hefty!

In general, how would you proceed with cutting targets? Or just wait it out?

Should I simply focus on CTR and cut the one that give me $1.4 clicks and keep the 0.51 ones? After how much reach should someone take action?

Any general rule?

Thanks for reading mates

---

Really hard to give you a good answer given the info you provided but...

If I am not familiar with click costs in a particular niche (as it seems like you probably aren't here) then I wouldn't really be super comfortable cutting ads until they'd spent at least a whole conversion if they have any reasonable costs. I would definitely cut earlier if I had a little better sample size and the costs were just absurd. For a $100 payout it seems like you could not reasonably make any assumptions given the data that you have. I'd personally probably easily expect to blow through $1,000 testing a $100 payout so I hope you've planned for that.

If I am more familiar with the niche, I will make cuts earlier on based on CPC's and lander performance. Expensive clicks can sometimes be very profitable though just as cheap clicks can give you nothing in sales. Sometimes an image is expensive but for some reason it just converts.

One final note...I hate that you are using OCPM here for two reasons:

1. You're in the early stages of testing. I find that OCPM sometimes produces some erratic short term results and you really have to let it run for a bit to kick in properly. This sucks when you are testing because some ads will out perform due to random weirdness with the algorithm. I like to split test CPC most of the time and then switch to oCPM sometimes down the road when I find some winners to see if I can get better results that way.

2. You said your target demo is 240,000. I think Facebook will tell you not to use oCPM on demos under 1,000,000 in size. I've had some good results on like 500,000 person demos personally, but I would probably avoid using it on an audience as small as this one

---

Yes, these are good pointers above.

1. For a $100 payout offer, ad testing is going to be expensive if based on cost vs revenue (in an ideal world...).

2. oCPM is intended for large audiences. 240,000 is OK, but you will get fluctuations in ad performance which can skew testing. I would test on a CPC basis first and you will likely pay lower per click if your ad CTR isn't great.

How are you funneling this campaign?

Ads > Lander > Offer?

I think the absolute best thing you could do for this is get some tracking metrics earlier in the funnel than the +7 day event.

Ask your AM and/or talk to the advertiser - can they place a pixel for you for signups? Another at x Day mark?

If you have lander CTR, sign ups, and some earlier retention data, you can far more confidently spend and optimise your ads. At the moment, a +7 day $100 conversion is as blind as you can get!

When it comes to optimising ads, I would then do so based on cost per signup or failing that, cost per lander clickthrough. Ignore ad CTR and CPC, look at cost-per-metric-that-matters.

---

Thanks bogeyguy and thanks Zeno. That's some very useful and actionable input!

I've actually chosen oCPM because it's a newsfeed ad. Wouldn't newsfeed ads based on CPC be too expensive due to the high CTR?

I had to choose newsfeed, because the device type I'm targeting (iPads) doesn't let me choose RHS ads, which is weird because iPads do in fact show them.

Yes, and split-testing the lander.

I'm currently waiting for the advertiser to add my FB conversion pixel, which will fire upon trial sign up. It it def. possible :-)

Thanks again for sharing your wisdom!

---

No. You should test each bidding mode for separate campaigns. If you are bidding in CPM mode, a low CTR punishes you and a high CTR is rewarding. If you are bidding CPC, there will be a bid floor and your CTR will influence your CPC, but you will have to test to find the actual CPC prices you pay.

The two modes are entirely different and you can't guess that one will be cheaper than the other, since you don't know what your ad metrics are or the demographic CPMs are until you actually run the ads. Don't make assumptions, test.

When starting out, CPC has the advantage that you (presuming a low % social actions) pay for actionable data. Whereas with a CPM mode, you pay but may not get any actionable insight other than 'my ad sucks'.

This is not precisely true. On iPads, if you are using the Facebook app, there are only news feed ads - this is consistent across all mobile devices through the app. However, since the iPad has generous screen proportions, users might browse the Facebook website instead, and will see RHS ads in this case.

Facebook probably offers mobile targeting only through the app because they can be sure of the device type.

---

Cheaper clicks don't always mean more money or better CR.
I am not really familiar with oCPM but with CPC it takes few days to get cheaper clicks (you should just let it run 24/7).

---

Home > Paid Traffic Sources > Facebook & Instagram