Target/keyword cutting (zeropark) (5)
Hey,
I've been running dating on ZP for a few days, now I'm looking through the stats:
So, how should I approach optimizing this? Obviously 2x payout rule is not so convenient to be used here, cause we have several hundreds of target domains instead of 10-20 banners. Can we incorporate landing CTR here? Like .5-1x payout and low CTR?
You could put in a lander CTR cut-off but I don't think it would be a great idea here. Why? Because looking at your conversion rate you can see that you have 25%, 50% and 100%. If those highlighted landers keep their current CTR the first only needs to convert at about 1 in 6 to be profitable and the second 1 in 2, which it still may do.
-45% ROI off the bat without optimisation is pretty nice to be honest!
Running on ZeroPark with a comparatively low-payout offer I'd almost always ignore CTR altogether unless it's truly, truly terrible, and go straight for cutting on ROI.
(This assumes you're running a regular A/B test campaign and not doing something funky with multivariate testing, at which point the rules change again).
Remember, whilst you have several hundred domains, 90% of the traffic will come from 10% of those - which brings you straight back to having, effectively, 10-20 targets at any one time.
Once you have a solid base of profitable or promising targets (taking into account LP optimisation too), I'd recommend starting a new campaign with just those targets. That stops you bleeding money on small targets that never quite get to statistical significance.
Zeno, thanks for answer.
But I'm not sure if I understood you completely.
If you're saying we should not put in LP CTR, then you can't operate with given CRs (25%, 50%, 100%) cause they are calculated based on landing page -> offer clicks.
Instead we must use "total CR". With my current setup minimal total CR = CPC/AvgPayout = $0.02/$3 = 0.0067
So we use 95% binomial confidence intervals.
If the target placement has 0 conversions, we need ~550 "visits" (3.7x payout) to be sure it's CR < minimal total CR. 1 conversion - ~850 visits (5.7x payout), 2 conversions - ~1100 "visits" (7.3x payout) etc.
Still looks like too much money needs to be wasted for one domain. And as I have said before, I have ~50 major target domains and several hundreds total.
Also, here's my LP stats, if that's gonna help:
EDIT:
(haha I'm so slow at writing)
To caurmen:
What CTR could be considered terrible? I've read your follow along where you mentioned CTRs over 20%, but that was mobile. Are my results acceptable if I'm running only web?
A "terrible" CTR is something that will be easier to figure out once you know what the average CVR of a successful placement is. At the moment, eyeballing your stats, I'd tend to drop anything with more than 100 clicks and less than 3% CTR.
It looks like you have enough data to start cutting landers, unless you're intending to cut them on a placement-by-placement basis. Check the top four with the split-test calculator at PeakConversion: http://www.peakconversion.com/2012/0...al-calculator/ (on conversions not clickthroughs)
Home > Questions and Answers > Traffic Source Questions