Again... How many clicks do I need? (5)
Beat all to hell I know. . . How many times has this question been asked?
I know all about the tools (see here and here) I use them, but I do not get the results that I expect.
When using the ad calculator, should I use this per landing page/ offer combination ? Or do I just look at my ads based on all funnels?
What about time... I feel that I am cutting banners before they have a chance to really shine, as we know with adult there are up and down days. Banners do not perform consistently, so is there a better way to determine which banners to keep?
Should I test larger volumes? Such as 100 clicks per banner before I use the ad calc tools?
Looking for some insights to make better decisions with my campaigns, thanks!!!
Minimum 100 clicks (banners) or 10,000 impressions (pops) to make valuable conclusions
Is that 100 going to be per funnel (banner, lander, offer) ?
I personally prefer a more personal approach to banner/Lander/funnel testing and determining when to cut it. Whoever asks me about it, I always recommend to develop a style or lets call it "the feel" that tells you when it's been enough for a decision to be made. Its fine to follow the general "rules" at the beginning, but as you get to know your markets you should be able to tell its been enough without using any calculator or tool.
Im not a big fan of any set rules, like 100 clicks per banner, because often its not enough while in other cases its way more than I need.
Here are some reasons for cutting banners after a low amount of clicks or impressions :
- CTR : after running in a market for some time I know what kind of EPC I can realistically get, so if a banner has so low CTR that the clicks are way more expensive than the EPC I know I can achieve from past experience, I simply cut it. These banners are easy to spot when testing batches of ads, you will see some of them simply fail and there is no need to waste money on them.
- Too big negative ROI : lets say you launch a campaign and all banners show like -80% ROI, there is obviously something wrong here. There are several reasons for this, and one of them is that the banners are really bad. It can be other factors (placement, offer, LP...) so you need to analyze the situation in more detail, but if something like this happens to me, I simply stop the campaign and very often I restart with a brand new set of creatives.
There is way more to this, I will probably make my personal guide about this soon, but just to quickly give you a short glimpse on how I do it, lets take adult dating as an example :
I usually start with 2-3 LPs and up to 10 banners. First I run all the banners and watch the performance online, I always try to be online when the campaign starts, or some of my guys if Im not available. Usually there are about 2-3 banners that I will cut early based on low CTR, sometimes I only need like 2000 impressions, sometimes its more, this is where the personal knowledge of the market comes into play.
Then I let it run for some more, until the whole campaign spends about 3 times the payout, not every banner, but the whole campaign. If at this stage I don't get at least 1 lead, I pause the campaign and look for possible problems. If the lead is there, I continue to run until every banner has spent approximately 1 lead worth of traffic. Then I look again and go over the sources and look at the CTR, Im cutting sources that have way too low CTR which is making the traffic from those too expensive to have a chance to give me profit.
If at this stage Im doing worse than -50-60% ROI, I pause again and search for the reason. If its better, I continue to run and wait until I see some banners standing out and pausing the bad ones until Im left with max 5. I don't have exact $ spent set for cutting a banner at this stage, but I rarely keep going after 2.5 payout has been spent and the banner didn't make 2 leads. Once I have the best 5 banners and I've cut the worst sources, I let it run for as long as I feel comfortable in terms of loosing money 
At this stage I like to not be doing worse than -30%.
I also start to watch LPs now, I try to cut one of the 3 as soon as possible, the cases when I can do it quickly is when the leads per LP look like this : 7/5/1 ... if its not like that, I wait until one starts to loose clearly. Now we are coming close to the end of "day 1", by now I should have 5 banners/2 LPs and be close to breakeven. Next its a question of cutting bad placements and testing more offer versions. I always use proven offers, which is not a problem after you are running for a while as there is not that many of them in the end But pretty much every offer has more versions of their landing page and some will outperform the others. So its necessary to test them too.
At this stage I should be loosing just "mildly" and now its about optimizations. I keep just the best banners, the best LP of the 2 (or both if they are similar in performance), I cut the bad sources etc. Now since this example is adult dating, we all know some days work better than the others, so the next step depends on what day I started the campaign on. If Im running on Saturday/Sunday and the campaign is at heavy loss, I can almost safely assume it will suck even more during the week. If I started on Wednesday and the ROI is not there, I have a good chance it will work better during the weekend ... Sometimes you will only be able to profit on a handful of sources, sometimes it will be a total loss that you simply cannot turn green, sometimes the campaign will be a big hit on any day... but to be really sure, you need to test every day of the week. Even banners perform differently on different days. The key is to quickly find out which of your banners/LPs work the best and then test them over the week.
The post is getting too long so lets end it here. But let me repeat, you need to develop your own system/style, don't think that you have to use exact numbers/steps or you don't stand a chance to profit, its not true. Many times you simply have to give up early, when the numbers look ugly from the get go, simply cut it and don't expect a miracle, no matter how many $$$ you already spent on the traffic.. a 90% loss wont turn into profit no matter how much you pray for it I see people doing this all too often, its like ... the banner didn't make a single lead but I didn't spend 3 times the payout yet so I keep on running it even tho it has 0.00005% CTR, you need to be realistic, when its 50 cents a click its probably not gonna make money whatever you try
All the best!
Originally Posted by oneano
I know you've asked the same question multiple times already - and the fact that you did indicates that there's something you're not understanding, and it's probably this:
You can't really set rules of thumb like "run for xxx clicks / xxxx impressions before examining and deciding whether or not to cut", because it requires knowing the conversion rate (and perhaps other additional data), as well as the performance of other test candidates, in order to be able to cut anything fairly and with accuracy.
In other words, the "xxx click / xxxx impressions" will change with each situation. If statistics was as simple as you suggested, we wouldn't be needing to use these calculation tools in the first place. We're buds and have talked at length on skype so you should know I'm not trying to be condescending. Stats is a complicated topic and one that many people don't understand. So I'm glad to have yet another chance to make this clarification.
1)If your landers and offers are generic, e.g. NOT niche (so basically adult dating in general, not niche ones such as BBW-only, Chinese-only, Latinos/Latinas...), then you could probably safely evaluate combined stats, without evaluating each LP/offer combo separately.
2)If on the other hand, you're testing niche offers, it would make sense to test different landers for each offer, by selling each niche on corresponding landers. Of course in this case you WILL want to evaluate each LP/offer combo separately.
I'm going to make a few more points which I believe may help:
1)There are multiple factors that will affect ROI - mainly they will be your offer, banners, landers, and traffic quality. You've probably done enough testing to know which lander and placements convert well enough, so like I've suggested in your follow-along thread, it would make sense to test more offers and banners (i.e. angles).
It would make sense to use your best generic lander for mass-testing banners. Then once you find a good angle with your banner-testing, to then create custom landers for that angle.
2)Regarding your concerns about time (i.e. test duration) and fluctuations - this is exactly why it would be a better idea to test MORE BANNERS at one time, and only compare banners from the same batch of testing. By including more banners in a single round of testing, the data will be collected over a longer period of time (e.g. several days instead of several hours). By only comparing the performance of banners from the SAME round of testing, you don't worry about fluctuations because the same fluctuations will affect all the banners involved in a similar fashion (at least that's the approximation you can make). This is why I kept suggesting for you to come up with more angles, creating like 3 banners for each angle, and just mass-testing banners - instead of testing a few banners at a time like you've been doing.
3)Regarding a better way to test banners: If you don't currently have a profitable angle, it may be better to just lay off that calculator spreadsheet altogether until you do. If you create several banners for each new angle, you'll be able to see which angles have promise. Whether or not an angle has promise will be a subjective decision, and therefore this is not something any calculator can make. Keep testing new angles, and keep creating more banners for promising angles, until you find one or more angles that can generate profitable banners consistently. At that point and after, you can use the calculator spreadsheet to cut any banner that is unlikely to be profitable.
Hope I've not just reiterated what we've discussed previously - I don't remember what I have or haven't said before. I hope there's at least one new piece of info in this post that will help you...
Amy
Home > Paid Traffic Sources > Adult Traffic (NSFW)