how long to test arbi niche
- Thread starter int_josh
- Start date
- Status
Tons of factors -
What is the CTR? Which ads are getting clicks?
What is the CPC? Can I drive more traffic, pay more but make more? or am I better off with less traffic? Does my CTR change based upon my position in search rankings?
What is the earning per click? Can I rewrite my page to target higher paying ads in the same niche?
What do you consider statistically certain - 80% 90% 99%?
We test everything with t-bars in very tightly controlled environments. For arbitrage, that is not as elegant a solution as an A/B or multivariate test.
Quick and Easy Answer is drive 1000 clicks. Did you make a profit? Answer questions above.
Were there any outliers that skewed your data (say 1 click for lots more than all other clicks, did you get 50 clicks that did not convert from one ip range(fraud?)....) Did any of the factors change in that time frame? CPC went up? down? what about earnings per click? did you get smart priced?
Testing must be done, but usually it is for comparing one scenario against another. Statistics are good at answering the question - Will I make more money with scenario A vs. scenario B. We use testing to refine successful campaigns.
Campaigns that come out of the box a small loser can usually be tweaked to be nice winners. Bad losers are usually not worth the effort, you can make more exploring a different segment and come back to this one later.
What is the CTR? Which ads are getting clicks?
What is the CPC? Can I drive more traffic, pay more but make more? or am I better off with less traffic? Does my CTR change based upon my position in search rankings?
What is the earning per click? Can I rewrite my page to target higher paying ads in the same niche?
What do you consider statistically certain - 80% 90% 99%?
We test everything with t-bars in very tightly controlled environments. For arbitrage, that is not as elegant a solution as an A/B or multivariate test.
Quick and Easy Answer is drive 1000 clicks. Did you make a profit? Answer questions above.
Were there any outliers that skewed your data (say 1 click for lots more than all other clicks, did you get 50 clicks that did not convert from one ip range(fraud?)....) Did any of the factors change in that time frame? CPC went up? down? what about earnings per click? did you get smart priced?
Testing must be done, but usually it is for comparing one scenario against another. Statistics are good at answering the question - Will I make more money with scenario A vs. scenario B. We use testing to refine successful campaigns.
Campaigns that come out of the box a small loser can usually be tweaked to be nice winners. Bad losers are usually not worth the effort, you can make more exploring a different segment and come back to this one later.
Well there are alot of factors besides "the niche" that decide if your gonna make $ with it. Basically if your making money keep it going, if your loosing money kill it, usually you can tell in a few days. if you want to know statistically if you will continue loosing money with it you could try splittester.com
I measure average Adsense payout per click versus the money I paid for the click. So if I start with the standard 50% CTR template and you pay $0.10 per click in 7search, then you need to make over $0.20 per click in Adsense to break even. This is my first factor.
If Adsense payout is really good, but for some reason CTR is low, I keep the niche, but my script changes the template. If CTR is still low, I'll dump the niche. This is second factor. Some niches are just that you can't get CTR up. Almost all computer related niches are like this, or at least for me. I just can't get good CTR from those.
And how long to test.. I usually test for 50 Adsense clicks. Then you'll see pretty much accurate average Adsense CPC.
Please remember that if you are going to do split testing, the keyword better pay REALLY well because other than that, it's just waste of time.
If Adsense payout is really good, but for some reason CTR is low, I keep the niche, but my script changes the template. If CTR is still low, I'll dump the niche. This is second factor. Some niches are just that you can't get CTR up. Almost all computer related niches are like this, or at least for me. I just can't get good CTR from those.
And how long to test.. I usually test for 50 Adsense clicks. Then you'll see pretty much accurate average Adsense CPC.
Please remember that if you are going to do split testing, the keyword better pay REALLY well because other than that, it's just waste of time.
Yeah, I said I have automated script that changes the template, but I still drop niches if they don't produce enough ROI.
For T-bars: Ack Really mathematical...Start here Wiki on T-Testing
Basically it is a graphical way to visualize error ranges on testing. Once a gap opens between the top of one test bar and the bottom of the other, you can usually stop testing the poorer performer. All of the functions are available in excel, so once you understand what it is and what the limitations are, you just need to give the spreadsheet data.
Basically it is a graphical way to visualize error ranges on testing. Once a gap opens between the top of one test bar and the bottom of the other, you can usually stop testing the poorer performer. All of the functions are available in excel, so once you understand what it is and what the limitations are, you just need to give the spreadsheet data.
Do you know some good books or online tutorials on using excel for statistical analysis? It's one thing I need to get better at.
I actually have a script that checks CTR per keyword, but I haven't found the time to write a script that would remove bad keywords from 7search. So at the moment I drop the whole niche. Maybe in the future I'll optimize the script finally
As I understand now, you submit a list of keywords to your traffic provider (7search.com i.e) and relate them to your ad. It's a one to many relation is it not? I mean, one ad can show up on numerous keywords from that list and therefore we get visitors to our page comming by different keywords. You can have different keywords feeding visitors to your page, but of course not all are good performers. Channel for specific page would measure CTR per page, it would not reveal to us which keyword has higher CTR and which doesn't.
So I was thinking in terms of coding up a script to track the performance of individual keywords and do some stats. From there I could increase the level
of automation and have my script automatically remove non-performing keywords from the keyword list.
If I have it all wrong, please correct me. This PPC game is new to me.
So I was thinking in terms of coding up a script to track the performance of individual keywords and do some stats. From there I could increase the level
of automation and have my script automatically remove non-performing keywords from the keyword list.
If I have it all wrong, please correct me. This PPC game is new to me.
You don't need to use any of the advanced statistical or estimation functions in Excel. Just set up a spreadsheet with formulas for calculating your minimum CTRs for each nice or keyword, your average PPCs and CPCs, divide by number of days where appropriate, etc.
I have set up mine so that I can copy/paste data straight from my 7search and Adsense reports into one and the same sheets. From there I can tell which niches suck and which ones I should keep.
But like ppl have pointed out; you should gather enough data (say, 50 adsense clicks or 7 days worth of stats depending on niche expense)
- Status