Forum Moderators: martinibuster
One has to be realistic. That drop in statistics might just be signalling the end of life as we know it - aarrgghhh! However, before you jump off the Golden Gate, or pour a bottle of small round pills down your throat, it is worth asking whether it might be another demon that is tormenting you: "Natural Statistical Variation". I hope this post might just save some from their suicidal thoughts, for NSV's bark is much worse than his bite.
Natural Statistical Variation is a statistical principle that, in essence, says, all other things being equal, your Adsense figures will vary according to the following pattern:
So, unusual variations are to be expected, and need not cause us to despair.
But, I hear you all cry in unison (or not, as you switch over to another, far more interesting, thread), how can I tell whether an earnings figure is normal; what is my normal range? That's easy, do not panic, for the knight in shining armour to defeat Demon NSV is here, in the form of "Standard Deviation".
Standard Deviation simply tells you what is your normal range. And the law of Natural Statistical Variation tells you to expect 1 in 3 days that are outside that range.
To calculate your normal range, do the following:
Your normal range is then from 'average - stdev' to 'average + stdev'. For example, if your average is 100, and standard deviation is 20, then your normal range is between $80 and $120 per day. Natural Statistical Variation predicts you would have figures in these ranges roughly 20 days a month.
To work out what is "just outside your normal range", simply add another standard deviation. In the example above, that would be $60 to $80 and $120 to $140 dollars. Natural Statistical Variation predicts you would have figures in these ranges 8 days a month.
Figures below $60, or above $140, are unusually high or low. However, Natural Statistical Variation predicts this will happen once or twice a month. Natural Statistical Variation also predicts that each month will be different, so in some months you might have a lot of unusual earnings, and in others it all seems super smooth. That is, it is normal to be abnormal. On top of this, market rates vary, the number of advertisers vary, and there are lots of other things that vary.
If you get dropped from the Google SERPS, there is only one direction that your earnings will go in. But, hopefully, the above will be reassuring to those who have felt depressed every time the tide goes out, and celebrated every time the tide comes in.
For a quick rule of thumb I use 3:1 best:worst day, if it gets any higher I start to look around to see what's happening.
Ok I dont get it! My GF is a maths teacher so I will ask her!
I think adsense go look at your sites and choose a figure. They program this figure in and plus or minus thats what you get! (Ok not really but...)
Because there has been times that hits and clicks have doubled on some sites due to a big site linking for a week or two. Income went up on *that site* and down on the rest. Net result same income again!
Nitrous wrote:
>So how come my income over 2 years is almost always (90%
>plus figures) within 10 dollars of the all time average of $90
>per day?
These two responses show why you should learn to love Standard Deviation, as these figures represent a couple of extremes.
Optirex's figures show a massive daily variation - 3:1 from best to worst - whilst Nitrous' are very consistent. Optirex would likely have a very large standard deviation, showing that his stats vary a lot, whilst Nitrous would have a very small deviation showing that his stats vary a little. But in both cases the standard deviation would help to provide some clarity on what the "normal range" is.
Nitrous, have you done the STDEV calculation? If any other members calculate their standard deviation, it would be interesting to see how it varies across different sites. Ours for weekdays is $58 (ie the normal range is $116 wide) on an average currently of about $300. Weekends are different (less) though.
The stat I really like to see is the one on the check at the end of the month. So far, on average, that has just kept getting bigger.
I use 3:1 best:worst day
This are extrem variations
In September 2005, best to worst day was 1:1.7
Optirex's figures show a massive daily variation - 3:1 from best to worst
Heck, I've never considered them as a massive variation since they've always done this and I look at it as "normal" for me and I view my earnings from them like this:
Monday to Thursday 100% best days
Friday 85-90% of Mon-Fri
Saturday and Sunday 50-60% of Mon to Fri
This has been consistent for the past 2 years however it has been improving this last month, therefore using $100 as the overall average and the best day of $120 and the worst at $40, we have 3:1.
Considering I have "only" 50-60% of the traffic on a weekend maybe I'll only have 3:1 happen once in a month, the more regular average is 2.5:1 however I've never seen anything like jettroheller's 1.7:1.
Your patterns are very similar to mine, in fact. Are your sites business? Business sites are more likely to have big variations between weekdays and weekends.
When analysing our site stats, I separate out weekdays from weekends. Weekdays only has a standard deviation of $58, but if you include weekends it shoots up to $86.
Your patterns are very similar to mine, in fact. Are your sites business? Business sites are more likely to have big variations between weekdays and weekends.
Business as in the #1 trade niche business directory portal plus #1 trade niche business sites.
During the week I "guess" the majority of visitors daytime are business, evenings public, weekends mostly public or trade people sitting at home on a lazy Sunday afternoon learning the Net and trying to find alternative supplies or what they actually want!.
This does not "fit-in" with the conventional US Internet accepted business model of people surfing from work for personal info and purchases, and I only base these conclusions on our Adsense earnings and "read-about" Net phenomena since they are very consistent with these reported findings.
Does that make sense?
If it's something else then I've got everything wrong...!...am I going to change anything...hmmm!
We have one for our site, initially used to try and understand which part of the world our markets were in.
Each site will be different, of course, and short term stats are subject to even greater variation, particularly with click dumps and stats getting stuck. However, the following table may be of use as a starting point for anyone wanting to develop their own. It is an average of how income progresses through the day for our (UK site). The time is California time (add 8 hours for UK time) and the percentage is the total percent of daily income achieved, on average, by that time.
1:30am - 5%
2:30am - 11%
3:30am - 18%
4:30am - 25%
5:30am - 30%
6:30am - 35%
7:30am - 44%
8:30am - 54%
9:30am - 62%
10:30am - 68%
11:30am - 74%
12:30am - 79%
13:30am - 82%
14:30am - 84%
15:30am - 86%
16:30am - 88%
Our daily totals vary roughly in line with Optirex's percentages - Friday down a bit and Sat/Sun down a lot more.
