Welcome to WebmasterWorld Guest from 35.175.120.174

Forum Moderators: martinibuster

Message Too Old, No Replies

End of Month Balancing

Can someone explain this?

     
6:07 pm on Mar 28, 2016 (gmt 0)

Preferred Member

10+ Year Member

joined:Nov 15, 2007
posts:620
votes: 2


I know there has been talk over the years about ceilings in adsense earning, and conspiracy theory stuff, so I' m just going to share an observation.

One of two things happens to my earnings/RPM during the last 4-5 days of each month. Consistent.

If my earning for the current month are higher than my usual, even by let's say 50-100 dollars, then the last few days of the month earning drop to almost nothing.

If on the other hand, my earnings for the current month are below average, then the earning during the last few days spikes. So rather than an average RPM of let's say $3.00, it might go up to $16.00 just for those few days.

So, in the end, I earn just about the same each month, although progressively lower. This is on sites that have been around for years, once made 3k a month and now sit at lower to mid $000.

So what could explain this? What mechanism?

Practically the answer isn't actionable, but I am curious.
9:06 pm on Mar 28, 2016 (gmt 0)

Senior Member from GB 

WebmasterWorld Senior Member redbar is a WebmasterWorld Top Contributor of All Time 5+ Year Member Top Contributors Of The Month

joined:Oct 14, 2013
posts:3371
votes: 564


Is that why I'm not earning hardly anything right now?
9:22 pm on Mar 28, 2016 (gmt 0)

Full Member

Top Contributors Of The Month

joined:May 11, 2015
posts:219
votes: 70


Can be explained by seasonality, type of traffic, the way agencies buy ads (end of month, end of quarter). There's no conspiracy. There's no hidden secret.
You should try adding URL channels to your pages and try to pin point the few that cause the ups and downs.

Before thinking about conspiracies, it's important to understand your own account, how ads perform against each other, whether they are cannibalising each others clicks, whether it's your social vs. organic traffic (latter pays better), seasonality, the niche you're in, whether the industry growths or not, etc.

Also check whether it's a weekend or, for example, monday, that's right at the end of a month. In e-commerce, ad agencies spend their most budget on mondays. In this case, if a Monday is in the last few days of a month and the agencies must spend their budgets quick, your RPMs would skyrocket. That's just one example of what's possible.
10:35 pm on Mar 28, 2016 (gmt 0)

Preferred Member

10+ Year Member

joined:Nov 15, 2007
posts:620
votes: 2


LuckyD, thanks. I cannot see how seasonality applies, type of traffic or agency buying patterns, the latter because the "end of month correction" sometimes provides more money, and sometimes less, depending on how the month has gone. Given we're talking about adjacent months (Feb. was above average, than died, while March was below average, and now is earning lots in the final days), rules out buying patterns, which just aren't likely to change suddenly.

When I see a 900% jump in the last days of a month, OR a 500% drop at the end of the month, and one or the other happens for every month, I can't fit your possible explanations to the data. I DO think looking at a more granular level might help, but I'm not that curious, given I'm not sure knowing would allow me to do anything about it.

Oddly enough, the pattern applies on weekends, too, so a few months, I've had weekend numbers way higher than they should be given weekends are slow for our traffic.
7:28 am on Mar 29, 2016 (gmt 0)

Full Member

Top Contributors Of The Month

joined:May 11, 2015
posts:219
votes: 70


Another explanation might be the lack of statistically significant data. If you only get 10 clicks per day, and suddenly have 20 one day, it can be just plain coincidence. One lucky targeted ad, for example. I see people like RedBar (no offence), talking about large percentage drops, when it turns out that we are talking about single digit clicks per day. With only a few data points, there is no way to draw conclusions. If that's the case for you as well, please refer to the A/B test significance calculator.
10:29 am on Mar 29, 2016 (gmt 0)

Senior Member from GB 

WebmasterWorld Senior Member redbar is a WebmasterWorld Top Contributor of All Time 5+ Year Member Top Contributors Of The Month

joined:Oct 14, 2013
posts:3371
votes: 564


With only a few data points, there is no way to draw conclusions.


Just so that you are fully aware I used to be 1,000 clicks per day per several years and almost for 13 years. I have watched what Google has done to my sites by stealing images etc and the resulting drastic fall in traffic however because I have all this data I can still correlate what is happening now v last year v 5 years ago v 10 years ago.

Statistically significant data can still be drawn from lesser numbers, not that I like them to be though!
12:26 pm on Mar 29, 2016 (gmt 0)

Full Member

Top Contributors Of The Month

joined:May 11, 2015
posts:219
votes: 70


RedBar, I see your point. From what I see, your problems are mostly related to the amount and quality of traffic you get to your site(s), rather than AdSense itself. Please correct me if I'm wrong.
2:28 pm on Mar 29, 2016 (gmt 0)

Senior Member from US 

WebmasterWorld Senior Member tangor is a WebmasterWorld Top Contributor of All Time 10+ Year Member Top Contributors Of The Month

joined:Nov 29, 2005
posts:10599
votes: 1128


I have no proof, nothing but a gut feeling, that g's "accounting" is a strict 29 days, not calendar month ... means there's a secondary cycle over the 12 months which peaks and subsides at "beginning and end" of each month. This is fairly close to the lunar cycle, so this might be a lunatic observation.

(Part of above is tongue-in-cheek, but the underlying different reporting period has some value.)
11:57 pm on Mar 29, 2016 (gmt 0)

Preferred Member

10+ Year Member

joined:Nov 15, 2007
posts:620
votes: 2


Tangor, actually I agree about the 28 or 29 day cycle. It seems to match the data I see a bit better. Most of the possible interpretations of these patterns "could" be happening -- buying cycles, but none of them seems to be able to account for a balancing up OR down at the end of the month. That's why, as much as I hate the notion of going back to a "quota" explanation, I end up pushed that way.

If I had more time and energy, it would be interesting to run correlations day by day over a number of years and looking at months. For example, when total income is below average, do the totals from the last few days of the monthly cycle go up. And vice versa. You couldn't establish a causal relationship, but you could at least demonstrate it IS happening beyond the realm of chance.