Forum Moderators: martinibuster
One of the things I've been working on is to try to get the best epc considering my topic and ads available. My problem is that I have no way of telling if epc has improved because I've dropped some low paying pages, positioned my ad units better or because the advertisers are just bidding higher during a given period. For example I usually have a poor December. My topic appeals to busy wives and moms and the holiday season means a very low visitor rate. But I did well this year because for some reason the clicks were paying really well. The epc dropped in Jan but then I did well because of a greater visitor rate. April did poorly because of holidays (less visitors) but also poorer epc.
I seem to have big shifts in ctr as well. Summer is slower in terms of visitors (kids on vacation then) but my ctr was great. I hadn't changed my ad units at all so was it more interesting ads? Or did those who did visit click more ads out of boredom. (maybe the ones with no kids tearing around the house). Or maybe there are surfing from work and business was slower. You can go nuts trying to figure this out.
I have had a lot more visitors this year than the same months last year. Is the increase in unique visitors the content I've added, are my pages doing a little better in the search engines or is it there are just more & more people are surfing the net?
Is it really possible to improve earnings using stats or not?
Although it's interesting to look at past data, data that is two years old may not be totally relevant in working out the effect of changes. In the two and a bit years since I've been in adsense, Google have changed the algorithms many times, introduced many new features such as optimised ad units, smart pricing, cpc and so on that comparing two years ago is not a true comparison in some ways.
I guess you can compare the traffic, and how much that traffic earned you both then and now, but I tend to use shorter time periods to work out if my changes have helped or hindered.
Have you given new content a seperate channel, so that you can see if it under/over/match performs your norms?
I'd use a year's worth of data (in Excel Charts) to make sure I understood seasonal ups and downs. Again, with enough channel data, you might see some topics do well when others are off.
This is just a tiny start. There's alot you can do to grow your business by studying the performance data.
I think the preview tool should have CPC & CPM data.
I don't know why we are forced to work in the dark.
I think the preview tool should have CPC & CPM data.
Would that be based on averages across all sites for a given ad? Or do you want Adsense to make a prediction for a specific site?
That's asking AdSense to do a lot for a tool that would primarily be used by high-paying-keyword hunters.
No thank you. I don't want that tool.
Example: If you have a site about Ford automobiles, and you find that pages about Mustangs top all other pages, with Taurus or Focus being way down there, then I would focus on getting content about Mustangs up on your site first. In other words, if you have three stories about Mustangs and three about Taurus, I would get the ones done about Mustangs first.
It's basically, IMO, a matter of catering to the majority of visitors.
If you have some really well-performing pages, you might even think about adding a third-party add - ala Amazon or BN.com or whatever (i.e. on a page about Ford Mustangs, in addition to AS, place an add for a subscription to a Ford Mustange magazine or Ford Mustang book).
Not that AS stats are not handy - if you see some weird stats, i.e. your big-money AS pages are not the same as your most popular site pages, then you might want to tweak the ad layout (that's probably a rare occurence).
Also, as others have mentioned, check for seasonal variations - if your site performed really well leading up to some kind of holiday or season or general time frame, then you want to maximize that by providing more content oriented towards that timeframe.
So what is really driving me nuts is trying to use stats to help me decide on changes I should make in my ad placements, which units to use, which pages may do so poorly that it would be better to leave the ads off and which ads should be blocked. I feel like I've tried so many things and I can't tell if they have done any good or not.
Perhaps I should just stick to content and forget tweaking adsense.
I feel like I've tried so many things and I can't tell if they have done any good or not.
This should not be that hard if you have sufficient data (traffic). First, measure the amount of normal variation in the desired quantity (e.g., EPC, revenue, whatever). Second, make a single change, site-wide (e.g. change to different ad placement). Test for "long enough" (if you've got enough traffic, that can be hours or even just minutes, if not, you'll need longer).
There's three possibilities: the desired quantity got significantly better than normal variation, significantly worse, or was not significantly different than the range of normal variation. If things got significantly better, switch to the new scheme and get ready to test the next change.
What exactly are you getting stuck on here? Why can't you tell whether your changes have done any good or not?
For now I think my ads are positioned pretty well but I gave up my 468 x 60 banners at the end of my articles to see if that would help with EPC. If removing the bottom banners did not help my EPC I'd like to put them back as they were the one thing people who read the whole article might notice when they finish reading. The CTR was always low but the earnings added up.
I think my biggest frustration is that EPC seems to be varying so much even when I don't make changes. My overall EPC has varied by 25% in the last week.
How would I calculate normal variation with this little traffic?
(BTW I always seem to have a good stream of ads so targeted ads is not a problem)
- Start with one week baseline to determine the typical quantity and quality of daily traffic and which days tend to be similar in amount/quality of traffic and which pages appear comparable. Use the later info to determine if there is any correlation between what you are doing on those pages and their doing well/badly. Never use data older than a week on a quantitative scale, only in a relative manner. (i.e. 'changing the color last week boosted me to $100 a day on this page' is wrong, while 'changing the color DOUBLED my prior income in this position' is more useful to know)
- Tantamount is keeping absolutely detailed logs of ANY changes you make down to the exact date/time and URL on which you make them. Without this you'll wind up guessing what you did (often poorly) or duplicating a lot of effort. When something dosen't work, reducing CTR/EPC, change it back ASAP and make note of it.
- Unless every page on your site is identical work on a page by page basis and optimize starting with your most trafficed ones first. Make lots of separate channels (at least initially) and track each type of ad on each type of page. Eventually you'll realize 1 ad tends to make 90% of the CTR on a page and that by eliminating the other(s) you can double(+) your CTR for that page. If you decide to keep the other ads, always be sure to place the first ad (code-wise) on the page in that top earning position to optimize overall CPM on the page as supposedly the first ads delivered are the higher paying. Then continue on to optimize shape and colors, if appropriate, to increase CTR.
- Use a spreadsheet to calculate Avg earnings per click, since G doesn't provide this on any of their reports, find your average (sitewide and pagewide) and concentrate on the lowest ones in order to raise your avg. I initially spent over 3 months doing this every day and tripled our EPC. However nowadays, with their new algorithms working more efficiently, unless you are concerned about losing visitors leaving your site and not coming back, the page eCPM is actually the more important of the values since you may have an ad which doesn't pay as much as the others, but which may get clicked twice as often. If you ARE concerned about losing visitors too cheaply and would prefer less, higher-paying clicks, try TEMPORARILY filtering the ads apparently getting high CTR with low EPC and see if new ones get displayed with lower CTR/higher EPC. If so, you should be able to remove the filter after a couple days and GAd will continue to display the new better ones.
- Use Adsense regions on pages producing silly ads and low CTR/EPC, to tell G what the most important parts of your page are (and those which you feel will produce the highest paying ads). However, don't expect high smartpricing if this differs significantly from the overall theme of your site or from keywords you are found for on the search engines.
- Finally realize, no matter what you do, there are going to be daily flucuations and there are probably some outside factors over which you will never have control, such as advertiser buying trends (end of month budget, high-rollers dropping out suddenly), algo changes, and other external smart pricing factors added by G. Among these COULD even be your frequency of changing content or subtle keyword changes. Follow overall trends more so than absolute daily quantities and you will begin to get a feel for what exactly is going on.
Compared to where they were 18 month ago, I think GAd has made some major strides in tracking their own results and optimizing CTR/EPC to display the ideal ads and maximize overall payout per page. For instance it is no longer as critically necessary to filter out obviously low-paying spam ads with the competitive filter like it was over a year ago. This takes a lot of the work off us publishers. But the things listed above will definitely help to further accentuate these efforts.
I haven't kept the best of records either. I have spent a lot of time calculating epc. What a pain. I need to focus on it in a more organized fashion. EPC data would be nice to have in our stats.
always be sure to place the first ad (code-wise) on the page in that top earning position to optimize overall CPM on the page as supposedly the first ads delivered are the higher paying.
Do you think adlink units are included in the calculation? I earn quite a bit from them but on some pages they are ahead of the regular ad unit. The regular unit always gets a lot more clicks.
