MikeNoLastName - 10:42 pm on Nov 17, 2004 (gmt 0)

I'd like to mention one more thing which darkmage touched on in the opening post but which may not be intuitive to all, and may help explain the apparent increase in traffic without an increase in earnings. It concerns UNIQUE visitors. We've just started using GAdsense, but have been serving PPC ads in many varied formats for over 5 years. Consider a very simple hypothetical page with ONE ad unit which is doing really well. Assuming you've set it up on it's own channel, let's say it shows:

Impressions clicks CTR CPM earnings
1000 100 10.0% $50.00 $50.00.

So you decide "aw what the heck, if it gets that good of earnings with one unit maybe I'll add a second unit and those visitors who didn't find what they were looking for in the first ad unit might find it in the second". At least I'll make a little SOMETHING extra.

Well, since GAdsense (should that be pronounced: Gawd-sense? :) counts EACH UNIT as an impression, and assuming you make the common lazy mistake of adding the second ad unit using the same channel code as the first, adding that extra unit doubles the APPARENT impressions (but not, in our example, the unique users) but the rest of the stats APPARENTLY drop precipitously, to something like:

Impressions clicks CTR CPM earnings
2000 110 5.5% $26.00 $52.00.

In REALITY you've made a whopping $2.00 more by adding a second unit, although it appears you've doubled your traffic (which you haven't, it's still the same 1000 visitors). Do the math. We've credited a 10% increase in clickthrus to the second ad unit, which is reasonable based on experience. As previously mentioned the second unit is almost certainly lower bidding ads, so the extra 10% clicks (off of 100% more impressions) do NOT produce 10% more income even though you showed more clicks. In fact in some real-life cases we had the relative earnings go DOWN, presumably because a lot of the people who WOULD have clicked on the first ad unit, found a better fit on the second ad units and left via a 'cheaper route'. More about this in a moment. Finally you have to consider, in this scenario, people clicking, who don't find what they are looking for and come back using the back button. They ALSO apparently generate some additional impressions as far as GAd is concerned, thus further increasing impressions without increasing the rest of the stats. So as originally mentioned don't fall into the "self-inflicted more impressions trap", Ideally you should add the second ad unit using a spearate channel, at least initially, to make it easier to track what happens to each unit's productivity. You also need to make sure you're looking at UNIQUE users before assuming your earnings numbers are skewed down. The best way to determine uniques (until G decides to provide it on the reports) is of course via your own logs.

With one ad unit (asuming all 4-ad units here) you have to look at your avg exit fee to be the avg of the top four bids (say 1.50, 1.49, 1.36, 1.30 = 1.41, all things being equal), while the avg exit fee for two ad units (depending on the visibility of each) is the average of the top EIGHT bids (say 1.50, 1.49, 1.36, 1.30, 1.25, .90, .75, .10 = 1.08). So unless the 23% decrease in avg exit fee resulting from adding the second ad unit is made up by increasing the number of ad-clicking happy campers by 23% it may be more advantageous to simply use fewer ad units. Has anyone come up with what they feel is the ideal number of ads/units? Has anyone tracked the number of visitors who RETURN from clicking ads, perhaps to click another or visit other places on their site?

Finally, if you do add extra units be sure the first one (code wise) is more visible, so that visitors see it first as it generally has the highest paying ads.