The network of official Google blogs has to be commended for the transparency in recent posts. Their admission and apology for data collecting via StreetView and their candidness over the closure of the Nexus One webstore are truly great examples of business blogging. Today, the Inside Adwords blog finally discloses the how AdSense revenue share split is calculated.
Large publishers are already privy to this information in their negotiations with the advertising giant, but up until now, Google has never revealed this information to its 'self service' customers.
Here are the details you need to know:
AdSense for Content, the ads which are contextually targeted to the words on your web page is calculated at a 68% split to the publisher. Google takes 32% of the effective CPM (eCPM) they charge the advertiser.
AdSense for Search, the ads which are plugged into your own website's internal search engine and triggered and matched to keywords used in the search query is calculated at a 51% split to the publisher. Google takes 49% of the cost-per-click (CPC) they charge the advertiser.
What can you do with this information?
As Google do not disclose the keywords that trigger the ads on your page or search facility, there is no effective way to perform a reverse calculation to estimate how much the actual prices Google is charging for it's products.
However, a reverse calculation on the total yield of your website could be a worthwhile benchmark for any webmaster. For example:
Say your Adsense for Content earning for a particular month was $100 you would DIVIDE* this figure by 0.68 to discover the total advertising spend on your site. In this case, Google charged $147.06 for all AdSense for Content ads on your site, took $47.06 for themselves and obviously gave you $100.
For AdSense for Search, you would do exactly the same thing, but the DIVIDING NUMBER would be 0.51.
This is useful because it gives you an external sense of the market worth of your site. If you were to go direct to an advertiser, you would now want to make sure you were charging them over $147.06 dollars in order to generate a better return that your AdSense ads.
Google also advise you to focus on yield rather than revenue split.
"we encourage you to focus on the total revenue generated from your site, rather than just revenue share, which can be misleading. For example, you would receive $68 with AdSense for content for $100 worth of advertising that appeared on your site. If another ad network offers an 80% revenue share, but is only able to collect $50 from ads served on your site, you would earn $40. In this case, a higher revenue share wouldn't make up for the lower revenue yield of the other ad network. "
Join the dots to start making more money
Using a product like Google Ad Manager, which allows you to run your own exclusive ad slots and control the advertising on your page, you could now start monetizing your own inventory at a higher rate and use Google AdSense for Content as a backfill network for any unsold inventory.
This would mean you could offer premium or 'guaranteed' exposure on your own website for advertisers who were willing to pay the higher rate. It's good for you, it's good for Google and it's good for advertisers.
By revealing this information, Google has empowered webmasters to competitively price their premium ad spots, in line with advertisers expectations. Although an advertiser could target ad space on your own site, more cheaply, via AdWords, they have no real control of their exposure. By collaborating directly with advertisers, you can now offer them that control via Google Ad Manager, for a higher, but not wildly unrealistic rate.
It just shows that a little transparency can go a long way. So, those numbers again:
AdSense for Content (AFC) payout: 68%
Actual cost to advertiser: (AFC reported revenue) / 0.68
AdSense for Search (AFS) payout: 51%
Actual cost to advertiser: (AFS reported revenue) / 0.51
What does Google's revelation mean to you? Does it shed any light on your niche?
*UPDATE: Many thanks to commenter, Steve Hudgik, for correcting my calculations. (Further proof that transparency goes a long way!)