Pune based online advertising optimization startup PubMatic has announced the availability of the PubMatic AdPrice Index – essentially a sort of an Sensex (or S&P 500 Index) for the online advertising world.
The AdPrice Index essentially reports on the average online advertising revenues earned by 3000 web publishers worldwide who are PubMatic’s customers. And it only counts the income earned through use of ad networks like Google’s Adsense, or Yahoo! Publisher Network. Specifically, it does not account for revenues earned through ad spots sold directly, or through other channels like TextLinkAds or Amazon’s Affiliate marketing program.
The index does give some breakup of the information – small niche websites earn more ($1.18 eCPM, i.e. earnings per 1000 pageviews) compared to large websites ($0.38 eCPM). However, variation across websites (based on the area, geography and other factors) is so large that I doubt that the actual numbers are directly useful to anybody. However, the main value I see is in the variation over time. PubMatic will release these numbers monthly, thus giving publishers a feel for the general state of the market – just like a stock market index for stocks.
I am sure PubMatic is also in a position to answer the question on most bloggers mind – “Is there an ad network for me that performs better than AdSense?” However, PubMatic is not releasing a breakup of earnings by ad network. Of course, they want you to not care about this question. Because, if PubMatic works as advertised, you wont need to answer that question – PubMatic will automatically only show the most lucrative ads and in fact, their claim is that by allowing PubMatic to broker the ads between different ad networks on a per impression basis, you will end up earning more than you could have earned through any single ad network.
Of course, this is not really guaranteed to work. The main problem is that information about earnings from individual ads is not directly available. Hence PubMatic has to guess. Here, it ends up using lots of complex mathematics (data-mining, machine learning, and I assume bayesian reasoning) to make those guesses. If all this guessing works in the case of your website, you will be richer (some websites report upto 90% increase in revenue), but it it does not, your revenues might actually drop a little bit. This can be especially true in the case of smaller websites in the early days of installing PubMatic – because its algorithms work better with more data. Smaller website + fewer days of operation = more chances of guessing wrong. The longer you keep PubMatic running, the better it should get. In theory.