On April 14, I broke the news that the second Google update targeting content farms, dubbed Panda 2 by the search engine community, had unexpectedly impacted eHow search rankings. Since then, the stock has dropped precipitously - but things have only gotten worse.
On June 9, Google's search engine spam chief Matt Cutts announced that Panda 2.2,was coming soon, and that it would target sites that copy content from other sites. That day appears to have come on June 17.
Here is a summary of how eHow ranked over roughly 5,000 keywords. For more details surrounding the keywords used, look here:
- Feb 19 - 1,680 top 10 hits at an average rank of 3.06 (Pre Panda 1)
- Mar 02 - 1,892 top 10 hits at an average rank of 3.05 (Post Panda 1)
- Apr 10 - 1,996 top 10 hits at an average rank of 2.95 (Pre Panda 2)
- Apr 16 - 1,341 top 10 hits at an average rank of 3.39 (Post Panda 2)
- Jun 17 - 1,216 top 10 hits at an average rank of 3.40 (Pre Panda 2.2)
- Jun 18 - 924 top 10 hits at an average rank of 3.56 (Post Panda 2.2 day 1)
- Jun 19 - 924 top 10 hits at an average rank of 3.55 (Post Panda 2.2 day 2)
- Jun 20 - 916 top 10 hits at an average rank of 3.56 (Post Panda 2.2 day 3)
You can draw your own conclusions, but in my opinion things aren't looking good for eHow. Our credit card site, NerdWallet, saw a 20% increase in search traffic in the exact same time frame as Panda 2.2, which increases my conviction that an update happened.
The Goldman Sachs Upgrade Makes No Sense.
Finally, I feel compelled to address the head-scratching upgrade by Goldman Sachs analyst Jordan Monahan, one week ago.
The timing is interesting - it comes ahead of the 180 day IPO lockup expiration in July, at which time about 72 million shares of stock should become available for sale by previously restricted shareholders. Interestingly, a few million of these shares are held by Goldman Sachs (see pages 47 and 183). However, since it is illegal for sell side analysts to upgrade stocks simply because their firm owns ~$100 million worth of said stock, this is clearly just a coincidence.
Point 1 - Panda 2 was just a one time setback
Management has identified and is in the process of fixing Panda-related issues, and “Panda 3” does not appear to threaten the company... The next round of Panda changes appears focused on sites that “scrape”, or plagiarize content, which should not affect DMD. - Jordan Monahan, Goldman Sachs
I think that a logical analyst should be modeling an earnings impact closer to 65%. Why the discrepancy?
Panda may impact 2011-2013E earnings by 3%-6%... Using traffic as a proxy for earnings. - Jordan Monahan, Goldman Sachs
Monahan gets to a 4% 2011 earnings impact by assuming a 35% revenue allocation for eHow, and multiplying this revenue % by a 12% decline in eHow traffic caused by Panda 2. This methodology is extremely flawed.
First, using revenue as a baseline to estimate earnings sensitivity implies that the domain registrar and 2 content businesses have the same margin structure, and that there is no operating leverage. I'd argue that using earnings allocations to estimate earnings sensitivity makes more sense.
Second, Goldman's assumptions imply that eHow has the same revenue per visitor profile as the rest of their Owned & Operated and Content & Media businesses. This assumption is off by at least a factor of 7x.
See table on Page 27 of their latest 10-Q. Owned & Operated pages earn $15.69 in revenue per 1,000 impressions, while network pages earn $2.16 in x-Tac revenue per 1,000 impressions. So how does "Using traffic as a proxy for earnings" make any sense at all - I wouldn't even use it as a proxy for revenue.
Furthermore, even within eHow's pages there are massive differences between search engine driven visitors and those generated by 3rd sites. Google traffic is free and massive in terms of volume. 3rd party traffic almost always requires competitive bidding against other content farms or other traffic arbitrageurs like AOL, Yahoo (YHOO), QNST, and MSN.
Demand Media's profit engine (and Achilles' heel) is the 100% incremental margin organic Google search traffic - not the low margin registrar business from eNom, and not low RPM paid traffic from media partners.
Simply removing pages about "how to deny a fart" won't fix the fundamental problem: 4 months ago the company could pay some guy $30 to write an article about a computer generated topic, and it would earn a huge IRR due to organic Google search. Now, that IRR is receiving a massive haircut, and may no longer even be profitable. And that same haircut is also impacting the 2 million pieces of content created by 130,000 freelancers last year.