As someone who tries to present marketing data on a regular basis that is accurate, meaningful, in context, and hopefully actionable, it really bugs me to read reports on our industry that appear to be divorced from reality and/or are internally inconsistent. My frustration is only heightened when these reports come from big players that receive big press and the message gets disseminated widely, often without context or added analysis, and alarms get raised unnecessarily.
This week, a report from Kantar Media, who the New York Times described as one of two large firms -- along with Nielsen -- that tracks ad spending, suggested that paid search spend by American advertisers fell 14.4% in the third quarter of this year. This flies in the face of pretty much all of the information I've seen, not the least of which are the Q3 reports straight from the search engines themselves, and frankly sounds impossible to believe.
To their credit, a couple of the media outlets that carried this story, including Clickz and AllThingsD, did express varying degrees of skepticism around the paid search number, but anyone who just read the headlines in their news reader or Twitter feed would have missed that. Clickz went so far as to solicit a response from Kantar, whose reply was essentially to point out that Google, whose revenue grew 33% in Q3, has multiple revenue streams, with paid search being just one component.
Does that excuse hold water? Could Google, who reaps the lion's share of paid search spend from American advertisers, see its revenue grow 33% year over year while overall paid search spending declined by over 14%? Since neither Google nor Microsoft provide specific breakouts for paid search, let's do a few back-of-the-envelope calculations to see if that's even a remote possibility.
So, what do we know? Digging through the Q3 earnings of Google, Yahoo and Microsoft gives us several key pieces of information. That said, there is some quirkiness in Yahoo and Microsoft's YoY numbers due to accounting changes surrounding their search partnership, so we'll select our numbers in a way that is heavily favorable to the Kantar report. For example to start, Google's revenues in the United States were up only 26%, so we'll use that instead of their 33% global lift.
Yahoo, who conventional wisdom suggests is trending significantly worse in paid search than Google, does provide a breakout of paid search revenues and they reported a YoY decline of just 13% after traffic acquisition costs. Microsoft reported a 19% increase in revenue from its Online Services division and claimed that search was the primary driver. Yahoo's overall losses in the Americas for the quarter were a little over double their worldwide losses, so we'll say that paid search was down 30% for them in Q3. We'll also assume that Microsoft's numbers are just accounting magic and give them the same figure. Note that the larger the paid search declines we assign to Bing and Yahoo, the smaller the divergence in YoY growth for paid search and overall revenues Google would have needed to see to make the Kantar paid search number work.
Because Microsoft doesn't provide many specifics for its Online Services we also need to estimate the total share of search spend going to Bing and Yahoo. Most marketers would probably peg this as around 20% or so now, but we'll double that for our 2010 estimate, which also compensates for the other niche players out there. Again, the larger the share of the supposed paid search decline we can put on Bing and Yahoo, the more the necessary Google numbers start to look possible.
Finally, we need to estimate what percentage of the search engines' revenues are generated by paid search and what percentage comes from other streams. Yahoo's Q3 2010 report put paid search revenue at 38% of its total. This is likely the lowest of the three major engines in reality, but we'll use that figure for Google and Bing as well.
Taking everything together and putting total revenues between the big three engines at a rough $5.9B in Q3 2010, we get a table that looks like this. The key estimates I used are highlighted and the other cells follow from there.
So, under this intentionally skewed hypothetical scenario, in order for the 14.4% decline to work, Google would have needed to see non-paid search revenue increase over 44% to compensate for a paid search decline of 4%. This would result in paid search making up just 29% of Google's revenues in the US for Q3 2011. This sounds completely unreasonable. We could start with a higher initial share of revenue from search for Google, but this would only drive up the growth they would need to see from their other revenue streams to ridiculous levels, as we'll see next.
Plugging those more reasonable numbers into our calculations, Google would have needed to see an incredible divergence in its paid search and other revenues in order for Kantar's 14.4% paid search decline to be anywhere close to accurate.
Here Google would have seen their other revenue more than double, while paid search revenue declined 14%. Since display would be making up the bulk of the other segment, this does not fit with Kantar's estimate of a mere 16% increase in display spend.
If you're convinced, as I am, that there's simply no way that paid search spending declined 14.4% in Q3, it calls into question the credibility of the entire Kantar report and the methodology behind it. There isn't a perfect way to estimate these things, but it just seems hard to believe one would come up with a number that's so wrong two months after the engines' earnings data was available. One final nail: Kantar has paid search spend falling just 2.1% for Q1-Q3, suggesting paid search spend growth fell off rapidly, even as Google's revenue growth accelerated and Yahoo stemmed its losses.
Now that I've written this, chances are I've made a critical mistake myself somewhere. Please let me know if you see one or if you've seen any other data that supports the Kantar paid search number.