Who got the £354M UK search revenue that did not go to Google?

Here is my problem:

• The IAB says that Google has 78% search revenue market share.
• Hitwise says that Google has ~80% search query market share.
• Google monetises each search much better than the market average.

How can Google's revenue market share be less than (or even the same) as its query market share?

Either:

• The IAB is wrong. Which is very unlikely, because the IAB/PWC study is pretty rigorous, it's been around for ages and is used by lots of people who would have picked up on any discrepancy.
• Hitwise is wrong. Which is unlikely given that none of the major search engines have really complained about this Hitwise data. If Hitwise was wrong, I would expect to adjust search query market share to around 60%, and I just don't believe a third of searches take place somewhere that isn't Google.
• I'm being an idiot.

So what am I missing?

According to the IAB, UK search revenues in 2007 were £1,619M.

According to Google, its UK search revenues in 2007 were $2,531M. Which using an average exchange rate for 2007, means £1,265M. Which is 78% of the IAB's UK search revenues.

Although 78% market share is a big number, it seems low given that Google had around 80% search query share (and also here) in 2007 (78% in May but growing to 88% a year later, so let's call 80% a good average for 2007). But Google monetises search queries much better than anyone else so it's surprising to see its share of revenues be the same as its share of search queries. Either the revenue market share is too low, or the search query market share is too high. Does it matter how accurate these billion+ numbers are? Not so much if you want to model Google's revenues - an extra £50M or £100M is not going to make a fundamental difference. But it does matter if you want to work out how much the other search engines make.

So who got the remaining £354M in 2007? [Note: I couldn't find out from the IAB if they count as Google revenue the money that goes to Google via its major partners such as AOL and Ask, but I will assume not].

Method A: Using overall market display/search ratios as a proxy for Yahoo/MSN/Ask

• Yahoo does not report its search revenues. I'm going to assume that Yahoo is typical of the online advertising market sans Google. So I need an approximation for a typical display:search revenue split (in the UK). The IAB says that UK display was £592M, and search-minus-Google was £354M. So without Google, a typical display:search breakdown could be in the ratio of 592:354 (63%:37%). OK, back to Yahoo. According to its annual report, no one country accounted for more than 10% of Yahoo's total $6,969M revenues. If that also applies to its $6,088 advertising revenues, then no one country accounted for more than £304M. Let's assume the UK is 8%, so £244M. How much of that is search?  Using the 592:354 ration for £244M means Yahoo UK search revenue was £91M.
  
• MSN's online advertising revenues were $1,940M in 2007. Assuming the UK is also 8% of total revenues, and applying the 592:354 ratio makes MSN's UK search revenues £29M.
  
• Ask: In the US, IAC's Media & Advertising unit (which is where Ask sits) made $759M or £379M. Let's guess that Ask accounted for £300M. IAC seems to suggest in its annual report that the UK is a significant share of its revenues, so let's call it 15%. Applying the 592:354 ratio makes Ask's UK search revenues £17M.

This leaves £217M of search revenue unaccounted for - and it cannot all be long tail because what kind of a long tail would add up to more than Yahoo+MSN+Ask?

Method B: Splitting the revenues by their May 2007 search query market share:

Yahoo, MSN and Ask have search query market shares of 8.6%, 5.5% and 5.0%. Splitting the revenues proportionately means:

• Yahoo: £140M (vs £91M above)
• MSN: £89M (vs £29M above)
• Ask: £81M (vs £17M above)
• Others: £44M

However, it's not right to segment the market revenues as if everyone had the same revenue per search query. If the IAB/Google revenue correlation is right (it must be - I imagine the first thing the IAB did was check Google's revenues), and if the other search engines have an RPS that is about 40% lower,  then we must assign to Yahoo, MSN and ASK (with search query market shares of 8.6%, 5.5% and 5.0%), revenue shares of 40% less - so 5.2%, 3.3% and 3.0%:

• Yahoo: £84M
• MSN: £53M
• Ask: £49M

This leaves £168M of search revenue unaccounted for - and again it cannot all be long tail because what kind of a long tail would add up to almost the same as Yahoo+MSN+Ask?

PS: two other data points

• According to American Technology Research, Yahoo global search revenue was $1,700M in 2007. Using the above 8% and 592:354 ratio, that makes Yahoo UK search £68M.
• Another way of cutting the American Technology Research data is to say that Yahoo's search revenue was 10% of Google's search revenue. In the UK, Google is stronger than in the US so if Yahoo UK search is, say, 7% of Google UK search then Yahoo UK search would be £89M.

So in conclusion:

• It doesn't feel right that Google has 80% query market share and 78% revenue market share.
• I need to account for the remaining £168M/£217M or thereabouts of revenue. The easiest way is to assume that everyone earns revenue according to their query market share, and ignore the fact that Google monetises better, but that's just wrong. So how do I account for it?

Is the IAB search market size too high? Is there really so much search revenue beyond the four search engines above? Or do Yahoo, MSN and ASK monetise much better than the above suggests?

Comments

[Tom Loosemore]

Hitwise has Google getting nearer 90% - not 80% - of UK search query market (87% in May 2008, same as for Jan 2008 iirc).

I'd argue that there's little incentive for search engines to complain about Hitwise's data. Yahoo and Ask won't want to shout about having 3% each, and Google won't want to shout loudly about having on for 90% market share.

Of course, this makes the apparent discrepancy worse...

My money would be on the IAB search revenue stats being overstated. Just because it's been around for a while does not necessarily imply that IAB data is highly accurate.

Over the past few years, accuracy may have been less important to its various stakeholders than universal acceptance...

[James MacAonghus]

I forgot to say, I used Hitwise data from May 2007, since the rest of the analysis for 2007.