BUSINESS BLOGS
BUSINESS BLOGS
category: business
13 Oct 2006

Months ago, I recall seeing Yahoo!’s contextual text ads on YouTube.  Then those got replaced with Google’s Ad Sense (though it could have been Google first, then Yahoo!  The order is immaterial to this story).

Anyway, neither ads lasted very long.  After testing both Yahoo and Google’s text ads, YouTube dropped the contextual ad idea and opted for one display ad per page at the top (of the “Press the Fart Button” variety, sadly). 

Will Google proceed to plaster Ad Sense on all of YouTube’s pages?

Just a couple of days after Google acquired YouTube, the search leader wasted no time and added its cash cow Ad Sense product on YouTube’s pages, once again.  This time alongside the right hand side of the page.  Note, however, that upon searching on the site, I did not see Ad Sense on all pages.  Will Google proceed to plaster Ad Sense on all pages?

Herd Mentality?

It should, because that’s what Wall Street expects it to do… but it might not, because Google has historically outsmarted Wall Street (not exactly that hard, sometimes). 

Google is a company packed with smart folks, their PhDs probably know basic math and have done it: the fact is that even if Google puts Ad Sense on all of YouTube’s pages, it will only generate a tiny amount in revenue and actually hurt its market-leading ad formula (more on that after the analysis).  Of course, it will not hurt its ad formula; if anything, if Google were evil, the ad formula would hurt small advertisers.  But more on that later.

Note that Google’s revenue from Ad Sense is simply:

Click Through Rate % x Ad Sense Impressions x Cost Per Click $ x (1 - Partner Revenue Share). 

Since Google bought YouTube, it will retain 100% of the revenue.

Some numbers to consider (some background: I worked in the search engine industry and the publishing sector; in other words, I’ve been both on the sell and buy side of search):

- Google sells text ads alongside its organic search results, and offers advertisers to extend their reach on Google’s contextual network: i.e. those text ads you see on many sites alongside the content. 

As you can imagine, a web surfer is far more likely to click on a text ad after searching for something than when reading something on a site.  As such, while click through rates on text ads alongside organic results tend to yield anywhere from 1%-30%, on Google’s network, I guesstimate that it’s anywhere from 0.1 to 2%.  I have worked in the search space and have advertised on Google’s contextual network, and I am largely seeing a 0.1%-1% CTR on Google’s contextual network from my five years experience using the product.  If others yield more with their campaigns, more power to you!

Everything else in my model comes from third party sources.

- 100 million video streams per day: comScore July, 2006.
- Average Cost Per Click: Fathom Online.

A few additional comments:

1) Adjusting CTRs 

In my model, I do not even bother with any CTR ranges between 0.50% and 1%.  CTRs can go that high, if say you are on a news site reading something text-based and you see a text ad for something related; the fickle YouTube faithful probably will have a lower propensity to click on a text ad. 

A quick note to back this further: according to Alexa (not the most reliable source, but a good metric to consider here), YouTube averages 10 pageviews (and thus video streams) for every one unique user… so right there, it is fair to assume that if a user watches 10 vidoes, they have passed up on 9 ad sense impressions, even if they click on the 10th one (and there is no guarantee they do), that is a 10% CTR.  Do you really think that every single exit traffic click would go to Google’s Ad Sense?  Of course not.  Even if 10% of YouTube’s exit traffic clicks on an Ad Sense link, that’s 1% CTR.

2) Adjusting CPC

I run the numbers using both Fathom Online’s $1.27 CPC average, as well as a more realistic $0.50 CPC. 

After all, most of the keywords that push up the average are search query terms like career training, mp3 players, camcorders, webhosting, isp, security, lawsuit etc.  It does not take a Google PhD to figure out that - much like on MySpace - the bulk of the context of YouTube’s content (either legit or of copyright violating nature) is entertainment related, a more realistic average CPC is in the $0.35 - $1 range.  I am using $0.50 in the additional models.

Here are a number of scenarions, with the different assumptions:

Here are two more sets of financial model:

So, what is our estimate? 

Of all of these scenarios, my answer is based on:

- YouTube does generate some 2/3, or 67%, of their streams on YouTube.  Sure, a lot of people see embedded YouTube videos on MySpace and other sites… but once you know about YouTube or need any random video, you go to YouTube and generate a lot of streams per visit.  If you doubt this, think of the 10 pageviews per user stat off Alexa.
- The average CPC that Google will get off YouTube will be much closer to $0.50 than $1.27.
- The CTR that Google will get off YouTube’s contextual inventory is 0.25%.

Assuming these stats, looking at the various numbers above, assuming on growth or falloff in YouTube’s traffic:

Google would make $30,568,750 per year off Ad Sense on Google. 

Suffice to say I think this is extremely low, and perhaps much lower than Google’s internal analysis suggested before the deal.  Then again, at $30 million per year, YouTube would probably rank very high atop Google’s top distribution partners.

But ask yourself, if Google was such a cash cow for YouTube on YouTube, why would YouTube - a company with about $12 million in annual bandwidth fees - test Ad Sense (and Yahoo!’s equivalent product) earlier this year and then take them off). 

Of course, note that this number can shoot up by simply tweaking one variable, if the CTR goes up to 0.5%, you’re looking at $60M.  Get the CPC up to $1, you are now at $120M, etc.

But, I doubt any of that will matter. 

So Why Did Google Buy YouTube?

Clearly, Google did not buy YouTube for the technology alone.  Some say that Google would have at the very most paid $100 million were it for YouTube’s technology.  Of course, Google shelled out $1.65 billion in stock for YouTube for its audience.  But everyone who says Google can make it back with Ad Sense is wrong for the simple reason that contextual ads do not work on social networks and video file sharing sites.

Google Learns From Math… and History

Google made this acquisition because it saw MySpace triple in size after News Corporation’s Fox Interactive Media (FIM) bought it, and to protect its bread and butter against Yahoo! and MSFT, it went out and paid $900 million to FIM for the rights to power search and serve contextual ads on MySpace.  Here are my comments on why and how Google overpaid there in a defensive move.

If time is money, Google bought YouTube because it had the former and lacked the latter: since launching Google Video in January, it failed to make a dent in YouTube’s lead, which grew to 100 million video streams per day.  And, what’s worse is that MySpace went out and launched its own video strategy and that immediately became larger than Google’s video offering.  If the MySpace acquisition was a manifestation of Rupert Murdoch’s ambition, well, Sergey Brin and Larry Page shot right back by acquiring YouTube, who had clearly become the cool kid in the city where everyone desperately tries to retain their cool, “it” factor.

Enter Display and Video Ads

This analysis is not meant to discourage Google from plastering Ad Sense all over YouTube (you know, since they always wait for my approval on things, of course…).  They will and should do that.  But what Google should do is use YouTube as the canvas on which it can finally compete with Yahoo! (remember them?) in display banners and video advertising, two areas that Google badly needs some exposure in, since video ads are set to grow faster than search, even though search will remain the main beneficiary of online advertising with 40% of the market.

Why Google Itself Might Yank Ad Sense Off YouTube

Believe it or not, after the trial period, Google itself might pull off Ad Sense.  Actually, that depends on whether or not they want to be evil or not.

Here’s the thing: Google will generate incremental revenue by having more text ad impressions served up on YouTube since Google charges clients per click.  More impressions, more clicks, more revenue.  That much is simple.

But, the catch is that the marginal CTR percentage on YouTube pages is going to be so low that advertisers advertising on Google’s contextual network will see their “content network” CTRs slide.

So, what’s the problem?  Read on.

Indeed, Google charges clients per click (Cost per Click, or CPC), but it does not rank advertisers based on who pays the highest CPC alone.  John Batelle covered this very well in his book The Search.  This is what Yahoo! is now trying to emulate. 

How Google’s Ad Formula Works 

Google’s top paid text ad goes to the client who can generate the most revenue for Google.  Total revenue is simply: CTR x CPC.  So if an advertiser seeks more reach and opts to be included in Google’s contextual ad network, it tends to see an increase in total clicks but a drop in overall CTR (its search CTR remains unaffected, but its content network CTR drops).

The thing is, for better or worse, Google forces advertisers to maintain high CTRs, which is good for the overall user experience: this ensures that the search text ads are relevant to the organic results.  But, and this is the key, when the CTR drops, Google prompts advertisers to increase their CPC bids… the implication is that Google’s Total Revenue must be a certain level.  This is why Google is a cash cow: it has taken the pay per click model to a whole new level.

So, if Google were to:

a) suddenly become evil: it will plaster Ad Sense all over YouTube, pummel CTRs, force advertisers to big higher CPCs to maintain its total revenue at a given level.

b) avoid from becoming evil: they will realize that not only does YouTube not stand to generate much revenue for them, but that they should leverage YouTube’s platform to enter display and video ads (these could be user-initiated video ads embedded in the regular traditional banner placements and not pre-rolls) and mainly, avoid forcing advertisers to further bid up keyword prices.  After all, while it might be tempting for Google to do just that (raise keyword prices), some are wondering whether or not there is a bubble in keyword bid prices.

Disclaimer: I own shares in Yahoo! but that does not mean that you should too.