There has been a lot of speculation lately about a pending purchase of Groupon by Google. A lot of the more recent news recently has focused on all the reasons why this is a bad idea, especially around the purchase price and Groupon’s human-intensive sales model. I wanted to share a few thoughts on this given some of the history of Google, other Internet companies I’ve watched, and the future we’re all looking at.

The Unassailability of Network Effect Leaders
The number one consideration here is network effects. I was stupefied by Google’s purchase price for YouTube when it happened. My first thought was “wow, how can they justify that price”. I’m used to thinking about purchase prices for companies in traditional enterprise software terms where the math is about long-term value of customers, maintenance revenue streams, customer attrition versus upsell. I talked about this a lot when Oracle went on a spending spree and bought PeopleSoft (and thus JD Edwards) and Siebel . In those cases, you could do the math and realize that Oracle basically bought the ERP market at a pretty fair price (since then I think they’ve proven all those purchases were worthwhile).

The question I asked myself that changed my own opinion of the purchase price was “Could Google spend 3 billion dollars in any other way to catch up to YouTube”. The conclusion I came to was simply “no”. Metcalfe’s law is at work every day on the Internet and one of the things about Metcalfe’s law is that once you get big enough, you basically are unassailable until a completely new model comes along. YouTube was, at that point, unassailable. They still are, even though sites like Facebook (and back in the day MySpace) have compelling reasons for people to upload video. They simply paid for the winner, bet on their N^N expansion path, and took their time figuring out how to monetize it. With their recent announcement about their quietly built 3 billion display ad business, they proved the point here undeniably. I still believe there is no way they could have spent 3 billion any other way to now be the leader (remember Google has its own video service still). One could make the exact same argument about 750 million for AdMob, 3.1 billion for DoubleClick (seems like this paid off too), and now Groupon.

Groupon is the breakout network effect leader in the local business deal space. As much as many other services are great, the simply won’t catch up with Groupon. Perhaps ever. If you want to use non-Google examples, consider eBay versus Amazon Auctions or Yahoo Auctions. Or Microsoft’s 6.1 billion counter-purchase of display network AQuantive. Same reasoning there. This is why Google’s in such a bind related to Facebook – Facebook has surpassed the point where they are unassailable in social. Google can’t develop their way out of the problem. While people love to gripe around the edge cases, folks have engaged too far with Facebook to pull back. All my friends, pictures, tagging, etc. – it locks me in. Some have speculated this is why they invested in Zynga – as a way to reverse into a social network built on top of gaming as opposed to gaming built on a social network. Network-effect-based user engagement is the new Increasing Returns. I predicted APIs were the second coming of Increasing Returns , and I think I got that pretty dead on (see iPhone, Salesforce.com, Twitter, etc. etc..). Now I’m predicting NEBUE (network-effect-based-user-engagement ) is the third. You heard it here first.

Local

I’m stating the obvious here but local is going to be big in the next few years. Google signaled this when they moved Marissa Mayer (@marissamayer) over to the local team. They’ve launched a ton of local products (10-pack a while back, Google Places, Street View, etc.). But what they don’t have is a local sales force. Yes, having folks running around strip malls selling services seems pretty incongruous for a company that prides itself on self-service (e.g. AdWords), but to win in local you need to have the biggest and best sales force you can. All the local players from ReachLocal to Groupon to Yelp to Verizon have gotten good at cost-effective local selling. Some of them (e.g. Verizon, ValPak, etc.) have been doing this for a long time and have huge installed customer bases. Local businesses still buy primarily from people. With local ad dollars moving online, there is simply a scale race going on in local sales forces.

Google can arguably afford to make this work much better than anyone else because they are masters of systems, efficiency, and already have a lot of experience doing this with AdWords (Google has a huge sales team for AdWords which few people actually know about). All they need to do is give those sales reps multiple products in their bags to sell, and they’ll become very profitable quickly because most of those products are annuity products. That’s how the yellow pages became so successful so many years ago. In the grand scheme of things, having 2,000 local sales reps (to pick a random number) isn’t that scary when you already have more than 20,000 employees. The trick will be to graduate your customers over time to more self-service products and reduce customer service costs. I’ve never talked to a merchant who uses Groupon, but I’m guessing it’s pretty low-touch by now. That’s the beauty of coupons. You don’t need to integrate into people POS systems, give them printers to print orders, etc. You print a coupon that they can bar code scan and let them worry about it. It works even better because of the likely high breakage (e.g. purchased coupons that never get redeemed). It’s the health club model extended to everyone.

Advertising Distribution
Keep in mind that a huge part of Groupon’s expense budget is display advertising. We all know about Groupon primarily from their flood of advertising around the web. Google has a massive display network and can get incredible breadth, localization and cost optimization by putting them into this stream.

AdSense – The Hidden Weapon
Many times people forget one of Google’s secret weapons – AdSense. It’s been a fascinating and untalked about product for a while. I think it contributes to about 5 billion in revenue a year. What is more important though is that it’s installed on more than a million websites. AdSense has evolved to allow display ads (banners), which are now a big part of Google’s business and could easily be extended to allow affiliate-based Groupon ads. As content monetization heats up in the next few years, affiliate fees will become a viable complement to traditional CPM offerings. Google already has GAN (Google Affiliate Network), and now will have one of the biggest affiliate merchants in the world to offer as another AdSense service (e.g. if I own a blog let Google put Groupon ads on my site and pay me if someone clicks on it and purchases). I wrote a lot about this in another blog on real-time advertising and the future of content monetization. To go all the way back to the beginning, I think Google already has the breakout scale in “being installed” on content sites. With more products (text ads, display, affiliate ads), this will only increase their distribution further as they offer different effective CPM products to various sites.

Right now Google makes it money in two primary ways: CPC advertising (AdWords) and Display (AdSense, YouTube). They are merging the models with YouTube’s new promoted video’s product, which is CPC based [and the increasing effectiveness of search retargeting. If they add a third model — affiliate ads — it becomes an impressive platform for massive growth. Consider what happens if you search on American Apparel, and they can retarget that to a Groupon coupon for that vendor if they have it, rather than a display ad. This is extremely easy for someone like Google to do simply because they have such massive computational scale. So $6 billion dollars for Groupon? Doesn’t actually sound that much to me.
At least that’s what I think, but what do I know.

Recently I had the pleasure of listening to Greg Maffei (CEO) and Michael Zeisser (SVP Digital) of Liberty Media speak at CU’s Silicon Flatirons Entrepreneurs Unplugged event. While the conversation was broad ranging due to the vast nature of Liberty Media’s holdings, one comment from Michael jumped out at me. The reason that television advertising (in his opinion) has value is simple supply and demand. What he meant by this is that television still has the ability to create aggregated demand (e.g. the Superbowl, American Idol, Jersey Shore – where people actively watch the show at a specific time) while the Internet currently does not. Advertising prices are related non-linearly to demand and thus why a Superbowl ad still costs you $2.3M for a 30 second spot.

Clearly these dynamics are shifting in all directions right now. Television is being watched on-demand on the Internet (thus diffusing its spot-demand properties) and Internet live-streaming companies are becoming very successful. Witness the aggregated audience of 5.3M people that UStream had for the Chilean miner rescue. Setting aside predictions of when the cross-over happens (television aggregates less demand than the Internet), I want to focus on the advertising side of things.

Part of what makes television advertising successful is the predictability of aggregated demand. You know the date of the Superbowl. You know what it’s going to be about, and you likely know the kind of people that are going to show up.  If you’re Coca-Cola or Chrysler, you can have your agency of record (AOR) to get to work producing $2M context- and demographic-specific ad creative with lots of lead time. In the emerging world of real-time aggregated demand, you can’t. So how do you design advertising for a medium that’s totally unpredictable?

Let me give a recent example from an advertising perspective. You can’t predict when an Icelandic volcano is going to explode and thus cause 10,000 of travelers to be stuck in a location where you might have a hotel to advertise. So your AOR is of no use to you. Frankly, you don’t have time to build a bunch of creative and respond to the event anyway. What’s really the solution here? Well I think my friends at OneRiot have a good idea – it’s about content not products.

One of the things that OneRiot has innovated is the idea that content itself can be the advertisement. If you’re writing an article about Icelandic volcanoes and how travelers are looking for last-minute hotel deals, why not use a snippet of this content as the advertisement to attract a user to your site? OneRiot basically is making a market between two things – those who have real-time searches (Google, New York Times, etc.), and those who have relevant real-time content.

By constantly pouring over the content that their advertisers produce, they can match it to real-time events and automatically create ad units for you. This all happens without human intervention which is important in the real-time world. Once everyone figures out the price, they’re willing to pay and accept on a cost-per-click (CPC) basis. Now you’re in business. I’ve personally clicked on these “content ad units” because they ARE relevant. Which is why OneRiot sees the click-through-rates (CTRs) of 3 to 4 percent.

But here’s the problem: how do you price the click? The landing page of the ad unit, unlike an ad on AdWords, is the actual content itself. So by definition you’re sending someone into an RPM situation. For those unfamiliar to the term, RPM means Revenue per Mille (1000 in French). In other words, how much money does your site generate from 1,000 visitors. This is only something that most sites have started to think about. To give you an idea of how new this is, RPM doesn’t even have its own Wikipedia entry yet (as of the time of writing this entry).

RPM is important for two reasons. First, the classic way that content producers have thought about revenue generation is through cost-per-impression (CPM) based ads. These are display ads like banner ads and Google AdSense text ads that we have all become used to. Recently a crop of “sponsor-based” ad companies have emerged to simplify the direct-buying process as well. (See isocket as an example which automated the buying process for ads on TechCrunch and others). With more content inventory coming onto the Internet every day, CPM rates are constantly being pushed down, thus making the RPM of the site smaller over time. While readership is growing, content production rates will outpace increased readership rates over time. So the CPM loses in the end.

Content producers will need to start thinking of two things: how to optimize their sites for CPM, and how else to monetize content (more on this below). Second, when you’re playing the traffic game in a CPC world, it quickly becomes all about the numbers. If you have a RPM of $50.00 on your site and your average CPC for a content ad is $0.06, you’re losing money on every visitor. I am sure that CPCs for content ads are very cheap because the market is young. But painfully, the more success content ad units have, the higher CPCs will go due to demand in the market. Therefore, the harder it will be to generate a cost effective CPC based content ad campaign. I think this will force one great thing – for content producers to actually look at their fully loaded RPM and cost and to understand what they can afford to drive traffic. Much like retailers driving traffic to their ecommerce sites, it’s more sophisticated than it first looks. The math I ran above was simple, but it was fatally flawed in two ways – expense and repeat visitors. If your RPM is $50 on your site and it costs you $25 in expense to deliver content to those readers (servers, bandwidth, staff, etc.), you only have 2.5 cents to play with in CPC. But if your readers come back once a month without clicking an ad again because they fell in love with your content, your allowable CPC is really 12 x 2.5 cents or 30 cents. This is very similar to the CPA calculation a lot of retailers go through with CPC advertising. If your buyers are repeat buyers, you can spend more acquiring them up front.

All of my math set aside, the more important topic I want to mention is that content monetization is at a very early stage. Right now, most people think about their primary content monetization strategy as CPM based advertising. But as my friend and colleague Oliver Roup, CEO of VigLink, likes to say, that means you’re only monetizing 20% of your real estate (the top and right of the page usually). What about the actual content itself? VigLink helps companies automatically affiliate any link they have to an external retailer (full disclosure, I am on the board). If you write a piece of content that recommends a camera, the link to BestBuy you put in the content as a convenience to the reader is actually something that can generate you a referral fee if they buy. There is a whole industry around this simple concept; it’s called the affiliate industry. It is a massive one that few content producers know about still (but the smart ones are making tons of money from). The great thing about VigLink is that it costs nothing, and VigLink does all the work for you. In other words, you don’t need to sign up with BestBuy’s program, and Amazon’s and Dell’s and Payless’ – they’ve already done it for you. VigLink is a second generation affiliate technology in the sense that sits on top of all the existing affiliate networks.

We see incredible uplift in RPM for some content sites that have installed VigLink (it’s just a few lines of Javascript embedded on your page like a Google Analytics tracker). I think that as more content sites contemplate CPC content ad units, they also need to contemplate how to raise RPMs. Affiliation will be the next big growth area for content monetization. The infrastructure is already built, it doesn’t cost anything, and like CPM ads, you can learn how to optimize your affiliatized links as you get more data and become smarter about it. A big part of VigLink’s value to content sites is around these analytics.

So all of that is to say that not surprisingly, advertising is once again evolving to meet the habits of the user base and the technologies that support them. It’s not only moving money from one bucket to another, but increasing the audience that will find value from it. To take advantage of this shift, site owners need to think carefully about how they make money, what their costs are and how the equation works out. I wish the best of luck to both OneRiot and to VigLink. I think both of them are creating and developing a whole next generation of content monetization on the Internet. It’s  exciting to watch.