“It’s funny how fallin feels like flyin,
for a little while…”
On July 1st, Google announced its intention to acquire ITA Software. ITA owns a primarily B2B airfare search and pricing system called QPX. Several of the leading online travel sites, like Orbitz, Kayak, and Bing Travel, use information from QPX to power their airfare search. Many in the industry view this move as a seminal event in Google’s history, as the company makes a decisive step from being a general search engine, into more structured vertical search. Certainly, Google already offers vertical search in Images, Videos, Maps, News, and several other categories. Despite that, ITA feels different. Perhaps the difference is that this is a step into a vertical where many independent incumbents, like Priceline and Expedia, who are material customers of Google, have large established businesses.
There are two reasons mentioned for why Google feels compelled to dive deeper into verticals. The most straightforward explanation is competitive pressure. Following its own acquisition of Farecast, Microsoft has subsequently launched Bing Travel, a much richer travel search product than offered on Google today. The second argument given for the move is that by moving deeper into verticals, and closer to the actual transaction that Google can actual make more money per visit. This argument suggests that CPA (cost-per-action) is a fundamental improvement over Google’s current business model, CPC (cost-per-click). The competition argument seems obvious and accurate. However, it is not at all clear that going deeper in verticals will raise Google’s revenues. In fact, there are several scenarios where they could actually go down.
Let’s first address the easy part – competition. Bing buys Farecast and Google needs to respond. This make sense, but it is not the whole story. If you are searching for a book or an author you go to Amazon, or at the very least you do a search like “Man in Full Amazon” so that you go directly to the page you want on Amazon. The same is true for hotels with TripAdvisor and for restaurants with OpenTable. These sites offer deeper and richer experiences for a vertical searcher precisely because they incorporate deep meta-data, faceted search, transaction connectivity, and typically a form of community or UGC (user generated content). These things simply do not exist in the simple but limited Google user interface that Om Malik affectionately refers to as “10 blue links.” So Google has competition in verticals not just from Microsoft, but also from best of breed vertical sites offering users a richer, deeper experience.
Some have suggested that Google’s move into a deeper vertical experience is more about greed (more money) than fear (competitive) response. The argument voiced by Barclay’s analyst Douglas Anuth and others, is that by moving closer to the transaction, Google can ask for CPA fees, which naturally carry higher margin. Clearly, a single CPA fee will be much higher than a single CPC fee, but you will also have much fewer of them. The variable that links the two is conversion. One can certainly argue that Google can drive higher conversion if they can help drive the customer closer to the actual result they need. This will require a materially better product. Even then, however, there are two key reasons Google may not see higher revenue with deeper vertical integration.
Reason #1: Irrational CPC Pricing on Google Today, “Uber-Optimization”
Some might argue that Google’s current bid-based CPC model results in “optimal” pricing. The argument would be that the market-clearing price settles out at the precisely rational price for each and every keyword pair. A market-place model naturally results in efficient pricing. While this makes logical sense, we all know that there are companies participating in the CPC purchasing game who simply are less sophisticated than others. Moreover, many of these buyers win bids and have huge CPC budgets. The point is that there are plenty of startups, arbitragists, and even large companies “experimenting” with CPC purchasing in an attempt to gain an edge. The winner in this bidding game isn’t the most rational, but simply the one with the highest price. Certainly, over the long run if a company irrationally pays too much for CPC ads, they will eventually go out of business. But the key word here is eventually. For a long period of time, they are paying an “uber-optimized” price for their keywords.
If there are enough of these players in the market, then Google’s CPC prices aren’t economically rational. Rather, they live slightly above that level driven by irrationality and experimentation in the market. If you have a hard time with this theory try typing a search term like “laser treatments” in Google and look through the list of CPC purchasers on the right side of the screen. Would you put money in these companies? Do you have confidence they will be around in ten years? Have you ever heard of them? It gets even better. Many believe that Google uses a low quality rating score on these “middle-men” to force them to pay a higher fee for a single CPC, thus getting an even higher price than previously discussed. That’s right, for certain CPC buyers, Google has a mechanism for extracting an even higher price, even if the buyer is already the high bidder!
As Google moves past its “10 blue links” model and connects directly to airlines, hoteliers, etc, it will be removing the irrational and arguably temporary middle-man from the system. This fleeting but determined participant, very likely has a negative long term ROIC, and Google, riding the brilliance of the CPC model, stands as the beneficiary. With this player out of the system, and with the connections directly to the service provider, the model will naturally trend to a more efficient pricing. You will have fewer larger players, who are all more rational, and all more experienced. As such, you would have to expect more rational, and therefore lower, pricing. Building a better product could actually result in less lead-generation revenue.
Reason #2: Moving from a Marketing Channel to a Transaction Channel
If you have ever sold anything on the Internet, ask yourself the following question. What is the maximum amount you would want to pay for a transaction fee? 5%? 10%? There is data in looking at typical affiliate fee percentages, which can range from say 4-15%. Amazon charges 6-15%. Ebay charges about 11% (with Paypal). Comparison shopping engines make even less. When an etailer assumes they are “always going to pay” for something on every single transaction, they are very sensitive to the % of revenues, as this payment will always reduce their margin. One could assume the general average for all affiliate fees or similar distribution type arrangements is around 10%.
Now, ask someone in your marketing department how much they are willing to pay to “acquire a customer.” While I don’t pretend to support this logic, the Lifetime Value of the Customer (LTV) model depicted herein mesmerizes many marketing managers. Using this simplistic but highly regarded model, many marketers justify “acquiring a customer” not as a percentage of revenue, but as a percentage of life-time value. The key to reaching this Zen state of marketing awareness is to believe that Google is sending you this customer only this one time, but for here ever after this customer is going to come directly to your own site, bypassing Google. This logic supports a much larger denominator, known as LTV. With LTV, ad buyers are easily willing to spend 25-50% of a first purchase in order to “acquire a customer.”
We could talk forever about the LTV formula, and we could argue back and forth about its efficacy, but that would miss the point. Consider the following assertion. People that are buying CPC ads are frequently marketers, and marketers are much more likely to think in terms of LTV. When you enter into a CPA deal it feels very transactional. When you do deep integration it feels very transactional. And, if Google is building a deep vertical site in travel that will pass leads to companies like American Airlines it will feel very transactional. It will be harder and harder to assume that you are “acquiring a customer,” and it will feel more and more like you are paying a distribution fee to a channel. As such, it may turn out that moving deeper into a vertical will puncture the illusion that marketers are “acquiring a customer” from Google, and get them in touch with the fact that they have a permanent CPA “transaction fee” they need to “pay” to Google. The end result is a lower overall rake for Google with the per transaction model.
Could this be wrong? Absolutely. Perhaps conversion rates will triple due to the incredible design, implementation, and ease of use of the Google’s new product, more than offsetting the two points we just mentioned. Or perhaps, Google will have such a powerful place in the travel ecosystem that travel companies will simply be “price-takers.” If this is the case then Google will once again find the exact right way to optimize their business model. It is equally likely, however, that Google’s current business model is highly, highly optimized and tweaking it may have as much risk as upside.