For work and personal reasons I’ve taken a 3 year leave of absence from the blog. Now that I’m “working” for myself again, there will be plenty of time to get back to the community and share random thoughts.
As the blog has taken on a decidedly technology focus in my absence, the first topic that’s piqued my interest is the concept of information distribution. I will be writing a series of posts on it, ranging from topics like how information distribution has evolved in the Internet Age, how business models are affected and moats are created and destroyed in the process, how its evolution differs between China and the West, and any investment implications. I suppose a lot of this stuff you already know because it’s just common sense and some of it, if you didn’t know, is probably pointless crap to you, so I’m going to break tradition with the usual long-winded style of this blog and keep the posts short for your benefit. Hopefully you will enjoy the new format.
Part 1 – The TAM of Search Engines
I don’t own Google but I follow the company a bit. This is not an expensive stock. With some reasonable adjustments it’s not hard to get the core search business trading at less than 20 times underlying earnings. The main pushback on the bull thesis that’s emerged in the past year or so is the extent to which the search engine business is nearing saturation, considering that Google and Facebook now account for over 70% of online advertising globally (ex-China). At some point the party has to end if you’ve already won all of the chips, right?
Not that I’m bullish on Google or anything but I think this train of thought really misses the point in the greater context of information distribution on the Internet. To truly appreciate the nature of information distribution, we need to think in a broader context and challenge some assumptions: what if “online advertising”, or even “advertising” is not the right way to measure the TAM (total addressable market) for the search engine business?
For starters, “advertising” is only a subsection of what we label as “marketing”. Just think about how much commissions department stores pay their salespeople. While these costs are not considered advertising in strict definition, it is quite conceivable that some of these dollars would get re-allocated to search engine advertising as retail sales move online and brands shift their budgets accordingly. Similarly, companies spend enormous sums on things like trade shows, business travel, sponsorships, cold calls etc., all of which are marketing channels for potential customers to learn about their products. All of that information could conceivably be acquired by the same customers with a simple click on a search engine, a shift in information distribution that is well underway today.
If we think in these terms, the TAM for search engines would be an order of magnitude larger than the figure that’s often quoted by investments analysts which limits the TAM to a narrow definition of “advertising”.
And we are just getting started.
If we correctly define the role of search engines, we can see that what they are really designed to address is actually something much broader – “search cost”. Search cost is the biggest component of what economists label as “friction cost” in an economy and it can exist and be addressed in many different forms. For example, before the invention of the Internet, the most widely used tool for addressing search cost was actually the physical location. Imagine you owned a 200,000 sq ft building in downtown Manhattan. If you used that space for storage, chances are you would be able to charge the same rent as a 200,000 sq ft warehouse situated in the suburbs; but if you turned it into a mall it would likely bring in many times the amount of rent that an equivalent mall can bring in outside the city. All of that excess rent is a form of marketing that brands and retailers pay to address “search cost”. In a purely online environment, the physical location is disintermediated and what companies would otherwise pay in excess rent in an offline setting would presumably get re-allocated in the form of Amazon commissions or Google keyword ads (By the way Amazon is really just another search engine with a “vertical” emphasis – even though Amazon commissions are not labeled as “advertising” in accounting terms, they are in every sense a form of advertising in substance; we will explore the topic of “vertical” vs “horizontal” search engines in detail in the next post).
In fact, in a market like China where online penetration of retail is comparatively high, the fungibility of rent vs. online advertising is already on full display as many retailers and brands in their expansion strategy now routinely conduct a cost of traffic comparison between setting up new physical locations and raising the budget for Tmall keyword bidding, as if those were interchangeable marketing channels.
As a thought exercise, imagine a world where at some point in the future all consumption moves online. What will happen to the way search cost is addressed? Excess retail rent will be fully eliminated, direct-to-consumer sales will no longer exist as a profession, travel agents will be out of their jobs and we will no longer pay late fees at libraries (which won’t exist). At least in theory, all the money spent on these belong in the TAM of search engines. Once you add up everything, the real TAM becomes quite considerable. Hell, I wouldn’t be surprised if search cost consumes 10 – 20% of GDP for a lot of countries.
The often debated metric of advertising spend as a % of GDP, in my opinion, is more or less a meaningless statistic as it is merely an output of how we label and account for different buckets of search cost. Take Alibaba vs. Amazon 3rd Party as an example, one is the largest advertising platform in China and the other reports close to zero advertising revenue. In substance both sites serve a similar purpose of generating traffic leads for retail merchants; they simply report the nature of their revenues differently. As a result, this causes “true” advertising spend in the US to be under-reported compared to China. Personally I don’t take this metric very seriously as it is too narrowly defined to have any informational value. What matters more is search cost dollars not advertising dollars.
On an unrelated note, this search engine discussion brings up an interesting point about the danger of the traditional top-down “TAM analysis” that analysts employ when attempting to size the potential of disruptive technologies. With any business enabled by new technology, while the existing size of the market provides a useful reference point for TAM, it is often a very imprecise one because it assumes a static relationship between the size of a market and the technologies that enable it. What we often forget is that the size of a market is actually an output of technology constraints and as technology evolves, the market size evolves with it.
I will give you one example. Vipshop is the dominant online off-season apparel retailer in China. The bull thesis for this name is relatively straightforward: apparel retail in China is a $300 billion market of which the off-season segment is about 20% of overall sales. This would suggest a TAM of $60 billion which given Vipshop’s $8 billion of revenue means the company is less than 15% penetrated. Selling off-season apparel online is a superior model that will displace offline retailers and Vipshop, similar to TJ Maxx in the US, is the undisputed leader with an unassailable competitive position.
I actually don’t disagree with the second part of this statement (competitive position) but the first part (TAM) is problematic. This is because as an off-season discounter Vipshop procures the majority of its inventory from excess inventory in offline channels. As apparel retail continues to shift online (already at 40% online penetration in China, even 50%+ in some cities), the availability of this excess inventory shrinks significantly because online in-season retailers have much lower inventory needs than their offline counterparts (online means no inventory needed for in-store display, shorter supply chains and better sales predictability). In other words, the more apparel sales shift online, the smaller Vipshop’s TAM will get – there are always two sides to every coin. Vipshop’s share of its TAM is therefore a fair bit overstated, which seems to explain why growth has slowed from 80% to 30% in less than 2 years.
Actually, most of the time, new technological developments have a tendency to shrink the TAM as technology is usually deflationary (the search engine being an exception), so be careful when you see IR slides of tech companies where management takes an estimate of the market size today and declares that number as their company’s financial destiny – it most likely overstates the actual TAM.
—- to be continued…