Why Maps Are Like Search

Published: by

As of this writing, there are several major mapping providers used on the Web and in mobile apps: Google, Microsoft, MapQuest (owned by AOL) and, of questionable quality, Apple. As of this writing, there are two major search providers: Google and Microsoft Bing. Notice that there is a lot of overlap.

Additionally, there are some secondary players in both markets. AOL still thinks of itself as a search engine, sort of, kind of, although its behaviour in that regard has been questionable for years, and has been schizophrenic: is it a search engine? A content site? An online portal? A connection site? Apple does not do anything serious in search (yet), although Siri is clearly an early foray into doing it via mobile, and I expect a text search engine, like Google and Bing, built into one of the coming versions of iOS. Yahoo is problematic as a search engine of late (comScore's May 2013 results had Yahoo down to 11.9% of total), and they appear to have deprecated their maps developer API in favour of that incredibly successful fast-growing company with solid experience working with developers and open-source, Nokia. (Yes, that was sarcasm.)

The natural business customer overlap between mapping and search is obvious. Maps give context to search results, allow search providers to place results in relative locations to each other, and enable them to present results relative to a mapped location. Since both rely primarily on advertising revenue to provides services to free consumers, the sales and marketing overlap also make sense. Hence the various acquisitions over the years of map engines by major search providers. Microsoft bought Vexcel and Multimap; Yahoo bought whereonearth; Google bought Where2, keyhole, several others, and, of course, Waze.

But there is another, less-visible, overlap between search and maps: capital investment.

Several years ago, I had the pleasure of working with Stephen Arnold of ArnoldIT. Stephen is an expert in the search business space, and has even written books about it. At the lecture he gave where I met him, he pointed out that almost no competitors could take Google head-on. A search engine is only as viable as it is useful, and it is only as useful as its search index. That search index takes a lot of time and computing power to fill, even more to keep up to date, and, even in the era of rapidly falling storage prices, a lot of storage. All of these take capital; a very large amount of capital. Google's total invested capital in crawling, indexing, and storing the Web over the years likely is in the tens of billions of dollars. Only a company with huge amounts to spend and at least some history of having done so can compete head-on. Hence, Microsoft's ongoing attempt and partial success.

This doesn't mean a smaller player might not disrupt them by providing "good enough" - say 10-20% of Google quality, although the exact percentage can be debated - search results that are acquired at a tiny fraction of Google's cost. It just means that a head-on attack - same business model, but we will give better results - is nearly impossible, except for someone with some history and way too much excess money on their hands.

It turns out, maps are very similar. It takes a lot of money to get the world mapped in detail. Think how many different towns, cities, streets, buildings, neighbourhoods, rivers, lakes, airports, train stations, etc. there are in the world. While some of the data is publicly accessible, an enormous amount of it must simply be mapped by hand, perhaps using an iPad instead of the paper Colonel George Washington used in the mid 1700s (how many know that a big part of his success was his intimate knowledge of the terrain and how to use it).

In business lingo, we call these "barriers to entry." It turns out that, while not as high as search, the barriers to entry for mapping are pretty high as well. However, like search, there are ways to bypass each: focus.

  • Search: It is incredibly expensive to map the entire Web or large portions of it. On the other hand, if you limit your service to a subset, e.g. medical research articles, the cost is immeasurably lower, and you can bring domain knowledge to boot.
  • Map: It is incredibly expensive to map the entire world or large portions of it. On the other hand, if you limit your service to a subset, e.g. Washington DC, the cost is immeasurably lower, and you can bring domain knowledge to boot.

Last, mapping has one additional advantage. The cost of search is primarily capital; the cost of mapping is primarily labour. Crowdsourcing and what we used to call "user generated content" or UGC can allow a provider to bypass the investment in staff to provide a "good enough" map for limited but growing areas. This is the basis of the OpenStreetMap model, and MapBox built on top of it.

Neither of these attempts to take an entrenched provider with high barriers to entry in the form of capital investment head on; rather they attempt either to focus on an area that has materially lower costs with high value for domain knowledge, or use a network of people to spread the costs. If you can get 10MM people to give you $100 worth of their time, you just saved $1BN!