One of the pieces of news that was making the rounds over the past week, at least in my little corner of the twitter-sphere was news that Watson a computer system baked up by the fun folks at IBM beat the pants off two of the all time best players on the popular trivia show Jeopardy. Like its predecessor Deeper Blue in 1997 who beat the pants off of the then world’s best human chess champion… Watson was designed from the ground up to perform a specific task, and to do it quite well thanks to modern capabilities around processing power, data storage and hundreds of simultaneous algorithms tasked with interpreting the natural human language.

But reading a bit more of the press about the event, something caught my eye, a reference to the fact that Watson doesn’t even use the Internet.  To which my immediate reaction (yes I think I’ve become a jaded and skeptical New Yorker) was something like this, well if it coulda, it woulda, so since it didn’t…  well something is up.   And my suspicions were confirmed when the author of the IBM post was kind enough to elaborate that yes he believed that accessing the free and open Internet for information would have been detrimental to Watson’s performance. Which upon reflection makes all the sense in the world…  why would you sift through the entire Internet of information, when you can carefully curate all the information you need for the job in less that 1 terabyte of data held in 90 servers stacked up back stage.

Now like most people, I don’t have a clue as to how much general knowledge trivia there is in the world, uh… a lot? But am I a little surprised that it  can all be crammed into 90 computers, well yeah I guess so, I’ve never really thought about it. I am certainly impressed with the fact that Watson can fish out any little corner of it in about the same time as it would take Alex to type in the question.

But I guess my whole point here is the bigger picture stuff… the folks at IBM wanted to solve a single problem… find answers to trivia questions.  And with the state of things in the world of processing, data storage and algorithms getting, storing and retrieving that information was best done in a closed environment…  the 1TB of factual data necessary, apparently not that big of a deal.

So getting back to local search. Now I had not dealt with buying or selling web search in quite some time now, but within the past couple of years I have started to go back to shows like SMX East, the east coast edition of one of the larger search marketing trade shows, where the subject of local search is a pretty hot topic these days.  What surprised me a bit is that finding places and things in the real world was generally just viewed as more or less the same as searching for and finding web pages about places and things.

Maybe it’s just me but this just seems downright weird.  The web is home to billion if not trillions of pages of content on topics ranging from what Julie in San Jose had for dinner last night to how to calculate the weight of an African tree frog… 99.9% of which has absolutely nothing to do with what’s in the ½, 5 or 10 mile radius around me where I live the vast majority of my life.

Now if someone can sort out the web stuff that’s relevant to my little physical personal world that would be an improvement, but still people put some pretty useless and irrelevant stuff on the web, thousands of people I don’t know or care about and a huge chunk of retail businesses with no significant web presence to speak of.

What I need is just access to the stuff that is relevant to me, good detailed information on stores and products, people I know or might like to know or communicate with, events, relevant news etc.  Much of this information already exists, but like with Watson, just because it exists doesn’t mean that one can get at it quickly and easily, and sorting through the entire web of content to find it, well just wouldn’t be the best way to do it.

Until last night I wasn’t very aware of what folks were doing in terms of creating hyper local networks, but at the LBS Apps developer meet up last night here in NYC,  a company called Nearverse demo’d  their app LoKast. The application allows an organizer to create an ad hoc local network between users of the application who all share a common space of between 300-1000 ft… using a combination of Bluetooth, wi fi or your cell carriers’ wireless data connection.  With file sharing at its foundation, the idea is to allow folks who share a common space to share, well…almost anything digital that they want, including videos, songs, photos, contacts or webpages.

Now Nearverse doesn’t seem to be looking at areas around local search necessarily, and I am not sure if it really directly fits, but it would seem to support the idea that just because there is the capability to push info out to the one mass Internet doesn’t mean that this is the best and only way to do something. Perhaps million of tiny micro locally relevant Internets that just contain locally relevant data, or a massive database of only locally relevant data that is built around location from the ground up would be two novel ways to help connect people with just the relevant info about the world directly around them.

I love Yelp and use it frequently, and to a degree I get the rationale on why its a potentially attractive acquisition, particularly for someone like Google that is looking so aggressively at targeting the local brick and mortar businesses, and their advertising budget. Yelp has many of the right relationships with local advertisers and at the same time has a nice content creation tool and user base who rabidly create tons of local oriented content… all a very nice fit with what Google does and where they’re looking to grow.

 But the deal between Microsoft and Localeze in my mind directly tries to fix something that is currently wrong when you search for local businesses. So many times I try to fire up a mobile browser in the hopes of searching for things around me, and so many times the experience is just so damn underwhelming. No sign of businesses that have existed for years and years, no understanding that I am a consumer looking to walk into a retailer and dont care if someone deep in the skyscraper in front of me runs a web LLC out of their apartment on the 25th floor.

I can still stand on the sidewalk peering into the window outside of Sohpie’s Cuban Cuisine staring at huge baskets filled with yummy empanadas and then turn and do a mobile search for the term ‘empanadas’ and end up being directed to go miles away to Empanada Joe’s, Empanada Mama’s, and Reuben’s Empanadas.

Currently the local search solutions are either far too simplistic in understanding the search query… ie search for ‘empanadas’ and get back a list of places with ‘empanada’ in their business names, or they return 372,000 page that match the place and the term ‘empanada’ which is an unwieldy mess to sort through if you’re just standing on the corner with 15 minutes of your lunch break left.

Localeze clearly has this problem in their sights and is looking to help connect mobile searchers with the closest place to fill their bellys with empanadas even if the place selling you those empandas is called Sophies.

For Google at least, there may be a day when all of those Yelpers will rave about the great empanadas at Sophies and the Google algorithm would put aside those 371k other pages, and put two and two together and match the Yelp content about empanadas at Sophies with the Sophie’s business listing and give me a great search result. But the problem for Google may be that it’s special sauce is built around links (and votes) between web pages not necessarily analyzing the content or intent of the words within pages… so whether or not Google can effectively and reliably make that connection, particularly without owning the content is still a big TBD.

Per John Hanke, VP of Google Earth, Maps, and Local from a recent TechCrunch article : ”PlaceRank is like PageRank for places, it tries to figure out how prominent a place is based on factors such as references on the Web, reviews, photos, how many people know about it, how long its been around.”

By the way I think it’s notable that the thing being “figured out” here is “prominence”.

Now I understand that you’ve got to start somewhere, but I would argue that the tactics used in web search engines don’t really apply to spatial search much and it should be treated as a completely separate animal.  Web pages are about text and the authors of that text linking to (and as a result voting on) other web pages, in order to determine a pages’ ‘prominence’. 

I think there are a few key differences when looking at location and spatial oriented search:

• The “prominence” of a search result is relative to things like distance and the convenience of alternatives in local/spatial search, versus something more absolute in web search where you’re simply clicking on a link to ‘get there’
• Determining “prominence” is very important when parsing through 1 trillion pages of “always available” information, but in the more dynamic yet much more limited options of local search something as simple as solving for “highest prominence” may not be the right answer
• The true “linking” happening to a physical place is not happening on a website, but through foot traffic and phone calls… and the traffic links between places is not captured on a webpage at all, but on a handset or a carriers’ back end logs
•  The stuff being searched for could and should exist in a variety of mediums, not just html on webservers… find a person from their mobile device, find an item from an inventory system, find a bus from a location sensor.

I suspect  that there will be some incremental improvements over time with matching online information to offline stuff, but I also suspect that we’d be better off by blowing up the existing search model and starting over from the ground up with a model designed purely around location specific spatial search, that merely taps into the vast reservoir of online content only when necessary… rather than serving as the foundation.

• Google already has over a million and a half advertisers across its global business, sounds like a lot huh? Well there are 16 million local businesses listings in Yelp alone, so there are still a ton of potential advertisers who don’t use Google… and the ones that the Yelp sales team deals with everyday are of the local brick and mortar store variety, the same ones that very often don’t even have a web presence, let alone an AdWords account.
• Google already developed Google Place Pages to give every place on earth a webpage.  Yes very philanthropic and all, and yes now it’s great that the Central Park Carousel now has its own web page.  But one of the expected by-products of creating PlacePages was that those brick and mortar retailers who never really cared much for the web and didn’t have a relationship with Google, would wake up and say ‘wow, now I have a web page? thanks Google, let me claim it, and build on it, and take care of it and oh yes buy some advertising to promote it.’  It’s just possible that the initial reaction Google is seeing is that it’s not going to be that easy and maybe they need someone like the sometimes heavy handed Yelp sales force on the phone to force the issue a bit.
• Google lives and dies off of the creation of content and information on the web.  In the same way that it wanted to own Blogger in order to be closer to those who have created 332 million pages of blog content on anything from pet Schnauzers to the history of the bowling ball, Yelp has established itself as an important tool and industry leader for folks wanting to create content  focused on the local community and businesses… some use the term ‘lifestyle blog’.  While it looks like Google only indexes a far more mundane 25 million pages of content on Yelp, the commercial potential for this content is in many ways much more rich, since there are real live companies with ad budgets behind many of the pages, and something there to be bought and sold which makes it fertile ground for advertising demand.  

Some interesting numbers to consider:

• Size of the quickly dying yellow pages business: $15 billion. Will your Yellow Pages Rep be replaced with a Yelp Rep that down the road could be a major seller of AdWords as the new alternative to that yellow pages ad? It would mean more money in Google’s pocket versus relying on local re sellers.
•  Yelp currently has 8.5 million reviews, with 3 million businesses having at least one review… that’s almost $65 per review at the rumored buyout price.  Seems high, but not crazy high considering what people pay for an Associated Content piece… and it will only grow from here at little to no cost to Yelp.
• Yelp sells an estimated $30-40 million per year in advertising via 200 local sales people, I couldn’t find an advertiser count for Yelp except for some off handed remarks about having “tens of thousands of businesses accounts”, but using as an estimated 1.5 million advertisers for Google generating $22 Billion in annual revenue, the average advertiser on Google is worth nearly $15k in revenue… I’d hazard a guess that this is at least 5-10x what Yelp is getting, probably much higher.  So the opportunity to get more out of the Yelp field sales team by giving them more to sell seems significant.

 Just to keep this simple, let’s say that there are 15 million brick and mortar retailers in the U.S. (rough estimate) and that Google has 1.5 million advertisers currently as customers. Even if all the existing Google advertisers were brick and mortar advertisers and in the U.S. only they’d still only have 10% penetration of available advertisers. Now in reality over half of Google’s revenue is international and you can bet that a huge chunk of those advertisers are pure e-tailer with no physical store in sight.

There is only so much searching happening on the web and only a small percent of advertisers engaging with Google to try to reach those searchers, so if you’re Google what do you do to speed things along? Well you try to provide more stuff to search for, make it super easy to search for it, and try to engage the 90%+ of advertisers that don’t current engage with you.

So lets look at some of the newly announced efforts made over the past few days, particularly around location and expanding beyond the virtual world to the physical one.

The Gold Star Approach for Small Businesses: If you’re really good Google will give you a sticker!

A few months back, google created placepages for every place on Earth. If one of those places happens to be a business, Google obviously hopes that the business takes an interest in its placepage and engages with them to improve it and make it better. This week Google launched a Google “favorite places” sticker program that encourages local brick and mortar business owners to claim their local business center listing and to add information to their PlacePages.

Google has annouced that it will initially select 100k of the most sought out and researched businesses on Google (~1% of the 28 Million U.S. businesses) and send them window sticker which passerby’s can scan with a special reader on their phone in order to read and submit reviews as well as receive special coupons.

I am not so sure this one will be a big driver… is a Google favorite places sticker likely to do much for retailers? I am not so sure… I suspect that only the geekiest of the geeks will be standing on the sidewalk whipping out a cellphone to snap a picture of that QR code and stand there flipping through reviews. By the time you’re standing by the front window of an establishment, I suspect your five senses and a guage of how tired your legs are or how bad traffic is will guide your decision. I am just not sure that standing directly outside the store is the time when folks are going to do 10 minutes of web research on their decision on whether to enter or not.

Just give consumers the information they want and figure the rest out later.

The mobile local discovery market is quickly filling up with applications that allow you to do either free form searches or to navigate through a directory to discover what is around you. But searching and navigating is work, especially on many small and clunky mobile devices. So Google also announced a new feature which will appear on the Google mobile hopepage where you can do away with all that messy searching and navigating and simply ask to see everything nearby me now.

 This is a great feature, sometimes people just don’t know what they want, so why force them to make decisions via search of directories? In the real world there is plenty of serendiptious discovery happening as you walk or drive down the street… why should the experience on your mobile device be any different.

Give people more, more, more stuff to search for… like real time local store inventory information

 Knowing that there is a nearby WalMart is nice. But knowing that the nearby WalMart has your favorite flavor of Ben & Jerry’s Ice Cream in stock for $2.99 a pint is nicer. Allowing searchers to go deeper than general store level details with their search is a no brainer. You don’t buy a WalMart… you buy the stuff in WalMart. But this is not anything new, there are a lot of folks that have gone down this path, the key will be how many inventory systems Google can get into and how well it can infer what I want. Google has shown that it can kick ass in the world of cachable, contextual web pages, text and hyperlinks, but the jury is still out in terms of organizing and sorting through large volumes of non contextual, stand alone raw data sitting by itself in an inventory system somewhere.

I suspect it will be a while before product level mobile searches are commonplace, either that or life for the local WalMart marketing manager is about to get a hell of a lot more complicated!

Give people new easy ways to search: If you can see it you can search it.

 Is typing or speaking just too much work? Well Google Goggles, was just announced for Andoid phones, and uses the phones camera to do the work for you. Simply snap a photo of something and Google will try to figure out what it is and return relevant information about it. For small businesses, you don’t even need to take a picture, just hold up the phone camera to capture imagery of the outside of the business, and Google provides a link to that retailers Placepage.

This by far the most gee whiz announcement from a technological perspective, and like the favorite places sticker program seems very much designed to generate excitement among the small brick and mortar retailer community to get them engaged with their Google PlacePages and with Google in general. It certainly seems like nifty technology and if it ever comes available for the iPhone I’d certainly give it a whirl. But as with the sticker program, I wonder how much use Google Goggles will see in the real world. So far most of the visual and augmented reality I’ve seen so far is pretty limited in their practical applications, or at least don’t provide much advantage over the readily available alternatives. But without being able to play with it hands on, its hard to say how impactful it will be, but let’s just say I am a bit skeptical.

But don’t fear, where one company is exiting, there is always another entering.  And the latest hot entrant in the area of local product search seems to be a company called milo.com. The company just announced that it had raised $4 million in a substantially over subscribed round which included some big name VC firms and private investors.

So what is all the fuss about? Well, in case you weren’t aware e-commerce was so late 90’s… after years and years of hype e-commerce still only represents a tiny percent of total retail commerce  (5%), and you guessed it people are still frequently walking around on sidewalks and into stores and handing a sales clerk their credit cards and cash and walking out the door with a product resting happily in the bottom of a bag.

So the relatively new found focus seems to have shifted to how people pick the stores they visit and the products they purchase… and as you might expect looking around for information on the web is a big part of that decision making process.  

Research on the web, but buy offline in a nearby store.

So there have been a number of site that have popped up over the years to attempt this: Yokel, Slifter, NearbyNow, ShopLocal, Krillion and even Google’s own Froogle.  But none seem to have shown much success. But with the market for ‘research online, buy offline’ estimated to reach $1 trillion by 2011 its not surprising that folks will just keep on trying to crack the code. 

So why would anyone think that maybe Milo will succeed where the others have failed?  Well I’ll take a stab at it… looking around at some of the competitive offering there doesn’t seem to be anything special about the real time inventory availability aspect… many sites have that and in fact in some categories like sporting goods, the list of participating retailers is stronger with competitors.  And doing an identical litmus tests across sites with a search for a Playstation 3 in Manhattan’s zip code 10024, Milo gave me the equally crappy result that I should swim across the Hudson river to find this easy to find item in Paterson New Jersey, even though its available about 50 yards from my apartment across the street at the GameStop I can see out of my living room window.   Evidently they don’t have a deal with GameStop yet.

However, Milo seems to have gone more broad (and long tail) than the others with retailers like Target, Walgreens and Barnes and Noble, the latter of which alone may have as many as 100k book titles in stock at any given store… Milo claims to have nearly 1.5 million products available, and knowing how many SKUs are carried by folks like WalMart and Barnes and Noble, I don’t doubt it. 

I think the unique part of the Milo approach is on the inbound marketing side, acquiring customers… in other words SEO, SEO and more SEO. When people research items on the web, where do you think they go? Well to Google of course. So rather than fighting this by trying to create a competitive search destination site/tool that you’d only use when search for products locally, Milo takes the logical step which is to try like heck to be sure that their product pages come up high in the search results from product searches that include a local intent (like Garmin 10019). Very similar to what Local.com is trying to do with local business listings, although as far as I can tell without the SEM arbitrage part. 

In fact a search for site:milo.com on Google reveals that there are a whopping 319,000 pages on the site in their index!  Just for some perspective the Wall Street Journal online has 204k pages, and CNN.com 415k. Now a similar search for NearbyNow reveals that they have close to ½ million pages… but a quick glance reveals a major difference… NearbyNow pages are generally a lot of the same templates with just  new location (like Orlando, Fl) swapped in for each page.  Milo pages are unique, text and keyword heavy product specific pages…  and there are over 300k of them! So if all those pages are product pages then that’s 20% of their available product inventory having its own dedicated web page…  and if you believe the good old 80/20 rule applies here, presumably those 20% are the items responsible for 80% of the searches.

After doing a number of Google searches with product names and zip codes, I was pretty impressed with how frequently a Milo product pages came up in the listings.

• A product search with local intent: you often get milo listings (Search: Garmin 10024)
• A product search without local intent: no sign of milo listings (Search: Garmin)

So the company has obviously figured out a way to do local product SEO quite well… with all those unique product pages a clear differentiator… and perhaps some behind the scenes link juice with major e-tailers. The company seems to do a fair amount of web crawling, likely of their retail partner sites, which may also help them to systematize the process of quickly adding new products to their index by grabbing marketing elements that live outside of the inventory system.

Certainly an interesting company to keep an eye on, and in particular  to see how they develop the revenue side of the business, now that they seem to have the marketing side well on its way.

Tech Crunch had one of the best summaries of why it makes sense and the nice fit between the two companies… the article is definitely worth a read. My two cent summary from that article and others on the deal:

• Google has talked over and over again on the potential they see for mobile in general as well as location aware mobile, and this is still early times so buying the leader now for $750 million was probably better than waiting another couple of years and paying closer to the $3.1B they paid for Doubleclick
• AdMob is somewhat reminescent of an early doubleclick, back when they still ran an ad network business, although without the customer concentration issues Doubleclick had with Alta Vista back then
• AdMob was supposedly doing ~$100 million in gross revenue with a 60/40 rev share with publishers/app developers, and already growing at a fast clip, with some particular success in applications
• the aquisition of Doubleclick didn’t help Google much in the mobile world
• AdMob had a strong presence among the long tail of the small publisher and app developer community… sound familiar? The mobile long tail of app developers and wap site developers is quite similar to the long tail of web pubishers that has been the goldmine for Google
• Apple Iphone/itouch devices made up nearly half of ad requests, so its a great way to profit from the popularity of that platform… and expect that Google will make this a key component of the offering available to Android application developers as well, creating a nice premium display smartphone mobile ad network for advertisers.

So, on the surface there is not much related to location awareness going on here… just a large mobile ad network that has great potential to be a leader on sophisticated high end smartphones given their current success on iPhones and soon on Andriod phones. But these same devices are the ones that will be leaders in bringing new location aware consumer services to market and you have to think that location relevance targeting will be on top of the development list.  The folks from IPG seem to agree… from their Future of media blog:

“Google does location very well. AdMob’s one “weakness” is a lack of hyper-location targeting. Google’s problem with hyper-location targeting has been one of scale – only a handful of apps were being powered by its in-app solution. With several billion ad requests a month, AdMob will add the scale to allow Google to roll out a location targeting mobile ad solution. It’s highly likely that this will be one of the first major things to come of this acquisition.”

I’ll be honest I don’t know very much about search engines, a few years ago I printed out a copy of Larry and Serge’s Stanford paper on The Anatomy of a Large-Scale Hypertextual Web Search Engine and read it on a cross country flight for a job interview on a related subject… needless to say I didn’t get the job and I pretty much dropped any more digging on the subject.

But after attending last weeks MetaPlaces conference and listening to the various speakers, including one on the opportunities in mobile search and also a refresher on Sense Networks, the subject is back top of mind again.

The consensus at the conference seemed to be that the existing web search leaders will continue to be the leaders in mobile and there may only be some crumbs of vertical specialty mobile search left up for grabs for new upstarts. And the existing leaders certainly do seem to be innovative as heck in term of trying to infer local intent… I think if you go give Bing Mobile or Google Mobile a trial run for a local business related search, you may be pleasantly surprised.

Here is a quick chart with some of what I think are particularly significant differences between pointing an algorithm at a stack of static text web pages versus the more fluid movement happening between geographic spaces:

As always, comments and discussion are welcomed!

Before getting into the details of this, a couple of interesting stats from the MetaPlaces conference that I picked up last week as background:
• There are currently 21 million small and medium sized local businesses spending an average of $4k per year on yellow page advertising (~$25B ) and other local advertising
• 75% of Internet user have used the internet in the past three months to look up a place on a map or to get driving directions

In other words:
1. there is a lot of money being spent locally, by local business to get customers to patronize their businesses
2. customers are increasingly consulting with internet base maps to find their way around and to find new businesses to patronize

So it’s just a matter of time before those local business owners begin to direct some of that $4k per year to the providers of internet and mobile map services.

For users of Google map, the Place Pages make it easier for you to find some of the important information about places in a single, simple webpage without having to interrupt your experience and getting side tracked on a mini research expedition to determine if the place is worth a visit. Place pages include an ‘entire page of rich details, like photos, videos, a Street View preview, nearby transit, reviews and related websites.’ According to Mike Sterling, Google is running a separate algorithm that ranks content providers like review sites within the pages.

While certainly a benefit for consumers and users of Google Maps, you can’t help but note that this will be a very compelling tool to help Google in its efforts to sell in advertising to local brick and mortar businesses which may not have had a meaningful website or online presence before now.

Businesses that have a big, or solely, e-commerce business have for years been big users of search engine marketing , but the e-commerce market is still just a tiny fraction of the total commerce market, and Google’s CEO Eric Schmidt famously noted in Davos last year how a truly mobile Web, offering a new generation of location-based advertising, is set to unleash a “huge revolution.” Google Place Pages are just the beginning of this revolution…. the battle on Yodle and Yelp Hill.

Google Place Pages pull brick and mortar businesses online whether they like it or not. They now all have a web page at http://maps.google.com/places/us/city/street/ZIP/-business-name which in many cases may be quite simple information, but all the better reason to engage with Google to claim that listing and potentially turn into an AdWords client.

A quiet, but nonetheless impressive and potentially game changing development.

I dug around to get the latest and greatest, and it looks like in the good old fashioned web world search is about 45% of ad revenue today with premium display close to a 1/3. So Kelsey’s predictions for the mobile world certainly seem to magnify the trend we’ve seen so far on the web with search having an even bigger role and display, a decidedly smaller role. To some degree this makes a lot of sense, since the small footprint of the mobile handset doesn’t leave a lot of room for all that lovely creative ad work, the sight sound and motion and all, and search is simple, quick and to the point which is well suited to the phone.

An even more interesting part, was not just the AMOUNT of search revenue, but the TYPE of searches expected… citing data from Google that mobile searches are 2-3x more often to be local in nature than searches done via a desktop, The Kelsey Group calls for over 1/3 of mobile searches to be “local” in nature in five years, and for over half of that whopping $2.27B in search revenue to be generated from “local” search queries.

Kinda makes you wonder what kind of local stuff all those folks will be searching for on their phones and who will be the benficiaries of that cool $1B+ in local oriented search ad spending.