Startup company NavDog has a solution in beta mode that uses an AJAX-powered mapping system and location intelligence based on its Geographic Business Intelligence Engine (GBIE) technology to support better results for local search. The intent is to provide more relevant advertising to users of local search. The GBIE uses expert system rules and data mining models and is integrated into NavDog's ad delivery system. The impact of this type of technology may be to help retailers better reach local clientele. For NavDog, "everything is local." Shawn Fredrickson, president and founder of NavDog, answered Editor in Chief Joe Francica's questions.

Joe Francica (JF): Can you explain more about the logic and algorithm behind the Geographic Business Intelligence Engine? You say the process is built on continued user interaction; please tell us more about this.

Shawn Fredrickson (SF): We are really looking at a front-end system and back-end system that work together. The GBIE is a server side (back-end) application that receives information sent via AJAX from various user interactions that are recorded by the (front-side) application. On the front-end we record interactions deemed as important by making elements of our interface smart; those elements can be the map control itself, the map icons and popup, the search control, list control, etc. We could make any relevant button push or control action on the site smart, as well. So the front-end collects the data, and we are taking a similar approach to that of a researcher; we are collecting and storing everything.

On the back-end, we are processing select pieces of these stored data using a system of scoring types of actions performed on types of content. For example, entering a keyword into the search control will produce a score for [search] performed against [keyword value], then further interaction with the retrieved map content will produce an additional score for [interaction] against related [map item]. Then the system will choose from the highest scores and crosswalk the database for broader [interests] that the high scoring [keyword value] and corresponding [map items] fall into in order to produce additional content related by those broader interest categories. All of this is maintained within the context of your [geographic position] and your [proximity] to these high scoring items.

In regard to the map: the system is always watching where you are going, and what level of zoom you are on, and how far that is from your starting point or home location, because your actions define what type of searcher you are. We have defined several types of interaction patterns that help us predict and score content that might be relevant to your activity. A good example of this would be starting in San Diego and moving the map to Las Vegas. We know you are traveling and we know where your destination is, therefore we can place hotels, casinos, and restaurants, but NOT something that would be more beneficial in a local search such as the local post office. With that said, any available map content is always within a user's reach and you can always include post offices in the mix if you choose. No aspect of your interaction goes unnoticed or unrecorded and just about everything you do will have ramifications to content, even if they are very subtle.

Please understand that even though the system is fully functional, we are still in beta and a few portions of the GBIE are not enabled for anonymous users. The reason for this is that the GBIE uses a combination of expert system rules and data mining models. Both of these require some "seeding" to be truly useful. We have a set of testers who are not only testing the system for usability and quality, but more importantly are actually generating "training data" for the GBIE. Over the coming months you are going to see the intelligence of the system grow by leaps and bounds.

JF: Can you provide more details about how location-based advertising will be better placed, and how the ads will be served - presumably through Web portals of retailers, hotel chains, etc.? Do you anticipate serving ads to mobile platforms?

SF: If you can understand the explanation above, then the wheels may be turning already. Our goal is to live at the intersection of who, what, where and when. This can be accomplished right now and there are many services, i.e. companies, that are in possession of these data and don't really even know their value. I could name a sparse few here, and they would start with travel websites, ticketing and event websites, and a host of others. Advertising is becoming less about the masses and more about the individual and even more so about the individual in the moment; it's having a system that understands the moment and is ready to complement that moment with the perfect offer or suggestion, which is paid placement of course.

As for the serving mechanisms, they could be websites, maps or embedded into follow-on emails or the confirmation email after a purchase, and they could come in the form of traditional banners, coupons or media. Something like this is made for mobile as long as the user has a GPS-enabled smart phone. Our years spent in mapping and the expertise we have developed in geographic solutions make location aware mobile a slam dunk.

JF: How does the GBIE support a retailer's own trade area boundaries so that you are matching ads with a prescribed geographic territory? (I ask this because I just had a discussion with a company looking to do similar ad searches.)

SF: I think you've got the right idea. We are a bit more fluid regarding our definition of "territory" but we will be leveraging data like you mentioned to better target users, businesses and points of interest. In addition, we'll be combining location with other factors to ensure the advertisement content adds value to the user experience instead of distraction.

Retail is a very interesting area for us because most of them still have a significant "brick and mortar" presence. Online advertising has always delivered the most value to companies that had a significant online presence. There has not been a good value proposition for a locally based brick and mortar business to advertise online. In the NavDog .com experience, everything is local. Businesses that have one or more physical locations have the advantage in our system.

JF: Can you provide more details about the content aggregator that will be used to support location-based advertising? Will this be a pay-to-play model that is weighted to whomever pays the highest amount for the ad, or will it be more dependent on proximity?

SF: The aggregator is the GBIE that has been integrated into an ad network's delivery system. We have several different levels of integration that we are pursuing with ad networks. Ad networks already have some kind of "rules engine" that makes a decision on what ad to deliver. The ad networks can decide how they want to plug the GBIE into their "rules engine." The GBIE provides them with relevance scores for those ads based on the user's interests, motivations, location, etc. These ad networks already have their own weighting system and we're not going to try to change that. It's up to the ad networks to decide the weight they want to give to the GBIE suggestions. We think this makes it easier for ad networks to integrate NavDog into their business model.

For us, on the NavDog.com mapping site, we'll be weighting ads the same as all other items on the map. The difference will be that paid advertisements will have additional features enabled that provide a more effective way of capturing a user's attention.

JF: What other applications do you see down the road for NavDog, and what are the potential business opportunities you intend to pursue?

SF: Over the last year we have talked with several senior level business development and product development executives with some major Internet players. We have had at least one acquisition offer from a small ad network and have gotten advice from some top Silicon Valley CEOs, as well as major investment firms. We have used these dialogues to help us define our mission and the insight we have gained has led us to this point.
Currently we are looking for partners in the ad space; we will need an established ad network to help us develop this technology into a product that best serves the future needs of advertisers. We are not interested in re-inventing the wheel, so working with an ad network that already has publishers and advertisers will enable us to get to market quickly. In the meantime we are pursuing content partners for NavDog.com in order to add further content to the current mapping experience there.


More about this author...