Announcing Bing Maps Geospatial Analytics Platform Preview for Enterprise Business | BRK1074
Okay. Good. Morning everyone. It's. Really curious about the timing, today on our last day afternoon. Lunchtime but. Then I thought, all. Those. Who are turning up are really interested in the topic so, big thank you for making to the session, and. Welcome, to the session. My. Name is Sasha Gauri I'm a senior product manager at Bing Maps. Today. I'll be introducing, our, new geospatial. Analytics, platform. Preview. I'll. Get into little details about our motivations, in building this product. How. It can help your business scenarios, and then. Finally get into details of our private. Preview. So. As you may know Bing, Maps offers. A rich set of location. API, number. Of our customers use it and I was, pleasantly surprised recently, when a customer's, customer. Started asking, us to increase the usage. Limits of this API for. Example most of them wanted us to, increase. The number of local, search results by almost tenfold. So. When I started talking to them we realized they, were getting into analyzing, locations. And using. That data to make decisions like, business. Expansion, for example what's the best place to open a new store and. Couple. Of cases. Like this we started to think if it made sense for Bing Maps to actually offer a geospatial. Analytics. Platform. As, a product. So. What. Exactly I mean by geospatial, analytics it's, how you process. Measure. Analyze. Basically. Slice and dice various data and, visualize. Location. Data. To. Derive business, insights. Number. Of businesses. From retail, to, insurance. Have. Successfully, used analytics. It's. Also called location analytics. To. Get various business insights for example, retail businesses, use it to find new locations, to expand their. Our new. Locations, to introduce. New products. Insurance. Companies, use. Location, analytics, to, measure. Their risk, exposure to. Various natural hazards. For example, and. Digital. Marketers, use this to do. A better profiling, of their customers, and come up with good digital. Marketing campaigns. In. Fact, we. Surveyed a number of businesses, trying to understand, how they see, geospatial. Analytics and. 85%. Of the c-level executives, actually said, location. Insights are critical, for their business, decision, making process. Interestingly. In, the same survey. At. Five percent of the analysts, said, they did not have the right. Tools. To effectively, apply. Geo analytics in, fact. Some. Of them actually said. They. Read about all this. Geospatial analytics, or location analytics in various business reports, but, they have never themselves, really seen such insights or how it can help business, so. That got us to thinking whether. There are any challenges, people face with, existing products or services and. The. Issues we found we, could categorize. In three buckets, it's. Really hard to find. The right data to get your analytics. Going. Most. Of the existing, products, or services have, some kind of a repetitive process and even. After you complete them the insights are kind of incomplete. And. We. Also noticed, most, of them do not really, liberate the latest technologies or latest developments, in spatial, and YAG. So. I want to get into some more details about what these issues were. Starting. With acquiring data. Because. Challenges you don't have a single place where you have all the required data, just. To give an example demographic. Data, like population age groups etcetera, are very, frequently, used but. Just the demographics, table in US, Census is about 1,200. Seven columns so just imagine the time. One spent in going. Through all this data and finding. The ones that's good. For you our scenarios. Next. Even if you have all the, required data. Most. Of it is at a census, track or block support levels but you may actually want to analyze. Something. Related, to your business regions, like for example let's say I want to look at sales in greater Seattle area or. In. The Midwest u.s., so. How can I go about aggregating. All this, statistics. At regions, that are of interest to my business that's. Still a challenge. For. Example, let's, say you want to study how weather can affect your retail, sales, what. You would typically do is get. Weather. Data from somebody. Like US government, or. You. Also want to pull. The sales data from your own databases, and then most. Likely engage a developer, to stitch, that data and make, it available for the analysts. Finally. When. You are doing. Location. Specific insights we. Notice there, are a lot of sources which have either, some, really stale all the data for. Example, businesses. May have permanently, closed so. Are they, misrepresent. Businesses, like some of the businesses, may be run from home and. Cases. Like this can really skew your insights.
Especially. When you are doing things like calculating. The business density, you are trying to understand, what, kind of suppliers, are there at a particular location for, a given product or service. Next. Challenge was. Looking. Into existing, products, or, services. Most. Of them have a paradigm, where you import, data onto, a Maps layer and, then. Zoom, into that data get. Some insight and then repeat that across for example let's say I'm looking at all. Of us so. It's a very repetitive process, where I have to do. This on, a per location basis. Hardly, any tools to do it at a country or state level. Most. Likely this is coming, from limitations. In spatial. Analysis, or such capabilities, for. Example, if I want to do a most. Products, offer a search in a you, know your specify a point and then a circular, radius you, want to search within that, if. We want to make it a little complex, like add. More parameters, I want to find businesses, which. Are of a particular. Type. Which. Are running in a certain location let's. Say I want to let's. Have a home-improvement. Store, and I want to find somebody who, can complement, my business. It's. Really time consuming in, existing, products to do these kind of search operations, and, then. Another. Limitation is in terms of scale. Like you, have millions of address, data either your customer, or sales, regions. Today. You have to batch them up in, sizes. Like I don't five throws and also and then, you repeat that so all of this adds to your time. One. More thing we noticed is. You. Know especially in retail. A lot. Of customers, start their. Product. Or. Buying, journey online. Like if I want a product buy a product I would, start online either being on or on Google and either. On Bing or on Google and, a. Lot of times we see that the, number of searches that happen for a particular product or service have. A direct, correlation with the demand for that in, that region but hardly. Any tools to leverage, this kind of insight when. Making decisions a. New. Technology, or new development, is the concept. Of user movement, a lot of mobile or communication, companies are now sharing anonymized. Data about, how users move in, different parts of a city, businesses. Can actually leverage this data to, get ideas about what's. The best locations, to promote their business or, try. To understand, the user habits, but. Again, hardly. Any way. To easily interpret, or analyze such, conduct, the. Last challenge is about. Special. And high technologies, for. Example I talked about. How. Data acquisition, and processing is hard in terms of locations. One. Common problem we see is when acquiring data people for. Example let's say I have a business and you acquire customer advice addresses. And they, make some. Input errors like the addresses either incomplete, or. Misspell. So. How do you recover as of today most likely you would end up throwing the way such data, but. Can we leverage some, AI or m/l techniques to recover. From, such user errors and still make the best of your data. Another. Interesting, trend is digital. Transformation, where. Businesses. Are investing in. Disguising. Their legacy. Content like scanning, or, legacy, documents. In. Certain businesses these documents, can have number. Of location. In sites like locations. You investigated. For a particular, investment, your, sales records, things like that, how. Can you artificial. How can you automatically, infer, data. From these documents and also build insights. We. Talked about how it's, a, long time, consuming, process to get insights, so. We're going to see how AI, can able can, enable getting. Faster, insights. For. Example. Talked. About user. Movement, data. If. I am now automatically, able to infer things like what our users home, and work locations, can i leverage that to see, or. To, see a better, profile to build a better profile of that customers. Businesses. Have tons of data can we leverage big, data or AI techniques, to build ml. Models based, on that data that can be used for either, sales, or demand forecasting. And. Also, as businesses, keep investing, on acquiring data how can we channel that data to fine, tune your. Models. One, good example is. Inferring. The right type of customers, for your like what kind of demographics. What age group or income, levels where do they live and how do the how. Do these factors, influence sales. Of your products. There. Was a automatic, way to infer. This, kind of information, that can, definitely have been faster, insights. Finally. Talking. About some new developments, in the industry like satellite, imagery, was really. Costly before it's. Now becoming affordable, and this. Is another great, way where, AI can actually help you get insights.
There. Was a very popular research. Article about how somebody used. Satellite. Imagery, of parking. Lots and used, that to infer number. Of customers visiting a retail. Store. While. It was a research article, before. We. Think yeah I can actually bring. That kind of technology, and power to all, businesses. So. That's where we, want to come, up with Bing Maps G Olympics we. Want to leverage our strengths, beam strengths in spatial. And AI technology. We're. Getting, data is hard. With. The, Bing Maps Journal it'll provide. Great. Ways to normalize. And merge. Different datasets in a easy way without having to port we. Have complex. Geospatial. Join. That. Can work at country, region, or various. Admin levels to, merge, data and. Then. Have. Some, really ready to use datasets that are very specific for your business scenarios. To. Speed up time, to insight Bing. Maps to analytics will. Offer. Rich, set of analytics, capabilities. This. Will help you, build a complete profile, of your, customers. And also your locations, and. You. Will be able to search and compare across. Different parameters, I will I have few demos showing that. And. Lastly. Leveraging. AI, in spatial so, we want the Bing, Maps geo analytics will offer inbuilt. Ml, model that can help with both. Sales and demand forecasting, not, only that we. Also want to enable businesses. To bring their own models, and run, on the data that. We, provide or a combination, of their enterprise, and public. Data. So. Let. Me show a few examples, of how this is going to work. Most. Of the, things. I'm going to talk are probably relevant or real retail. But, as such we. Do want the product to support all other, businesses, as well. So. Let's get started a few scenarios, let's. Say you are a retail store and you are trying to understand how weather, or air quality can impact, your retail sales, typically. You would, get. The, air quality data from. Climate. Data online it's a long indata sauce and. Compare. That with your regional sales and see if there's any correlation. To. Do that go. To the, analytics. Store get. Into the data manager, we have options, to pick data from different. Sources in this case I'm going to pick from a local. File.
I've. Already pre-loaded. It, so. You get a brief, overview of what kind of data is in that file what, was, the schema looking like but. Each of the data sets have the ability to transform, and in in this case since. I want to join with another data, set I bring up the, geospatial. Join capability. That. And as you can see on. The Left I have my custom, enterprise. Data which is coming from my own business Lucianne. Pull. Up and pulling up the store, sales, data and, one, of the columns of interest is store, location, on the, right you see is the data from. Climate. Gov, and. I'm, interested, in the weather station, rota so. The thing to call it here is. We. Are automatically, infer the schema of the file you provide, and we, think that that. For, example here the weather. Station, data is, move, the most likely candidate for you to join with store. Locations, you. Can specify it Thresh. The idea, is it's. Going to join all records, where. A store location is within 10 miles or 10 meters whatever you provide from. The weather station. So. That now you have your store locations, and then also the air quality. Provided. As a clawback. Next. Example is. So. Lot, of operations, when you are working on analytics, are related to search for. Example you, want to find where. Your competitors are located, or where businesses. That can compliment you they are located. Bing. Maps geo analytics will offer four different ways of searching in the current case I am showing. Let's. Say I want to visualize all my locations, on map and I also want to find certain types of businesses so I can bring those locations to the map, layer and then. So. Whatever I am seeing right now is my viewport or viewing region and. I, can search for a particular type of business let's say home improvement, within. That region you. Can. Also do a, national. Circle, radius, search where I specify, a point and a radius. Within which I want to find all the businesses. Interestingly. We can also we. Also support, a polygon search where you can either draw a free-form, polygon, close, a shape on the, map surface, and search within that yeah.
You. Can actually search for both business, types and also business names and. In. Some cases. Let's. Say, there's. A. Gas. Utility, company and they are trying to find what, are the best locations, to open, a new gas station along the highway so. You have this notion of a search along the path where user can draw, a freeform path, on the map surface, or import a path, and then, search along, that path again here, also you can search for like. Find, all businesses, along that. Path. Or you, can specify a particular type like restaurants. Or a particular, business name. Next. Year scenario, I want to talk about is a, lot, of cases you, want. To find which businesses, can actually supplement or, you also want to find, what. Are the cases where a computer, is in, a within, a certain threshold from your location so. This is. What we call as a collocation. Search where you can specify on, the, left let's say I'm fourth coffee and, I. Want, to find a car. Let's. Say I want all. Instances. Where there's a bank within with a particular, distance, and. I can also specify, where. I want to do this search for example in at a country level or in this case I'm doing at Washington, so. Run that and I. See that there are four look for cases where, there's. A bank pretty close to my. One. Example of how you could use this is like let's say you are a home-improvement store, and you want to find all cases where electronics. Let's, say you want to have a scenario, where, people. Buy electronics, from the, other store and get, home installation services, from yours so, you want to find out, all. Locations, where these two businesses occur. Within a certain distance. You. Can specify up to five different, criteria, for the search and you can also search at. Across. The country state and zipcode levels. A. Lot, of times in the analytics, you. Want to do some ad hoc measurements. Like distance between places, or calculate. Stats. Like area. Or volume, so. In this case we have the, ability to just, randomly, point or draw, on the map various. Shapes and calculate, their length. Parameters. Or area. You. Can also do this. Closest. Path analysis, where you, can plot a number of points, again random on the map layer and then, see what is the shortest path connecting, these to these, locations a. Variation. Of this is where you want to have regions across the, country. And you want to find the closest, road network, for these points. So. Isochrones. Our catchment. Areas are one, really. You know frequently. Used tool. Catchment. Area is basically the boundary. Of all the, customers. Who, visit your store and. For. Retail stores you can actually define that with a, driving. Distance or a driving, time. For. Example here I have chosen show, me all what's. The boundary of all the locations that can be reached within 30 minutes of my store. Location, so, centered at the store, there's, a boundary of all the regions that can be reached. I. Can. Actually, repeat. That to get insights.
Like What, are the regions that are with, across, different. Traveling. Distances, or traveling times. An. Interesting, thing I can now do is. Pull. Demographic. Data and, see. Within the catchment, area of my store how does a demographic. Or other data, look like in this case I am pulling up, demographics. Data and I get insights about, what. Percentage of my users are within fifty. Plus ten, minute, driving distance from my store or and, I, can see that it's like 99 percent of my customers. Are within a 30 minute driving time. One. More analysis like I was saying you can actually search with any custom, polygon, shape so here I want, to see. Where. My competitors, are within my store catchment, area so. I can pull, bring. Up the search and look for let's F worth coffee is a computer, I want to I can search for that location I find. That not. Only that I can, actually bring the same demographics, profile, or data again and analyze. How, my. Store. Population. Or my, customers, compare, with a competitor's, or catchment, area, for. Example. One. Case where, you could use this kind of insight is let's say you are trying to evaluate whether. You want to close, one of your stores or whether it make to, open. A new store or, how, does the computer, affect the sales at your own location. Comparing. Having the ability to do a side-by-side comparison can. Really speed up the process. Next. Up I, talked. About how online search, volumes, or. Search. Habits are, a strong, signal for assessing, the demand for a particular product or service so. Let's see how we can do that, so. Here. I'm. Interested in seeing the demand for let's say I'm a home-improvement store and I'm interested in seeing how. Many people are searching for furniture, in, a region I'm interested, in and how. Many people are actually looking for my store, name if I mind I'm. Yeah. Let's. Say if I'm doing home services or furniture, how, many people are actually, searching. For these products. Online I can. Then correlate, that with, the, number of times my own business, is actually coming up online, so. Here you can see that there's a good demand for. Furniture. Or home services but, my. Own impressions. Which, is a bottom graph the. Number of times my, business came up on, search, results, is kind of low so maybe it's indication, for me to run a online. Advertising, or coupon, kind of campaign. One. Common challenge, businesses. Faced in. This kind of scenario is you, would run a online. Campaign, but. Then you run into problems like you don't know whether it was successful or not like, you run ad advertising. Excuse. Me you ran a online, ad but. You are not sure how many people actually saw that and came to your store. So. For. That what we want to do is have, the ability to import user, location, movement data this. Is part of the standard product. Offering you can leverage this data to, get insights like. One. Quick thing before that the. Bing, Maps analytics would give. You rich capabilities. Like. Parts. Through all the fields in your data and pick. Which fields you want to visualize how so. In this case I pick the, number. Of. So. In this case I'm picking my. Data from. Number. Of users visiting my store without advertising and. With advertising, so now if I plot that I have, a strong correlation between how. Many people actually search for my product how, many times did I come up on search results, and then, how, many people actually came, to my store, after. Seeing those ad campaigns. From. The graph it's clear that whenever, I run a campaign, I can see more people, actually. Coming to my straw. That's. When. Increasing. Trend in retail, is. Retailers. Want to compete online and, a.
Strong Trend is where customers, make. An order for, a particular, product or service online, and then. Pick it up at a physical, store so, buying online and then picking up that store. Now. Let's take a hypothetical, case. Where I am, seeing that a, number, of customers are actually you, know you know one. Of my particular stores, there's. A strong trend of people, doing this they're. Not really visiting, stores to buy there but they have already placed an order and, they. Want to pick up at the store this. Let's say this trend is interesting, to me and I want to see. What. Are the various contributing, factors, so. What I can do is. Pull. Up my sales, data and also, the. User movement, data. From. The user movement, data I can build some. Rich insights like. How. Many people actually visited, my store and what do they actually do after visiting, my store so. For example here let's say I'm a home-improvement store, and I. See that 10% of my users are already also visiting, a bank then. That's maybe a cue for me to have. A partnership, with a bank and have some kind of check processing, or some other similar. Capabilities, within the store and save time for the customers, I. Think. A popular example is how you find coffee. Shops in grocery. Stores. Another. Thing I can do is bring, up all my sales data and. Again. Visualize, that so. Here now I'm saying. For. A particular given location, what. Are the various products, what are their volumes but, I can also visualize, the customer locations, or the order locations, on a map and. You. Can see some insights coming up right there, are a few areas where the, really dense pockets of customers, who, are interested, in actually. Buying online and then, picking up at store so. Then I want to understand. What. Are the various demographic. Profile of these customers, so, I can get into the customer segmentation, and. Geo, Analytics would actually, infer all the various traits of that customer. Base it, could be things like their median age income, levels. What kind of localities. They live at or, their, online habits like what devices. They use for browse browsing. Things. Like that you, have the ability to edit, to. Rule out if, something, don't make sense but more importantly, you. Can actually search. For. Other locations. Which. Have population. Which are very similar to this kind, of profile. So. Do that I find that these are the locations, which have customer. Bases. With. Very similar traits, to that iPhone, so now if I want to let's say grow. My sales of a particular product across, other regions, these. Are the best places to target. So. These. Are some of the interesting scenarios. We want to build. Now. Getting into the details of private preview. We. Are excited about a lot of these features but we also want to make sure these. Are what. Our customers, really want and hence. Our private. Review program, during. The program we, basically want to validate some of our core assumptions, around what, feature sets what models what ml models can really. Help customers. Before. I get into details some, I want to get into some of the benefits partners. Can get by entering the partner, pay-per-view program. Number. One this. Would be genetics, that really works for your business. Bing. Maps team would work very closely with your team. To identify two or three business scenarios, where. Analytics. Can really make a difference. There's. Also a great chance to influence our product, roadmap. Give. Us feedback about what, features data sets or particularly. The ml models can really, help your business scenarios, and. Then. We. Think this is going to be a game changer so, this, could potentially be a competitive, advantage for your own business from. The inlet by. Opting to the program early. We. Do have few asks, from the program like this. Is this program is open for businesses, that have open. Do. Analytics, need. We. Want to be able to identify at least one or two problems that we can help with. During. Our survey we found that analytics. Is most suitable. For detailed. Digital marketing, insurance, and. Finance. Kind of companies but we are also open for other businesses, if, they are interested in the private preview. But. More importantly, we, want them to be able to spend quality time, on. Evaluating. The product and also giving us feedback, so. In. Case you are interested just. Drop, us an email at geo, analytics at microsoft.com. We'll. Start off by setting up NDA. So that any, confidential, info for both parties, are safeguarded. But. Once that's done. The. Bing Maps product team would actually sit with you and again. Find. Out what are the two or three most. Critical problems. For your business and how analytics, can help. We. Have a Minimum. Viable Product, as we call it that has the core functionality, required. To solve those issues, it. May not have do all the bells and whistles but definitely.
A Workable solution. And which. Is tailor-made for your scenarios, so. We want to work, with you to identify your, success. Tirion ik is. The product really helping your scenarios, or not and, then. Our partners can evaluate the product for one or two months or. Extended. As a case may be and provide. Feedback we. Will iterate based on the feedback and in case you are interested you. Can actually license the fully. Built with, all the bells and whistles or, if. Doesn't work, out then you could basically drop out with only without, any obligations. So. That's. It I had for today thank. You very much. Any. Questions. Yeah. If. You can come over here. Do. You have any plans. To integrate this into like, power bi for visualization yeah. That's. Cool thanks, yeah, like I said we, really want to get a good. Understanding of what works for our partners so if power bi is the best way to visualize our reporting, then yeah that's, definitely underlined. Thanks. Anyone. Else. Okay. Thank. You very much. Please. Do take a minute to evaluate, this session and, provide. Us your valuable feedback. This. Session recording, will be available at this link.