#087: Offline Data in an Online World with James Fogelberg

April 24, 2018

If you have a smartphone nearby and you are not wearing a foil hat, chances are that some brand somewhere — and probably several brands in many places — know where you are. Is that creepy? Maybe. It’s likely removing a few taps when you check what the weather will be like tomorrow, and there might just be a coupon for a discounted hamburger just waiting to pop up when you get near your favorite QSR around lunchtime! In this episode, James Fogelberg from Landmarks ID joins the gang to discuss the ins and outs of using the ubiquity of mobile to the advantage of both brands and consumers.

Checking What Your Device Is Tracking

As promised on the show, below are instructions from James for checking what you are allowing your mobile device to track.

iOS:

Settings > Privacy > Location Services > all the apps which have requested to access your location and the permission you gave (e.g., ‘Always’, ‘While In Use’ or ‘Never’).
Scroll to the bottom, and then: System Services > Significant Locations.  Once it loads, it will show where Apple has located you recently

Android:

Settings > Security & Location > Location > Recent Location Requests — this  is where you will see the apps which recently checked your location.
To turn off ‘Google’s Location Services:’
  1. Settings > Security & Location > Location > Mode – select ‘Device Only’ — this mode uses only GPS to determine location (i.e., no Wi-Fi signals, sensors etc.)
  2. Settings > Security & Location > Location – select ‘Turn off Location’ — this will disable location for all mobile applications on your device

Miscellaneous Mentions

Episode Transcript

[music]

00:04 Speaker 1: Welcome to the Digital Analytics Power Hour. Tim, Michael, Moe, and the occasional guest, discussing digital analytics issues of the day. Find them on Facebook at facebook.com/analyticshour. And their website, analyticshour.io. And now, the Digital Analytics Power Hour.

[music]

00:28 Michael Helbling: Hi everyone, welcome to the Digital Analytics Power Hour. This is episode 87. We play in digital, but we live in the physical world, a world where understanding behaviour has always been important, but is now amplified by the ability to track location based on any number of different methods. But what do we get with all this offline tracking? Well, sit back and relax as you’re driving in your car? Or is that you sitting at your desk? Or wait, over there on your balcony? Oh we see you, we see you all. [laughter] Okay, not really. But increasingly, the connection of offline data to online brings some interesting results. One of my co-hosts located in Columbus, Ohio is Tim Wilson. Welcome, Tim.

01:20 Tim Wilson: Hey Michael, let me just reach across the desk and shake your hand.

01:23 MH: That’s… [laughter] And my other co-host, probably sipping a flat white in her white flat in Sydney, Australia, Moe Kiss.

01:36 Moe Kiss: Hey guys, how’s it going?

01:37 MH: Welcome.

[laughter]

01:40 MK: One day, you’ll get used to it. One day.

01:43 MH: What?

01:44 TW: That was…

01:45 MH: I didn’t even miss a beat. There was no thing there at all. This is a normal…

01:52 MK: But actually, yes, I had a delicious flat white this morning.

01:54 MH: Yeah. That’s great. And I am Michael Helbling. [chuckle] But we need to add someone to the conversation who can help us triangulate, if you will, the intersection, the cool versus creepy, in all this location data. So we invited our guest, James Fogelberg. James is the founder of Landmarks ID, a mobile location intelligence company based in Sydney, Australia. Prior to that, he spent many years in the media marketplace and different companies, but most recently as head of innovation at Adshel, where he worked to help understand audience movement location. Welcome to the show, James.

02:36 James Fogelberg: Kia ora, how are you?

02:37 MK: That was a New Zealander welcome, so just to confuse our guests further.

02:42 MH: Dude, are you from New Zealand?

02:44 JF: I am, I’m a kiwi. Hopefully I’m your first kiwi on the show.

02:47 MH: Oh, this is great. No this is really good ’cause most of my references are more New Zealand and not Australian. And I’m afraid to use them because I think I’ll offend Australians. But since you’re from New Zealand, this gonna go great.

[laughter]

03:01 JF: Yeah bring it on.

03:06 MH: There you go. Well, let’s get started with maybe a little bit of background on what you’re doing, the work, and some of the technologies that underlies it, because I think that will help spur the conversation in terms of what’s possible and what is happening in the marketplace, in terms of these location-based data collection capabilities.

03:25 JF: Sure thing. At Landmarks ID, we help brands and marketers learn more about their customers using privacy-compliant mobile location data, powered by the smartphones consumers carry with them at all times. And what we do, so we work with brands and we help them understand where their customers move in the real world, against nominated points of interest. So basically we’ve got two core technology platforms. One is mobile device-centric, so it’s a SDK that gets integrated into a mobile app. And the second is our mobile location intelligence platform, which sounds like a mouthful, but basically is a network of geo-fence points of interest. Much like you might create tags on websites and understand people’s visits to websites, we’ve created a micro geo-fence network of points of interest in the real world. So basically, if a building exists in the real world, we can build a geo-fence around it and then activate that geo-fence on behalf of the brand. What that allows brands to do, is understand again, which of their customers who have provided their consent, as they enter those locations. It abides by both iOS and Android privacy policies and we work at the discretion of what the brand wants to accomplish.

04:39 MK: Do you wanna just give us an example, and maybe let’s use THE ICONIC, right? Because we’re 100% online retailer, we live the dream because all of our data’s online. So what kind of value would we gain by looking at that offline geodata?

04:56 JF: Sure. I think THE ICONIC is actually a really good example because I catch public transport to work. I see THE ICONIC handing out $20 off customer vouchers all over the place. A lot of your shipments probably get shipped to people’s work addresses. So do you know where your customers live? If you look at your advertising and marketing purposes, where are you spending your money at the moment and are you targeting people that are existing customers or actual new customers? If people get their shipments sent to their CBD work address, do you actually know where they live? When you do outdoor media campaigns, for example, or handing out promotional flyers, you could target those in the geographical areas on people’s early morning commutes, I.e, they’re coming from home into the city, and target them appropriately.

05:45 MK: So do you wanna just take a step back and talk a little bit more, ’cause obviously I’ve spoken to James in life a lot about this, but for our listeners it might be interesting if you just explain a little bit more about how geo-fencing works? And what that actually means?

06:01 JF: Sure. Geo-fencing is, basically, creating a point of reference of a point of interest. In the most simplistic sense, a geo-fence is a circle over a real world location. If you think of a sports stadium as a circle, you can create a circle over the sports stadium. It acts like a virtual tripwire, and as and when a consenting customer enters that location, their device informs the brand owner they’re at that location.

06:29 TW: What’s the mechanics of that? I have to know you’re my customer. And then, you said the customer has to kinda opt in, they have to have an app with location services enabled, they have to have… What do they have to do in order for them to be detectable as having crossed that virtual tripwire?

06:47 JF: Sure. So, this kind of gets into the privacy versus non-privacy compliance. There’s a bunch of people who use location data through the mobile bid stream. In programmatic advertising a publisher who’s selling an ad unit puts a piece of advertising up for sale, often attached with a longitude and latitude. And people scrape that data, and infer that person’s in a location. Now, the people who scrape that data don’t have the consent of the customer. Whereas, when you work with things like a mobile application, for example, like a bank’s application… So I have, like everybody, everyone has a relationship with a bank. I download my bank’s application, I turn that application’s location services on, because I want the inherent benefits as a customer, to turn it on. So for example, I wanna find my closest ATM bank branch, when I’m at the international airport. I want the bank to know that I’m flying, passively in the background, so they don’t block my cards from transactions, etcetera. Because I’ve provided my consent to that bank, or the brand in that case, I’m with Landmarks ID, we build a additional value propositions for that bank to learn more about their customers.

07:53 TW: Do they have to have the app open? They just have the app running in the background? How does that… What’s the threshold? And I guess, maybe the broader question is, generally, what sort of a scale? If I’ve got 10,000 active users of my app, can I expect to have 500 that I’ll be able to use, or 9,500 or somewhere in between, or this is totally varied by industry?

08:16 JF: Yeah. All great questions. There’s probably about 10 minutes of answers there. But at a high level, so you’ve got two platforms, Android and iOS. Just to go through the permissions or opt-in process. So, with Android, you’ve got permissions. Now, anybody who’s listening to this podcast can actually go to the Google Play Store, look up an app that’s on your phone, if you’re on Android user, or anyone can do it actually publicly, scroll to the bottom of the page of that app, and click on “Permissions.” So, I don’t know, click say, United Airlines Android application. Click on “Permissions”, and at the bottom of that page, you’ll see all the permissions that you, as an application user, grant United Airlines. [08:58] ____ when you download that application. And with Android, being owned by Google, they’re slightly more open to providing those permissions all the time than say, Apple. So, on the iOS platform run by Apple, location services is specific opt-in, versus with Android, just by downloading an application that’s turned on location services, you by default, granted your permission to Android. So with iOS, some of you may be familiar… Like, take Facebook as the ubiquitous example. You download the Facebook application, and it’ll ask you to enable location services.

09:16 JF: And, you as a consumer say, “Allow” or “Don’t Allow.” If you click, “Allow”, you have granted Facebook the right to access your location. Now, Facebook can choose to always access your location to select “While in Use.” Or if you say, “Don’t Allow”, they put you to near that. So, this is how the privacy compliance piece of location services works. With “Always” that app owner has access to your location, if they so choose, like if they say “Always”. “While In Use” with iOS works in two capacities, either one, when the app’s in the foreground I.e. You’re looking at the application, or whilst the app’s in the background, I.e. You haven’t hard closed the application. And if it pulls your location in the background, the blue bar will start to appear. So, it’ll tell you that… Using the example before, “United Airlines is using your location in the background.” Again, it’s fully transparent, so the consumer can choose to turn their location setting off in their settings if they so choose.

10:34 MK: So, you keep referring to United Airlines. Can you explain a little bit about, how would this be beneficial to a company like that?

10:41 JF: Yeah. Sure. It can be, in a couple of different capacities. They might have a feature within their application that allows you to book flights. So you, instead of you having to enter New York City as your current location to book a flight from, it will automatically by default know you’re in New York City, and try to find you a flight to wherever you’re looking to go. It speeds up the journey for the customer. There’s airlines that we’re speaking to that want to understand where the people are, through security, or through immigration at an airport, so they can start to understand when someone’s likely to miss a flight, and then also, provide advice for walking times through terminals. So for example, if you’re at Heathrow, and you’ve arrived at the terminal, and they know that based on previous data, it takes a 40-minute walk to get to the gate, they can tell you, “Your current walking time today from check-in to gate is 40 minutes.” Or alternatively, if they know that you’re still, I don’t know, stuck in Duty Free, or you haven’t gone through security, and they’re on final call, they can start the process to look for your bag to offload you, ’cause you’re not gonna make it in time.

11:41 MK: Wow.

11:42 TW: While you’re standing in security, they could be reminding you, “Hey, you’re not gonna make it. Give it up, buddy.”

[laughter]

11:47 JF: Or, the idea would be that they could…

11:49 TW: Oh. Now, I get pissed off at that, that family of seven in front of you that is gonna take forever, you’re gonna want to murder them. [laughter]

11:56 JF: Or, for example, they could remind you 40 minutes before that time was to come that you need to leave now to make your flight on time, just like Google Maps services provides notifications of your meeting times…

12:07 TW: Oh wow.

12:07 JF: Saying, “Hey, it’s time for you to leave for your meeting because traffic’s heavy.” Like airlines wanna take off on time, it costs them tens of thousands of dollars at the gate window. So there’s huge business expenses that companies are trying to optimize, and one consumer or costumer who’s late to a plane or whatever has huge reprobations for the business…

12:28 TW: The husband and the wife who are arguing about when they need to leave for the airport, they could just defer to the app.

[laughter]

12:34 TW: It could be a marriage saver. I think, wow…

[laughter]

12:37 TW: That’s, that’s fantastic.

12:38 JF: Or you just do what my fiancee does and you just leave and make sure that she gets herself to the gate on time and if I make it it’s a miracle.

[laughter]

12:48 MK: Do you know, it’s really funny actually, so about a month ago I was at a conference in the US called CXL Live, and we had Garret Angel was speaking about offline data, so Garry’s actually been on the show twice before, and he is a former digital analyst who’s now applying a lot of his learnings to how people navigate through a physical store, and one of the discussions that we had kind of after his session was talking about, with a lot of location data people do sometimes think it’s a bit creepy, and there’s this inherent value for business, but how do you make the customer… How do you actually make it a better experience for the customer so that they want to opt in? And those examples that you’re giving is like the first time that the penny’s actually dropped for me of like, “Oh shit, yeah, this can make my life easier, it’s not jut about it being beneficial ’cause the plane leaves on time, it’s like, you won’t miss your flight or… ” That’s actually something I don’t think someone’s explained to me very well before, so…

13:49 JF: Yeah.

13:50 MK: Yeah.

13:50 JF: There’s another great example in the quick service restaurant space. So Domino’s, pretty innovative in what they do in terms of being a pizza company, but how they’re using technology. They’ve got mobile location technology executions, where they start to cook your pizza when you arrive in proximity to the store, and they know the speed you’re moving at based on whether you’re walking, biking or driving.

14:12 MK: Wow.

14:12 JF: And so they basically have perimeter geo-fences setup around their stores, and as you come in proximity to the store, they cook your pizza so that they can basically guesstimate your optimized arrival time so your pizza’s fresh.

14:24 MK: That’s cool.

14:25 TW: But it’s weird in that all you get asked is, “Do you wanna enable location services?” There’s not a mechanism when you’re getting that prompt to actually tell you this is why you might want to and this is the benefit to you, right?

14:40 JF: Sorry, I probably breezed over that. In the latest versions of iOS the brand owner, I.e, the application owner is through Apple is able to provide a sentence or two as to why they’re trying to turn on location services for “always” or “while in use”, and I can provide some examples of those.

14:57 TW: So do you have a… I know I asked like four questions at once, so do you have a sense, one, I guess between iOS and Android, and now this totally makes sense and I think I knew that, was is different mechanisms for opting in. One, is there inherently one or the other have a higher general opt-in rate for location tracking and then two, what is the generally, and I’m sure it varies by brand, by app, by everything, but do more people… Are we at a point in 2018 where people are like, “Yeah, if you’re asking for location services I’ll give it to ya,” or is it more often to say, “Oh, hell no.”

15:35 JF: Yeah, good questions. So, I guess as a rule of thumb in Android, if the application owner puts location services, and there’s two different accuracies of location services in Android, so there’s approximate and precise, and broadly speaking they kind of speak for themselves, but if they put that into their application, by default 100% of people who download the application have locations services consentual permission using Android’s terminology to Android. In iOS it does differ, but it really comes down to the value proposition of what the brand or the application owner is doing. So 100% if you’re doing something creepy or untoward or whatever, like if you’re Words With Friends and you’re asking for location services to be enabled, there’s no inherent value proposition for the customer, but if you’re a brand and you put a legitimate value proposition for a customer could be that you’re offering special deals or promotions, it could be that you’re providing relevant content and news or weather, it could be that you’re making sure that people get to flights on time, or you optimize the app for what they’re doing, like if it’s a shopping or whatever.

16:43 TW: In both cases, can just mobile app tracking actually include in its reporting, in its tracking whether you’ve got that, like, “You had 10,000 active users, and of those active users these were the ones that had, this percentage had it”?

16:57 JF: Yeah, so when you download an application as a consumer, you share a bunch of data sets to the application owner, okay? There’s some that are mandated by the platforms, so like what location setting consent did the person give, what push notification consent did the person give on iOS, but you also share things like, what’s you mobile advertising ID, what’s your operating system, what telco are you connected with? What software version of your operating system are you on, etcetera. So with that, if you think of a bank that has [17:29] ____ million customers, of those customers not every single one’s gonna download an application, because some people might not be mobile first or whatever, so they’re gonna have then some segment of their customers, let’s just say it’s 7 million of their customers have downloaded the application, and then a subsegment of those have turned on location services. In the banking industry it’s pretty high, it’s north of, depending on the market but it’s normally 80 to 90% of people turn on location services.

17:58 JF: And so they’re able to match that one to one to the customer because you downloaded a mobile application. So I bank with Commonwealth Bank of Australia in Australia and I log into my application with my customer number. So the bank knows my customer number. I log in, they create a safe device, or a trusted device for me, and they link the mobile advertising ID of the device which is a resettable ID. Think of it as a license plate for your phone, so it’s not like your VIN or chassis number that’s stamped into your vehicle, it’s a license plate that you can update or reset. So, your mobile advertising ID then links the customers, the bank would know the customer number and then they now know the mobile ad ID of the consumer. And with that, you’re able to now link different data sets for the brand… Well, the brand does. So at Landmarks ID, we don’t know anything about the personal information of the device, we don’t know who owns it or anything like that, or what their banking relationship is with their customer, how much they earn or anything like that. We just provide data to brands, and then they ingest that data and use it for various business purposes.

19:09 MH: So is it a marketplace for the data itself in that case?

19:13 JF: No, not at all. We’re an enabler for brands. Brands own the data. So if you look at things like GDPR coming in, think of us as sort of like, a bit of an abstract example, but Salesforce helps brands build data sets on their customers from a CRM perspective. They don’t have a direct relationship with any of those brand’s customers. Landmarks ID, we’re an enablement pipeline to build location insights for brands, and then we provide the data to the brand. All those data and insights are 100% owned by the brand, it’s their relationship with their customer. We don’t want anything to do with data ownership or anything to do with any of their own personally identifiable information, it’s a one way data feed. So we feed data into brands.

19:57 MH: But in theory could someone like a bad actor go out and capture this kinda data if they had an app that people were using to then sell off to people? Has that happened? Or is that a risk, I guess, of this space?

20:11 JF: Sure. So there are organizations out there that get access to location data through other means. So when you’re talking about mobile app, you’re talking about GPS-labeled data. And you’ve got the capability to, potentially, at the discretion of the app owner, get access to that location always. Now, it’s at the app owner’s discretion, it’s their relationship with the brand. There are people out there that use mobile advertising, but stream data, to start to build up pictures of where people move. Now, what their motives are is their business’s discretion, but there are people out there that want to learn more about where people go and then next, sell you the data for advertising purposes. Much like online retargeting exists today but they use where you go in the real world to understand your behaviors, purchase intents, and brand [21:03] ____.

21:04 MH: Yeah, or Facebook to understand everything about you always.

[chuckle]

21:09 JF: Yeah. So we see ourselves at Landmarks ID is very much like a marketing tech solution. We work for the brands and it’s what they wanna do in their relationship with the consumer. There’s people out there that scrape things like your messages coming off your phone or that create free wifi networks. Inherently in society there’s always gonna be some form of bad actors regardless of the industry or the purpose. The point for us as our business, is we want to help good brands have great relationships with their customers, create new value propositions for where customers spend the majority of their time, which is offline, and where customers spend the majority of their cash which according to PWCs, 92% of transactions take place offline, yet insights on where people go in their offline behaviors have lagged online.

22:01 JF: We just wanna help brands go along their path, and at the same time I think, brands themselves recognize that they don’t want… Like you mentioned before, Michael, Facebook, brands don’t wanna continue to pass data through Facebook, they wanna learn more their customers and create new value propositions for their brands. And likewise, they have licensed software and technologies and hopefully more so with Landmarks ID on the way, to be part of the process and learn more about their customers.

22:28 MK: I’m gonna take us off a cliff, I love doing this, where I just take us down. We’re having this great discussion and I’m like, “Oh, let’s go over here, there’s something shiny.” Talking about geodata, there’s obviously lots of different mechanisms, like beacons and connecting to wifi networks and all sorts of stuff, and obviously using the mobile device is the most accurate. But are there potential flags for businesses in the fact that, the type of customers, and I know this from my own analysis at THE ICONIC, the type of customers who have our apps are super engaged. They’re our loyal customers, they spend the most, they come back regularly. I’m just thinking about coming at it from an analyst’s perspective, if you’re using this data to understand your customers, but only your most loyal and engaged customer actually have your app, are you potentially gonna have some kind of blind spot on…

23:23 TW: Sample bias basically, right?

23:26 JF: Yeah, I think, and just for everybody’s benefit who’s listening, I’m not a data expert or an analyst or anything like that.

[chuckle]

23:34 JF: I’m from the media world.

23:35 TW: That’s okay, neither are we.

[laughter]

23:40 JF: But Moe, using your example, I think THE ICONIC and the example of people who download the application being super users or whatever, it does depend brand to brand. For example, a sports bidding application is kinda ubiquitous across… A lot of the sports bidding takes place only online or on mobile. Or if you’re a bank, or an airline, or whatever, the people who download your application’s fairly ubiquitous, so it’s use case by use case but yes, you’re right, the brand knows that themselves. You just get an example of, you know that your app owners are your super users, the point is that there’s data sets that are available through mobile that are being untapped that can be used to understand more about your customers. You just need to unlock them.

24:22 MH: Well, there’s two separate aspects. One, you’re saying that you’ve got this data and you can operationalize it to improve or enhance the customer experience, or even then hopefully drive more benefit for the brand just in absolute terms of touch points and offers, and interactions you can have. Then there’s the second, which is okay, using it for getting a deeper understanding of the customer. Just, you need to go in with your eyes wide open, like that’s using it for analysis and having that recognition. So, it seems like they’re really two different uses of the data, and it may be that you’re like, “Hey, we can totally make a great customer experience mostly for our most loyal customers.” But that’s fine, they’re your most loyal customers. And that’s kind of a, if you have enough volume, that totally makes sense, and from the analysis, so I think Moe to your question, it sounds like James, you’re saying, “Yeah, don’t think that you’ve got necessarily a representative sample of your customer base but it’s still potentially valuable.”

25:31 JF: Of course. So, a great example would be if you take a geo-fence over a McDonald’s store. Like, we’re not going to be able to tell you every single customer that turns up to a McDonald’s store. But, we’ll be able to give you a really good insight as to the time of day they turn up, where else they go, how many of them turn up to Burger King, or Hungry Jack’s in Australia. That’s data that is available on your customers that you may not be capturing.

25:57 MH: That was a fantastic on the fly localization of your message. I remember being very confused by Hungry Jack when I was in Australia. I’m like, that logo and the menu looks exactly like Burger King.

26:09 JF: Yeah, the Whopper exists in Australia.

26:11 MH: Yeah.

26:12 JF: And I think the other thing is, like, if you look at traditional market research surveys and panels, like this is data directly matched to your customers, one to one. You’re able to use those insights to inform a bunch of different strategic business opportunities. Say for example, just talking quick service restaurants, we’re talking to a company locally that’s looking at where they should build their next restaurant based on where their customers are. So, that’s a multimillion dollar exercise. They’ve got to buy the land, build the restaurant, find a franchisee, so where should they build their next restaurant?

26:45 JF: They’ve got over several million mobile applications downloaded by their customers, so at an aggregate level, heat mapping where their customers are by time of day and day of week, and analyzing it against restaurant breakfast, lunch, dinner, etcetera, and understanding where to build their next restaurant. Or using the airline example of where their airline customers fly, that’s not gonna be a 100% of those airline customers being located at Buenos Aires Airport, but if you now know that 300,000 of your customers go to Buenos Aires Airport every year and you don’t fly there, that’s substantiated empirical data that you should start looking at putting an airline there, or flying your plane there, sorry.

27:24 MK: I keep being really selfish and thinking about this from my own perspective and how I would manage this. Like, having this data set coming in as one of my points. What about when we talk about stitching? ‘Cause we all know I love talking about user stitching. The context of a mobile IDFA works really well, and I’m really lucky ’cause we collect that using Snowplow Analytics. But I’m just thinking about, because people always, because I talk about user stitching…

27:52 MH: Time out, time out. A mobile, what’s the, give me…

27:56 MK: IDFA, which is the thing that Fogelberg spent like five minutes explaining earlier.

28:00 JF: So, it’s your mobile advertising ID.

28:02 MH: That’s your advertising ID, okay. He said the mobile advertiser.

28:05 MK: Okay, okay, fine.

28:09 JF: That’s for iOS. The Android is AAID?

28:12 MK: Yes.

28:12 JF: A bit lengthy [28:12] ____.

28:14 MH: That’s the one that’s the license plate for your phone?

28:16 MK: Yeah.

28:17 JF: Yeah, I think. So, anybody out there who wants to understand what their mobile advertising ID is, you can download an app called The Identifiers. Which you just go to Google Play, or App store, download The Identifiers, and it will show you your 36-digit character which is your mobile advertising ID. It’s a sequence of numbers and letters that’s unique to your device, however, it’s consistent across every single application.

28:43 MK: Yeah, and you can reset them as well.

28:45 JF: Correct. However, it’s incredibly more robust than a cookie, and every single time you re-download your application. So if you upgrade from an iPhone 6 to 7, and you download the latest version of, let’s just use the example I talked about before, the banking application, you’ve then shared your new mobile advertising ID with that bank application, and then you log into that bank application with your customer number, that bank now knows your latest mobile advertising ID.

29:11 MK: Yeah. So one of the questions that I always get asked is, when it comes to user stitching, how do you stitch users if they’re not logged in? And I’m just really curious from your experience, as you start talking to different brands. And I always say to people, I’m like, “Look, I’m really lucky because we have Snowplow, I can get their mobile IDFA, which as far as I know you can’t store on the Google stack. Also, secondly I’m even more lucky because we make all of our users log in, in order to transact.” So, for me, it’s actually a lot easier to go back and stitch users. But, I’m thinking about the context of like, if you are in a company where people don’t log in like, I don’t know. Can you think of any other ways that people could potentially stitch this data back to online behavior? I’m, yeah, I’m just thinking out loud.

30:00 JF: Okay, so for the non analyst on this podcast, user stitching as connecting data sets?

30:07 MK: It’s basically using identifiers like their customer ID, their mobile IDFA, their email address, and then trying to link back to different online visits that they had so that you count them as one visitor instead of 10. Does that make sense?

30:21 JF: Yeah, so the example you’re giving is that the mobile application that doesn’t have a login process…

30:26 MK: Is that common? Or do most companies make…

30:30 JF: Well, you could… Well, a good example it might be like a newspaper or a news publication that doesn’t have a paywall, so in that case they don’t necessarily need to know who you are, they just look at your offline activities, I.e. Do you did visit a [30:45] ____ you’re now a car intender, and the advertisement that you see is Toyota, Honda, Mazda, etcetera. So you in THE ICONIC since you’re wanting to understand, maybe where someone goes in the real world and do they transact? So, I guess it’s somewhat case by case depending on the brand or application. I know if that got… If that build into their application no names to link one customer can each [31:12] ____ a mobile app with a website, then it’s difficult to do that. However you could potentially have a collaboration between multiple brands where they might understand… Let’s just say you’ve got two non-competing brands, an airline and a quick service restaurant, and both have a relationship with the customer, I.e. The same mobile advertising ID, and the airline knows the person’s email address and the quick service restaurant doesn’t. If those two brands, for [31:40] ____ wanted to collaborate, then they can share data based on their own privacy policies and consents with the customer. It could be something there but that’s getting quite in a rabbit hole… Of likelihoods.

31:52 MK: We’re very good at going down rabbit holes, aren’t we?

31:55 JF: Yeah. I guess I’d put it back on brand owners or people working for brands that don’t have connection points between… If you’ve got, I don’t know, a person’s physical address and you’re not linking that in any way to the online activity, and you’re not linking that in any way to what mobile connection points you’ve got with them, then of course it’s gonna be difficult to pull those together. Or stitch them, using your words.

32:21 MK: Yep.

32:23 TW: So, we’ve been talking a lot about the app but I think in kind of our… Some of our prep you were sort of noting, “What if its mobile web?” Sounds like app… Once they enable location services, you’re in good shape and you have a good data it’s reliable. What are kind of the other…

32:38 JF: Sure. So just really quickly just to wrap up on that mobile app pace. So there are brands out there that do things that are not privacy-compliant through app. So, for example if you’re inferring someone’s location based on the wifi they’re connected to, so certain companies that drive around the streets sniffing wifi addresses and tracking, the fact that it’s like Bob’s free wifi, they know the longitude and latitude of where Bob’s free wifi is, so if you’re in an application and you’re connected to Bob’s free wifi, they can interfere your location. So with mobile applications the privacy consent is… If it’s done through the platforms like iOS and Android and the customer opts in, then you’re complied.

33:18 JF: But if you’re doing something untoward, then there are work arounds, so just to be really clear on that. In terms of answering the second part, you’ve got a raft of different ways of all other sources of location data. So with mobile web you’re getting sort of suburb level accuracy when you’re doing wifi sniffing or scraping a mobile device that’s walking into an area, common in where there’s free wifi networks so you wanted to understand like footfall analytics, retail shopping malls do a lot of it, so they deploy “shopping mall free wifi” and then they scrape the MAC address of your device and where it moves. So your only way as a consumer to stop that happening, let’s just assume you see the smoke sign that they placed by the door, saying that they’re doing it, is to turn your wifi off when you walk into one of these areas. So that involves basically going in and turning off the capability where your phone automatically connects to a known wifi network. Your mobile device has what’s called a MAC address, good news for those Apple owners listening, your MAC address automatically rotates, so it’s changing all the time. With that they start to understand numbers of devices moving to certain areas and then a shopping mall might start to, for example, charge a tenant more rent for an area where more footfall goes than an area where there’s less footfall.

34:36 MK: So, how accurate is that?

34:37 JF: MAC address? Scraping?

34:40 MK: With connecting to wifi’s? Wifi networks. Could you tell where someone is in a store, or more like in a mall?

34:46 JF: Yeah.

34:47 TW: Depends on their routers, right?

34:48 JF: Yeah, depends on what they’ve set it to. But you can pretty much know which aisle of the supermarket they go down, how long they dwell in that area. So depending on the mobile device and the operating system that you’re on generally every 8 to 20 seconds, your device is trying to find a known network. So it’s basically sending out a ping saying, “Is there anything out there?” and if it sees a known wifi network like “Moe’s,” “Aunt Jenny’s,” wifi network that connects so they hook up. So if you’re spending time in that area, depending on your device, every 8 to 20 seconds, it’s counting your dwell time, so you set the heat map warnings movements. So, you might see if your’re at a supermarket that people spend a lot longer dwelling in the pharmacy or cosmetics and health care aisle, than the chips and confectionery.

35:33 MH: Interestingly, for listeners who are interested in learning more about that I’m pretty sure Gary Angel covers that sort of thing. If they come to Marketing Evolution Experience they can choose not to attend are recording in front of a live audience and instead attend Gary’s session that we somehow got stuck head to head with, where I’m sure he will address that mobile tracking option. I’m not bitter, #yesiam.

[chuckle]

35:54 JF: Yeah, actually having seen him present and trying to spur less people to come to our session, it was actually really interesting talking about applying exactly these to shopping malls and alike, and about your staffing levels and all that sort of stuff, so…

36:14 MH: It’s possible that he appeared on a podcast called the Digital Analytics Power Hour and sort of talked about some of this stuff as well. So I hear.

36:22 TW: But we’re clearly saying, “Please do not come to our session, go to Gary’s.” I think that’s the clear message.

36:29 MK: Is this like voting where you tell people not to turn up, and then they actually don’t, and you get really upset when you lose the election?

[laughter]

36:36 TW: No.

36:37 MK: I dunno, I’m just throwing it out there.

36:39 MH: Will you be upset, Moe? Alright, let’s get back to our regularly scheduled podcast.

36:44 TW: Episode number 64: Analogue In-store Analytics with Gary Angel.

36:50 JF: So just on the MAC address piece, the interesting piece about that is, from GDPR perspective, you can’t tell, when I say you, is the retail store owner who’s deployed this technology or the shopping mall owner or the airport, train station network, whatever it might be, you can’t tell if a MAC address is linked to a UK or EU citizen’s phone.

37:14 MK: Oh.

37:15 JF: So, suddenly you would be storing MAC address data based on potentially any, let’s just use Sydney, any UK or EU tourist who comes through one of your areas, you’re suddenly collecting data without their consent.

37:28 S?: Oh, interesting.

37:30 MK: Oh, that’s scary.

37:30 MH: That’ll be interesting to work out because there’s no way for you to even delete that data if they ask you to because you have no idea who they are with just a MAC address.

37:38 JF: And on iOS the MAC address is constantly rotating.

37:42 MK: Wait, but couldn’t you argue that if a MAC address is rotating so frequently, you couldn’t actually tie it back to a user, and therefore it wouldn’t meet… I feel like someone was talking about this recently but I really should stop talking about GDPR because I am not…

[laughter]

38:00 MK: I do not know my shit. But if you can’t tie it back to a user because it rotates so heavily then it wouldn’t be considered personal information.

38:08 JF: So, I guess first of all the GDPR policies are still, at least from the advertising and media industry perspective, are still being ironed out and we’ll have to wait post their go-live date to see how that plays out, but you could argue that as one stance, Moe. You could argue that it’s come from a consumer’s device that they didn’t provide their consent in any capacity. So, which side of that do you sit on? And then you’ve got Android which is the majority of mobile users both in Australia at 65% roughly and then globally who don’t rotate their MAC address.

38:43 TW: How long has Apple been… I mean, ’cause it used to be MAC address was kind of the immutable link to the hardware and I am assuming they started rotating it specifically because it was such an easy kinda permanent cookie but I completely missed that they were rotating it now. Is that like, for the last five years? How out of touch am I, is the question?

39:01 JF: No, it’s definitely not the last five years. I would say it’s the last ball-park 12 to 18 months. It was one of the platform updates. It might have been iOS 8 or 9. I can check and give it to you for the show notes.

39:12 TW: Okay, yeah. I missed that. That’s interesting.

39:15 MK: Okay, if we’re talking about this, isn’t all of this whole industry about better leveraging location data somewhat dependent on the Apples and the Googles and all that sort of stuff? I mean, ultimately, what happens? ‘Cause I’m thinking about the context of the mobile IDFA. Suddenly it’s like a MAC address and they start regenerating them or they decide to not have them at all or… I mean, I don’t know, isn’t that kinda scary that so much could be built on companies who, if they change their policies…

39:47 TW: Are you constantly chasing that they’re gonna keep… Is privacy concerns go up, they’ll keep trying to… Like the rotating MAC address was a way to foil somewhat, some level of tracking, and are you playing a constant game chasing, trying to keep working around whatever they’re putting in place?

40:04 JF: Sure, yes, good question. So I guess there’s lots of businesses and industries out there built on platforms or reliant on contingencies of others. In the case of the mobile advertising ID it was created because previously people were using the UDID, which is the unique device ID. That’s more like your chassis number in your phone. So, basically they created the advertising ID for, as it says, advertising purposes. But interestingly, Apple themselves, you know they play a high privacy moral standards and things like that, but they themselves use location data for their own advertising businesses. There’s a platform called Apple Ads, and just to quote them, they say, “Device location: If the location based Apple Ads system service is enabled, then your location may be used to serve you geographically relevant ads. Your device location is not stored by Apple’s advertising platform and profiles are not constructed from this information.”

41:01 S?: Right.

41:02 JF: However, they’re using that to sell advertising within, say, Apple News or whatever.

41:06 MK: So, basically, the whole world’s corrupt.

[laughter]

41:08 JF: I wouldn’t say that, I’d just say that people are using that location data for commercial purposes.

41:14 MK: I know, I just feel like we’ve just been talking about privacy and GDPR so much. And I do like that we talked a bit about the customer perspective and how you can add value, but yeah, there’s a bit of me that’s like, “Everyone’s doing it.”

41:29 JF: And I think the interesting thing is by default, every Apple phone has this enabled and turned on. I could give you in the show notes how every consumer, sorry, how every listener can see, if you’ve got an iPhone, where you’ve been recently, so you can actually see what Apple’s tracking of your location and you can choose to opt-out if you want.

41:50 TW: You’re gonna single-handedly drive that Android adoption in Australia from 65% to 75%.

41:57 JF: Yeah, but Android’s owned by Google so…

[chuckle]

42:01 JF: They’ve got their own commercial purposes as well. So I think the interesting thing though is it comes down to, as a consumer I think there’s one of two pathways. One is, if it makes my life easier and it creates value out of for me then I’m in for it. If it’s creepy and it’s doing something that I’m not comfortable with then you can opt out. I don’t know. At least from my perspective I kind of see these two unique camps and you either sit in one… The great thing is you can choose it at a brand to brand level. So you don’t have to sit there saying, “I’m either in or out for everything.” You can say, “I’m gonna do it for Uber, ’cause there’s a value creation there, but I’m not gonna do it for brand X, Y, Z, ’cause I can’t see the value.” Consumers have the freedom to be able to do what they want on a one-to-one level with each individual brand, consumer to consumer.

42:49 MH: What about in context like what’s happening with WeChat right now, where they’re actually bringing apps into their app? Have you explored that or dealt with that at all?

43:01 JF: Creating like the Chinese mega-apps?

43:02 MH: Yeah. Yeah. So do those just share the device location, permissions of the app, they live within or?

43:10 JF: Good question, I’m not 100% certain or knowledgeable on the Chinese market or how their platforms operate. My assumption would be that there would be some level of data granted to the sub-apps that sit within the master apps. They’re essentially almost platforms themselves like the WeChats, the Tencent, etcetera. Almost like the App Store or Google Play. So, if you look at the idea, that let’s just say, they’re trying to move towards being an App Store or Google Play, then the master platform by Google Play or App Store gets access to those data points and consents. And then they pass that on to the individual app owners. So if you used to house within WeChat, for argument’s sake, then my assumption would probably be that those people at those individual service offerings will get access to their data. But, it probably would be warranting further investigation.

44:07 MH: Yeah, sorry, kind of a rabbit trail, but kind of interesting at the same time.

44:12 JF: That’s alright.

44:14 MH: It seems like that’s probably the way a lot of these large companies who have, sort of, big user platforms will try to go because it frees them from the platform owners a little bit, or from the device owners.

44:28 JF: I think in general, in data, I’m a big believer that brands themselves will start to collaborate with non-competing brands and share, not in a creepy capacity whatsoever, but start to share insights on who the customer is. And I believe from a location perspective there’s two ways that you can identify that. That’s either A, through a common mobile advertising ID. So you download an airline app and a quick service restaurant app, you share the same mobile ad ID. Or vice versa, one person’s mobile ad ID turns up at another persons physical location. So therefore, by default, those two brands have a crossover with that customer. So I think what we’ll see moving forward is more and more brands that are non-competers, start to collaborate from a data perspective and not rely on the Googles and Facebooks of the world.

45:16 MH: And then putting on my sort of philosophical cap for a minute, this reminds me a lot of how data collection started on the web, which was… The device cookie was never intended to be used, really, as a tracking mechanism or an identification beyond just helping to know whether you’d been to that site before or not based on the data stored in the cookie. And then over time all of us, people who were interested in knowing more, we grabbed onto those things and started building tracking capabilities on top of them. I just wonder if maybe, technologically a lot of times we put a technology in place without thinking through how the measurement piece or the tracking piece will interact, and I wonder what it might look like if we were able to kinda develop that simultaneously to create kind of a mutually shared platform or technology.

46:16 TW: You’re adorable. You and your little…

46:18 MH: Hey, hey, listen.

[laughter]

46:21 MH: I just read an article, it’s not even my last call, but they’re starting to develop the little headset that reads your mind. So like that’s the obvious next step past the phone.

46:31 MK: Oh, Jesus.

46:31 MH: So we had computers, now we’re gonna have phones, and eventually we’ll have little ear pieces that are reading our thoughts. Maybe before we start giving up all of our thoughts for data collection we think about measurement beforehand, certainly I don’t want Google knowing everything I think.

46:47 TW: I stand by my “You’re adorable,” comment.

46:50 MH: Well, be that as it may just. I…

46:56 JF: Do you not want Google knowing all your information if it makes your life easier, better, more enjoyable, efficient, happier?

47:03 MH: I already give Amazon and Google a lot of information for that purpose, but as we progress, and again I’m creating a future frontier ’cause honestly, I don’t know what’s gonna be next after the phone. But what I mean is, it seems like all of the tracking follows the technology and gets creepy real fast because we’re doing things that frankly are hacks in a certain sense, around the technology to create the measurement tools that we want to keep. To create the commerce and even the value. It wasn’t built necessarily to do that, it’s all coming in behind those kinds, the development of the technology itself. Somebody someday was like, “Hey. We can go gather the Mac address or the Wifi or something of all these phones as they pass by us with this beacon.” And then off to the races we go, but that wasn’t a starting point like, “Hey. As we design Android or as we design iOS, let’s create a method so that we can connect people in realtime to physical locations.

48:07 JF: I think it’s an interesting observation, and if you look at the human psyche and the competitive nature of who we are as humans against one another, whether you’re a human being or a business, you’re trying to outdo one another. If you compare, say, the analogy of people finding a data sphere or a methodology to create new value, versus trying to build something like, hey… I look at a Beacon technology, for example. I look at that going that’s widely overhyped, and people basically try to create a technology without a value proposition attached to it. And now they’re trying to segment Beacons and create a value proposition back to it, but they created the hardware first without looking at what the value proposition to the customer is.

48:54 MH: Yeah, I mean, it’s certainly an interesting set of challenges and very interesting conversation one in which, unfortunately, we’re gonna start to wrap up because that’s what we do. We get into a really great discussion and then we end the show as quickly as possible because Tim Wilson’s on a tight schedule… No, I’m just kidding. It’s not Tim’s fault.

[chuckle]

49:14 MK: I feel like this discussion was really one where James and I could have benefited from having a drink with us.

49:18 MH: Yeah, again, that’s where I’m coming to. Everybody’s saying, “Hey, let’s get rid of this irrational stigma against… ”

49:26 TW: Morning drinking.

49:27 MH: “Morning drinking in Australia.”

49:28 TW: Make it a thing.

49:29 JF: I’ll happily accept gifts of gin coming from America. Any particular gin distilleries I’ll happily accept gin gifts and continue the conversation.

[laughter]

49:39 TW: They don’t have juniper bushes in the…

49:42 MK: We make a much better gin in Australia. Let’s be honest.

49:45 JF: Oh, I just like trying new batches.

49:47 MH: Really? Is it a highlight? Gin as a beverage is well done over there?

49:53 MK: Yeah, it’s huge. It’s huge. When you guys decide to come record a show in Australia, I have got all the Australian gins for you to try on my bar.

50:01 MH: It’s like you know that I’m not in Australia. Are you tracking my phone?

50:05 TW: That’s right, are you tracking his phone?

50:07 MH: Alright. One thing we like to track on this show, as we wrap up, is something we call a last call. That’s where we go around the horn, talking about something we found interesting that could be at any location. But since you’re our guest, James, do you wanna go first?

50:21 JF: Sure, so you kinda dropped me in it, but I think…

[chuckle]

50:26 JF: Given the subject matter, I think an interesting concept will be in the future will people create their own, let’s just call it data banks, where they provide their data and their license to brands, sorry. I don’t think it’ll be… Or even other consumers, I don’t think it’ll necessarily be from a revenue perspective but it’s more the idea that consumers themselves could own the data that’s on them and they provide access rights to brands. And then they can remove a brand as quickly as they add one. Instead of Facebook owning all the data on the customers, what if the customers created their own data bank on themselves and gave certain rights and permissions to brands.

51:07 MK: That’s so scary.

[laughter]

51:09 JF: Scary?

51:09 MK: Yeah, I’m a bit scared. I’m normally good with change.

51:12 JF: But you as the consumer would have the control.

51:15 MK: I know.

51:16 TW: It’s interesting. Actually, the very first Accelerate in San Francisco years ago my little five-minute or ten-minute little thing was kind of around that. What if it totally got spun on its head and reimagined as opt-in, trust, total tracking, how might that work? And a couple of people came up and they’re like, “Are you working on that?” And I’m like, “Oh, hell no.” The assignment was to just like, pie-in-the-sky, dream of the future. [chuckle]

51:45 MH: Some of it is already opt-in. Anyways, we’re getting… Last call, Tim. Moe, you ready?

51:53 MK: I’ve been flying about recently and one of the Australian airlines, Qantas, has actually put a whole bunch of the LinkedIn Learning sessions on their video service when you’re flying, which has been really good. And one question that I often get asked when I start talking about women in analytics is how men can better support women. So, I was watching a video not long ago by Stacy Gordon on the LinkedIn Learning thing about unconscious bias, and I really recommend going and checking it out because I think unconscious bias is one of those things that we probably talk about, but don’t necessarily completely understand. And Stacy was very good at giving an idiot’s guide to unconscious bias so I recommend people go check that out.

52:39 TW: Nice. Michael, you wanna not go last, just for giggles?

52:43 MH: Yeah, I would love to. Recently I was brought back to this, which is something we’ve actually talked about on this show before. In fact, we actually even did an episode about it but I doubt they’ll ever let me do this again, which is The CMO Survey. I’m a huge fan of this survey and the February results recently came out for 2018. And actually, apropos to our topic, there’s a whole section on mobile marketing. So they ask… This is done by the Fuqua School of Business at Duke University, and they talk to CMOs about all of these different marketing topics. It’s always a good read to get a perspective of sort of where senior leadership in the marketing organizations around the world or in the US, anyway, are thinking. Check that out at the cmosurvey.org.

53:36 TW: Nice. My last call will be to keep him the most referenced individual ever in the history of the show is gonna be a Matt Gershoff. A few weeks ago he wrote a post and I know he’s been working on it for a while. And it was inspired by a discussion with Elea Feit, past guest, and Kelly Wortham, past guest, and he was making a point that the… Post is, “Do no harm or A/B testing without P values.” And as most things with Matt, it requires a couple of readings, but basically he’s actually positing that if you think about what you’re trying to get out of your tests and he says, “Look, you’re either trying to do a do no harm test or go for it test, and if you’re really clear on that, that if you’re doing a go for it test, you don’t necessarily need to worry about P values. Just take the one that has the better results and run with it.” And I think as I’m on my little data science journey, it’s one of those that actually has some really pretty profound ideas, which is just not how we think about A/B testing in the digital optimization world, and yet, it seems like a pretty powerful thought that kind of has broader ramifications about how we think about when we’re doing testing. So, a hearty endorsement of that post.

54:55 MH: Nice. Awesome. Okay. You’ve probably been listening to this and you’ve probably been listening to it on your phone. And some of you, I already know, have been listening to it at 1.5 speed. But…

55:08 TW: At McDonald’s.

55:09 MH: At McDonald’s, geo-fenced, anyways, by anyways…

55:14 TW: In route to the airport for a United Airlines flight.

55:15 MH: Yes. And you have only 15 minutes left before your boarding starts. So, make sure your TSA-pre at least in these coming geographies, which we’ll insert in post production. Okay. No, you may have comments or questions. Now, we would love to hear from you. The best way to do that is through either our Facebook page or our website or on the Measure Slack. And James, are you active in the realm of social media? Could people reach out to you on Twitter if they wanted to learn more?

55:47 JF: Yeah, sure. Just @jamesfogelberg.

55:50 MH: At James…

55:51 JF: Fogelberg. F-O-G-E-L-B-E-R-G.

55:53 MK: I’ll tweet something, I’ll tweet it.

55:56 MH: It will be tweeted. And we’ll make sure it’s in the show notes as well. But we wanna give people an avenue to reach out as well. Lot of valuable information.

56:04 JF: Sure. [56:04] ____ conversation.

56:06 MH: Yeah. Thank you James for sharing some of that with us today and our listeners. And as always for my two co-hosts, Tim Wilson, Moe Kiss, you’re out there, we know you are, we know exactly where you are, but we also want you to keep analyzing.

[music]

56:26 S1: Thanks for listening and don’t forget to join the conversation on Facebook, Twitter, or Measure Slack Group. We welcome your comments and questions. Visit us on the web at analyticshour.io, Facebook.com/analyticshour or @analyticshour on Twitter.

[music]

[background conversation]

 

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.

One comment on “#087: Offline Data in an Online World with James Fogelberg

The Digital Analytics Power Hour © 2018