Location Intelligence

Why Companies Can’t Afford to Stop Innovating in an Economic Downturn

When the economy suffers and times of crisis arise, businesses look to survive by finding places to cut costs. Too often, this involves scaling back investments in innovation and doubling down on what has worked in the past. However, companies need innovation to keep up with the rapid and drastic changes brought on by economic downturns, or risk losing in both the short and long term.

Let’s explore what past periods of economic turbulence can teach us about the importance of innovation during tough times, with a particular focus on how location intelligence can help organizations not only survive in a crisis, but thrive.

Making tough choices: why investing in innovation during economic uncertainty is better for long-term growth

Economic downturns are pivotal moments for businesses. Across industries, organizations are faced with difficult choices about where to cut costs and optimize resources. While most leadership teams respond by cutting spend, particularly in new technology, historical evidence actually shows that companies investing in innovation during a downturn fare better in the long term.

The initial impact of a bad economy on innovation

History has taught us that inevitably, some organizations will fail while others manage to overcome the challenges of financial stress. There are many things we can learn from examining past economic events as we possibly prepare to weather one ourselves – particularly the role innovation plays in company survival.

The first instinct during a financial crisis is to cut costs and play it safe. Expenses deemed ‘non-essential’ are eliminated as companies double down on their tried and true strategy for success. While this can mean canceled company retreats and hiring freezes, it also tends to stifle investments in innovation. In June of 2020, a McKinsey survey found that in response to the economic downturn brought about by COVID-19, executives were, “deprioritizing innovation to concentrate on four things: shoring up their core business, pursuing known opportunity spaces, conserving cash and minimizing risk, and waiting until there is more clarity.”

Tightening spend is essential during uncertain times, but pulling back on innovation can actually have long-term negative consequences for an organization. Many of the challenges a business faces in these periods can be analyzed and even alleviated by technology. The same McKinsey report claims that “playing it safe may be a shortsighted decision,” and encourages companies to leverage innovation as they adapt to changing market conditions and prepare for the future.

History shows that investing in innovation is better long term

A Bain study on the 2008 recession and its aftermath led partners to conclude that, “the eventual winners in any downturn are companies that aggressively captured the opportunities unique to recessions.” The study shows that one of the main ways companies can go on the offensive is by finding innovative solutions to the challenges brought about by a bad economy. McKinsey’s survey also looked at the 2008 crisis as a data point, where results showed that companies investing in innovation during the recession outperformed those that did not by more than 30% in following years.

“The companies that power into a downturn and accelerate out of it aren’t complacent. They define the scenarios they’re likely to confront and lay bold plans to profit from whatever they encounter.” – Bain

McKinsey found that companies that invested in innovation and technology through the 2008 recession not only out-performed non-innovators by 10% during the crisis, but continued outperforming them by upwards of 30% in the following years.

Similar research suggests the same. An article published in 2022 by Northwestern University’s Kellogg School of Management argues that economic crises are more nuanced than we think, and that strategic investments in innovation can determine which companies will thrive, merely survive, or not endure at all. Harvard Business Review (HBR)’s 2019 examination of past recessions and company outcomes found that, “any of the companies that merely limp through a recession are slower to recover and never really catch up.” In other words, hunkering down and withholding investment in innovation can help preserve cash in the short term, but seriously damage a company’s ability to grow in the future.

The case for innovation during a crisis: short-term benefits to investment

Innovating during an economic crisis is not only beneficial in the long term. Oftentimes, companies that invest in innovation during a recession perform better in following years because of the short-term benefits of innovation – mainly increasing operational efficiencies and leveraging technology to adapt in a changing market.

Preserving resources by improving operational efficiencies

Regardless of the economy, innovation is often geared toward increasing operational efficiencies that drive growth. So while saving cash by not investing in tech may seem like a smart short-term solution in tough times, it can prove detrimental when considering the resources that could be saved by empowering employees and business processes with innovative solutions.

Businesses that continue investment during an economic crisis avoid missing out on technological developments that will be beneficial when the crisis is over, and instead can emerge as leaders in innovation. In fact, HBR says that, “the resource scarcity that typically accompanies recessions forces innovators to do things they should have been doing already.”

For many companies, this means accelerating digital transformation to increase efficiencies and cost savings. During the height of the COVID-19 pandemic, organizations made steep investments in technology to help their teams be productive at home, things like video conferencing, project management, and instant messaging tools. While investing in these technologies was an expense, doing so enabled these companies to create a productive environment for remote work, weather the storm, and emerge from the pandemic well-adjusted to a hybrid world.

“Crises are like adrenaline for innovation, causing barriers that once took years to overcome to evaporate in a matter of days. Entrenched orthodoxies on ‘the way things are done’ are replaced with ‘the new way we do things’ almost overnight.” – McKinsey

Adapting to a changing market

It’s important to remember that an economic downturn does not only affect a business; it affects that business’ customers, its suppliers, and other critical players that allow an organization to function. Filippo Mezzanotti, Associate Professor of Finance at Kellogg, emphasizes that, “these are times when people are forced to rethink the way they’re organizing their life, their business, or their activity.” Given these changes, doubling down on what worked before a crisis is actually not a safe bet. Instead, by analyzing the situation and adapting a business model or strategy, companies can come out ahead in a bad economy.

“Many businesses simply cannot operate as they have in the past. What made a company successful historically may no longer be possible during or after the crisis…The assumptions that supported years of stable, predictable growth may no longer be valid.” – McKinsey

HBR points out that, “the first reason to prioritize digital transformation ahead of or during a downturn is that improved analytics can help management better understand the business, how the recession is affecting it, and where there’s potential for operational improvements.” Investing in innovation can help companies understand how their business must change to meet evolving customer needs, identify areas of opportunity, mitigate risk, and double down on what really moves the needle.

Leveraging location intelligence to adapt and grow during strenuous times

There are many different innovative solutions companies can invest in during financial turbulence, and prioritizing should ultimately depend on the needs of each organization. As pointed out before, investing in tools that enabled remote work was key for some businesses to survive during the lockdown stage of the pandemic. However, one area of technological investment that is always a safe bet, regardless of the economic landscape, is location intelligence.

Let’s again consider the two main short-term benefits to investing in innovation: increasing operational efficiencies and adapting to a changing market. Location intelligence provides organizations with the insights they need to optimize operations and tighten up spending on nonessential initiatives. For example, with geospatial data and mapping tools, retailers can visualize where their customers are in relation to top performing stores and reallocate resources to support these locations.

Understanding how locations must adapt to changing customer needs is also imperative for retailers in these times. As we saw during the pandemic, many people relocated outside of urban areas. If retailers did not realign their strategy to meet the changing customer landscape, and instead doubled down on old assumptions, it’s likely they did not fare well. Location intelligence data and tools enable organizations to stay up-to-date with a dynamically changing world and to adapt their strategies accordingly. The insights derived from location-based analysis give companies the information they need to weather tough conditions, while also laying a solid foundation for succeeding in a changed market when the crisis resolves.

Making tactical adjustments quickly will not only help a company make it through a downturn, but also set them up for long-term success in a changed market. At Foursquare, we deliver high quality location intelligence solutions for businesses across industries. Many of the workflows our customers utilize most become even more critical as the economy shifts. Here are a few examples of how common location intelligence use cases increase in importance during financial duress, and why investing in geospatial technology and data is a smart way to spend when trying to save.

Site deselection

Real estate is a large expense for many businesses, particularly during a crisis. Understanding how local market conditions impact specific store locations can help companies make informed decisions about which venues to close, and what that impact will be on overall operating expenses. For instance, a restaurant chain may need to cut costs by choosing a location to close. Location intelligence can identify which neighborhood is most likely to underperform in a time of crisis- making it an ideal tool for site deselection.

Predictive modeling

Bain’s study emphasizes that stress testing and modeling different scenarios can help companies adapt quickly both during and after a crisis. This is particularly valuable for organizations as they navigate an uncertain economy and prepare for multiple outcomes. For example, if a nearby competitor is closing a store, geospatial data can predict how the closing might impact your own store’s performance. Similarly, forecasting sales based on possible changes to customer needs can reveal new market opportunities and mitigate risk.

Trade area analysis

As stores open or close, consumers migrate, and customer needs evolve. This means change for trade areas. Geospatial data can reflect these changes to give businesses the necessary tools to adapt their strategies in response. Consider the dramatic drop off in commuting that occurred during the COVID-19 pandemic; trade areas around popular office lunch spots suddenly became irrelevant. Instead, these lunch spots had to adjust their trade areas to include people now working from home nearby, or perhaps even pivot to a delivery model. Assuming pre-crisis trade areas are still intact can result in misallocated resources and lost opportunities. Location intelligence can help avoid these costly mistakes.

Product development & sales enablement

All of the dynamic factors that result from a changing economy can actually be an opportunity for companies to develop new products or discover new sales channels. HBR says that, “downturns can be great times to introduce offerings that connect with consumers who have tighter purse strings or are naturally frugal given continued uncertainty,” and that there is an opportunity to create solutions for challenges posed by a crisis itself. With geospatial insights, a company can identify target markets for new products or services aimed at solving new problems. As an example, a consumer packaged goods (CPG) brand can layer various geospatial datasets together to identify stores where their new discount products would be most successful.

Competitive intelligence

Companies understand the ins and outs of their own business, but often struggle to maintain the same level of intelligence about their competitors. During an economic crisis, keeping tabs on the competition is essential to making informed decisions. With geospatial data and tools, companies can visualize where their competitors are located, where they are closing stores, and determine how that ultimately impacts their own business. McKinsey points out that economic crises are also times when companies pivot and new competitors emerge, so it’s important to stay up-to-date in your analysis.

Supply chain analysis

Customer behavior is not the only factor that changes during an economic crisis. The same economy is impacting all businesses, including those that form the supply chains of others. When a key warehouse or manufacturing facility in a company’s supply chain is closed or repurposed, location intelligence can guide businesses to reoptimize routes and avoid wasteful ‘fixes’ that are not actually the next best option.

Investment insights

Financial firms investing in other businesses rely on data to perform due diligence, monitor portfolio performance, and make decisions on both the buy and sell side. Geospatial data is a critical component to these investment analytics, as it provides firms with an otherwise impossible level of local insight. For instance, if a firm is choosing between two businesses to purchase during a downturn, looking at how consumers are interacting with and visiting store locations enables predictions on how each company will fare during and after the crisis, ultimately helping the firm better hedge their bets and invest in the company with higher growth potential.

These are just a few examples that demonstrate the ways in which geospatial data empowers organizations to make smarter decisions during a crisis. To learn more about the wide variety of applications for location intelligence and how specific companies are implementing Foursquare data in their strategies, explore our repository of case studies.

Placing the right bets by investing in location intelligence

With all of the changes brought about during an economic crisis, companies simply cannot skip out on investing in location intelligence. At a time when every penny counts, it is critical for organizations to understand how the fluid market conditions impact their core business and what they must do to adapt.

To mitigate risk and adapt to a changing market, both companies with a history of innovation and those just beginning their journey are choosing to invest in location intelligence. As a trusted leader in the geospatial industry, Foursquare is a safe bet for organizations looking to invest in innovation and reap both short and long-term benefits. With high quality, up-to-date, and accurate location-based insights, businesses across industries can actually grow during an economic downturn, while setting themselves up for success in the future.

Get in touch with Foursquare today to see how our location intelligence solutions can help your organization survive and thrive during an economic crisis.

Interview with Humayun Irshad: Leveraging Street View Imagery and Deep Learning Models to Build Parking Augmented Maps

One of our MLconf Program Committee Members, Reshama Shaikh, recently interviewed Humayun Irshad, Lead Scientist of Machine Learning at Figure Eight regarding his work using Deep Learning Models to Build Parking Augmented Maps.

Finding proper parking spot in cities such as San Francisco is a driver’s everyday struggle as streets are reaching their capacities. Parking signs are installed to inform drivers about parking regulations applied to the curbs. However, in many cases parking signs can create confusion, cause more traffic and harm the transportation, environment and pedestrian safety as drivers spend a lot of time circling around the streets and reading parking signs to avoid tickets. Parking rules often change a lot among curbs. Multiple rules may be applied to curbs and parking signs may contain too much text that drivers need to read and understand correctly to avoid tickets. On the other hand, public street-level databases such as google street view contains tons of visual information about parking regulations which is locked in the images. Computer vision models can leverage such databases to extract parking rules from images. Parking rules in digitized format then can be used to build tools which presents parking regulations in a format that be easier for drivers to understand and follow or send them notifications about change of rules. The key benefit of exploiting public street-level imagery databases such as google street-view imagery is that once parking signs are detected and read from images, they can be located on the same platform and used as metadata with maps to provide parking related services to drivers navigating through those maps. Building a computer vision based system for extracting and digitizing parking regulations have several steps. First, parking sign data is collected and labeled for training deep learning models. Afterwards, state of the art object detector are used to find the location of parking signs in images. Parking rules are extracted from parking signs using multiple text reading API combined with human verification. In final step metadata about actual location to which panoramas belong are extracted to locate parking signs on the map. In this study, we focus more on the data collection, annotation and training and evaluation of deep learning models for parking signs detection from google street-view images. –Humayun Irshad

RS) Tell us briefly about yourself and your work.

HI) I am currently the Lead Scientist of Machine Learning at Figure Eight, the essential human-in-the-loop AI platform for data science and machine learning teams. My expertise is in developing machine learning, more specifically deep learning frameworks for various applications like object detection, segmentation and classification in fields ranging from medical, retail, self-driving car, satellite, fashion, etc. Nowadays, I am building Active Learning frameworks for selection of training data from labeled or unlabeled dataset to build model to avoid over-training and dealing corner cases. I have 3 years PostDoc experience at Harvard Medical School where I developed machine learning and deep learning frameworks for Computer Aided Diagnosis system including region of interest detection, nuclei and gland detection, segmentation and classification in 2D and 3D medical images. I received my PhD in Computer Science from University of Grenoble France.

RS) Regarding one of your current projects at Figure Eight, “Finding Proper Parking Spot”, at what stage is this application? When will it be available for the public to use?

HI) Parking Sign Recognition is a research project that I started early this year. I proposed and developed Naive Active Learning framework to select the training data for building object detection model to recognize parking sign in unlabeled dataset using human-in-the-loop (crowdsourcing). This framework manages to reach more than 86% accuracy in prediction of parking sign in San Francisco Street View images (downloaded from Google Street View), after 5 iterations of Active Learning. The paper is under-review in one of the AI related conference. Training and Validation dataset is available on Figure Eight website for download. We are brainstorming about how to make this service available to public either with web API or mobile app.

RS) How has your experience in cancer research augmented your skill set for your current role as Lead Scientist of Machine Learning and Computer Vision at Figure Eight?

HI) I am lucky that I got extensive training and experience working with wide range of medical images including 2D and 3D image modalities in H&E stained, immunohistochemistry and fluorescence images. These images are acquired with a wide range of scanners and microscopes like bright field, confocal, lightsheet and super resolution microscopy. At Harvard, I developed a number of computational image analysis frameworks for cancer diagnosis and prognosis, more specifically for Breast Cancer and published in the top Journals and presented at well know conferences and meetings. I faced different challenges like image acquisition issues including noses inherited from scanners and tissue preparation, problems in collection of ground truth labeling from medical experts, handling imbalance dataset and building models to outperform in any type of images, not only across scanners but also across hospitals. These experiences increased my domain knowledge and expertise in machine learning, computer vision, image acquisition, image analysis and statistical methods. Eventually, it gave me skills and confidence to deal with a wide range of image analysis problems and issues.

RS) What tools are you using for computer vision? What are the deep learning approaches that you are incorporating into your current projects?

HI) Nowadays, I am extensively using current state of the arts frameworks and models like Tensorflow, Keras and PyTorch with Python. Mostly, I am working with image classification, image enhancement, object detection, instance segmentation and semantic segmentation problems. Depending on dataset and complexity of problem, I selected different models for different applications. For image classification, I built a pipeline of all state-of-the-art deep models including VGG, Resnet, Inceptions, DenseNet, NASNet and all their derivations. For image enhancement, I tried either traditional computer vision techniques or GAN (Generative Adversarial Network) models to improve image from nose or reconstruct images from occluded and missing parts. I built customized object detection pipelines of Faster R-CNN, SSD (Single Shot Multibox Detector) and YOLO (You Only Looks Once) models for a wide range of object detection in number of domains but not limited to self-driving cars, drones, satellite, medical, fashion and retails. For object segmentation, I developed frameworks of auto-encoders and GAN models for instance and semantic segmentations. All these deep learning and machine learning frameworks are also part of our crowdsourcing platform to assist and increase the performance of our crowd on different labeling jobs.

Now a days, I am working on Active Learning techniques and strategies for optimal selection of data set for labeling and training the machine learning models.

RS) What are some of the recent research papers you have read?

HI) Here is a list of research papers that I read in the past month:

Large scale GAN training for high fidelity natural image synthesis (ICLR 2019)
Learning transferable architectures for scalable image recognition (Barret Zoph, 2018)
3D MRI brain tumor segmentation using autoencoder regularization (Andrily Myronenko, MICCAI 2018) – won the MICCAI 3D MRI Brain Tumor Segmentation Challenge
Dermatologist level of classification of skin cancer with deep neural networks (Andre Esteva , Nature 2017)

RS) I noticed that you released a video entitled An Active Learning Approach to Image Recognition. Can you elaborate on that approach?

HI) Deep learning models have been used extensively to solve real-world problems in recent years. The performance of such models relies heavily on large amounts of labeled data for training. While the advances of data collection technology have enabled the acquisition of a massive volume of data, labeling the data remains an expensive and time-consuming task. Active learning techniques are being progressively adopted to accelerate the development of machine learning solutions by allowing the model to query the data they learn from. In this approach, we introduce a real-world problem, the recognition of parking signs, and present a framework that combines active learning techniques with a transfer learning approach and crowd-sourcing tools to create and train a machine learning solution to the problem. We discuss how such a framework contributes to building an accurate model in a cost-effective and fast way to solve the parking sign recognition problem in spite of the unevenness of the data associated with the fact that street-level images (such as parking signs) vary in shape, color, orientation and scale, and often appear on top of different types of background.

RS) As someone who has transitioned from academia to industry, have you been able to continue your research and publish?

HI) Yes, I am lucky in a sense that my current employer (Figure Eight) also encourages to do research and publish in well know conferences and journals along with development of frameworks and models for customer and platform. I have submitted two research papers (which are under-review) in well-known conferences of AI and Machine Learning. Besides, I am also finishing few of my research projects and publishing in Journals which I started at Harvard.

RS) Here are a few of the trending topics in data science. What are your high-level thoughts, in 1-2 sentences, for each topic.

Algorithm reliability

Algorithm reliability is one of the top measure to evaluate the algorithm. In term of classification algorithm, the reliability of classification as an estimated probability that the classification is in fact the correct one.

Reproducibility

Reproducibility is very critical in deploying the model in industry. It minimizes variations in case of rerunning the prior experiments, makes the model fault tolerance and iterative improvement of models.

Diversity

Machine learning models should be better fit for special requirements of different tasks. Many factors can affect the performance of the ML process, among which diversity of the machine learning is an important one.

Open source

I always have support for open source frameworks and encourage other as well.

Opportunities for entry level data scientists

Entry level data scientists should focus on one aspect/problem at one time and after mastering and gaining significant knowledge, then go for next one. Try to avoid ML cocktail in start of your journey.

RS) Thank you for participating in this interview.

About Humayun Irshad

Humayun Irshad is currently the Lead Scientist of Machine Learning at Figure Eight, the essential human-in-the-loop AI platform for data science and machine learning teams. He has expertise in developing machine learning, more specifically deep learning frameworks for various applications like object detection, segmentation and classification in fields ranging from medical, retail, self-driving car, satellite, fashion, etc. Nowadays, he is building Active Learning frameworks for selection of training data from labeled or unlabeled dataset to build model to avoid over-training and dealing corner cases. He has 3 years PostDoc experience at Harvard Medical School where he developed machine learning and deep learning frameworks for Computer Aided Diagnosis system including region of interest detection, nuclei and gland detection, segmentation and classification in 2D and 3D medical images. He got a PhD in Computer Science from University of Grenoble France.
A computer scientist with expertise in machine learning, deep learning, computer vision, bio-medical image analysis and statistical methods. He likes to use his analytical mind not only when building complex models, but also as part of his leadership philosophy. Humayun also enjoys sharing his experiences in technical and non-technical audiences in conferences, seminars and meetings.

LinkedIn: https://www.linkedin.com/in/humayunirshad/
Twitter: @HumayunIrshad
Blog: https://sites.google.com/site/humayunirshad/

Interview with Abhineet Gupta

Q1: Tell us briefly about yourself and your work.

AG: I am the Resilience Engineering Lead at One Concern, a Palo Alto-based benevolent AI company, with a mission to save lives and livelihoods before, during and after disasters. I have been involved with One Concern for almost 3 years, essentially since the company’s inception. I developed the resiliency framework that lays out the conceptual architecture for us to meet our vision to achieve planetary-scale resilience, quantify it at a community level, and constantly improve resilience over time. As part of this effort, I have created specific metrics for measuring how access to critical resources is impacted in disasters and to identify those communities that are most at risk before, during and after disasters. Besides working on resilience, I have also contributed to development of our Seismic product to estimate near real time impacts in an earthquake, created evaluation metrics for our machine learning models and set up the simulation pipeline to generate earthquakes for preparedness trainings and mitigation policy design.

Prior to One Concern, I completed my PhD thesis from Stanford on creating the framework to estimate community level seismic risk from man-made or induced earthquakes, with a case study implementation in Oklahoma. As part of this framework, I implemented and developed novel statistical methods to quantify how seismic rates change over time with changes in oil and gas extraction processes responsible for causing these earthquakes. Prior to my PhD, I worked as a structural design engineer at Magnusson Klemencic Associates in Seattle and was responsible for the design of an iconic building in Boise – Jack’s Urban Meeting Place – along with helping out with design of buildings in San Francisco, Delhi and Kuala Lumpur.

Q2: I had not previously heard of “resilience engineering.” Tell us a little bit about that field. Is it new? How many people worldwide graduate with a degree in resilience engineering?

AG: As far as I am aware, resilience engineering is not a degree offered by universities. Through our work at One Concern, we realized that most of the work around disaster impacts occurs in silos by hazard or by field of study, while resilience is an inherent characteristic of communities that spans across both expected disasters and unexpected events and across many fields like risk engineering, economics, data science, behavioral analysis, policy and education, urban design and social sciences. As a result, we decided to create this position and team at One Concern to focus on holistic resilience analysis. At One Concern, we have defined Resilience as the ability of communities to maintain their normal functionality and to adapt and grow in the face of acute shocks and long term stresses. We are also starting to see universities like Stanford, Harvard and MIT create multi-disciplinary programs to bring together experts from different fields and companies to work on resilience.

Q3: What is the story behind the founding of One Concern?

AG: One Concern spun off from a machine learning project in Andrew Ng’s class at Stanford. The co-founders, Ahmad Wani, Nicole Hu and Timothy Frank along with Yawar Aziz demonstrated in that project that rapid damage assessment after earthquakes was possible using available data and machine learning. Ahmad was also motivated to help people during disasters from his recent experience being stuck on his roof while floods in Kashmir inundated his town. Combined with the success of their project and encouragement from Andrew Ng, One Concern was founded to rapidly measure impacts after disasters, prepare for and mitigate against them and thus save lives and livelihoods.

Q4: What sort of open data do you use for your models? And what type of data do you (One Concern) collect on your own?

AG: We use a combination of open data, proprietary data and our self-collected data. Our open data sources include the US census data that we use for understanding demographics and building inventory, weather forecast data to estimate flood inundation and USGS earthquake data to simulate earthquakes.

One Concern has also created a Digital Anthropology team which is tasked with collecting disaster relevant data from across the world. Some of the regions where our team has been deployed include Bogota to collect data about buildings, demographics and socio-cultural parameters, and Mexico City and Lombok to collect post-earthquake damage data. We use a variety of data collection tools like apps, cameras and remote sensing imagery. A critical component of data processing for us is that individual sources do not accurately capture the required features of our models. As a result, our data processing team uses several statistical and data mining techniques to augment data from different sources to ensure that the final datasets represent the ground truth.

About Abhineet Gupta

Abhineet Gupta is the Resilience Engineering Lead at One Concern. He works on quantifying resilience through cities’ abilities to withstand stresses and shocks from natural and manmade disasters, and to provide the technological solutions to increase resilience. Abhineet graduated from Stanford in 2017 with a Master’s and PhD in earthquake engineering and PhD minor in Computer Science with a focus on artificial intelligence. For his PhD, he developed tools to evaluate seismic risk from man-made earthquakes in regions like Oklahoma. Before his PhD, he worked as a structural design engineer at Magnusson Klemencic Associates in Seattle and designed buildings in Boise, India and Malaysia.

LinkedIn
One Concern

Guest Blog by Keshav Dhandhania, Co-Founder, Compose Labs

What Data Scientists Should Know About the State of Location Data

Introduction

The increasing availability of continuous, precise, location data will be a game changer in data science, giving rise to new analysis tools and bringing about a wave of collaboration between data scientists and product teams. It will give data scientists yet another way to influence product decisions and lead to better experiences for users. Along with that it will add to their responsibility of helping users without hurting their privacy.
You may remember being prompted by the Uber app, last year, to allow it to track your location in the background. It promised to only collect data for a small window before your ride, but you couldn’t be totally sure. The only reason cited for the request was a marginal improvement in the Uber pickup process. It sounded like a raw deal for users, but apparently, Uber had done its research and knew that many would tap accept.
In being over-eager for background location data, Uber is not alone. The Facebook app requests it to show you nearby friends. Google Maps promises the forgetful to remember where they parked their car. Even dozens of retailers such as Urban Outfitters and Bloomingdale’s, ask for the permission in exchange for pushing coupons of nearby stores. And there’s evidence that people are willing to make the trade of convenience over privacy. Perhaps users aren’t afraid to let apps see where they go because of so-called security fatigue–it’s just too much to think about; maybe they trust certain brands to use their data judiciously; or they think they have nothing to hide.
Many of the apps asking for location data have no malintent, simply using it to enable features that can’t otherwise exist. Starbucks notifies you when you’re near one of their stores; Credit card companies offer fraud protection by making sure you’re present where your credit card is being used. Perhaps most of the time, your data is used for nothing but the advertised features and never leaves your phone. Or the collection is transparently communicated and you’re given a chance to delete it, as is the case with the timeline data Google collects. There are instances, however where historical collection is thinly veiled under the guise of real-time, functional features.
As it becomes easier to put this data to use for better customer engagement, personalization, customer segmentation, ad-targeting, per-user dynamic pricing, and more, we suspect this practice will become more and more common. The trend will create an explosion of this new type of data, give data scientists more powerful tools, and along with it, more responsibility.

Google allows users to view their timeline in detail and delete data retroactively

iOS stores frequently visited locations on device and uses them for ad-targeting and other purposes.

It’s important to think about the opportunities and dangers this presents. If companies, large and small, are asking for location data, we should consider whether it can reveal sensitive information about us that we otherwise wouldn’t consent to share. What are we actually sharing and what are we getting out of it? Is it a stretch to worry about mere coordinates data being used to compromise our privacy? On the other hand, as a company, how can we use location data to build good enough features to make it a fair exchange? Finally, what are big changes coming to the space in the near future? These are questions we’d like to answer here.

Signals Hidden in the Noise

Location data in its simplest form is a continuous stream of latitude and longitude pairs, sent from users’ devices at regular intervals. It’s determined by a phone’s GPS sensor under open sky, or otherwise collected by triangulation through fingerprinting a set of nearby cell towers, wifi access points, or bluetooth-emitting devices that are known to correspond to fixed coordinates. It has properties that make it amenable to analysis: It’s structured, relatively homogenous across users, and usually reliable and not misrepresented like volunteered data can be. Of course, that doesn’t mean that extracting meaningful, human-interpretable signals from the data is easy work. Years ago, we had the opportunity of working in data science teams with access to geotagged data from millions of users, but this data was virtually unusable for many of the use-cases we dreamt up. The cost of making sense of the data was simply too high. Raw data had to be massaged to remove outliers–there can be lots of these, especially when indoors–and the data had to then be accurately clustered, and subsequently turned into actual addresses using a process called reverse geocoding. That way, a series of 50 coordinates can be translated to say, the specific path between a coffee shop and a restaurant and the duration of the dwell time at each location. On top of the difficulties in analysis, sampling the data in the first place is hard: using the GPS chip can drain a phone’s battery–a pet peeve that can lead to your app being deleted.

APIs to the Rescue

Given the daunting challenges of making sense of location data, it may seem as though we shouldn’t be worried or excited about its power in the right or wrong hands. However, new tools are beginning to change that. Throughout the past decade, talented teams have built solutions that power their own location products and are also being made available to other companies large and small.
Let’s go back in time for a minute. You may remember Google’s Latitude app, one of the first location tracking apps with a large audience. When Latitude first launched in 2009, it used a significant amount of battery power for its data collection, and suffered from terribly inaccurate readings, which it failed to correct. It meant the app would, at times, see the user jump to the middle of a body of water and back in a matter of seconds. Through the years, however, these issues were ironed out by improvements in hardware, mobile operating systems, and algorithms to better analyze the data. GPS readings and wifi triangulation became more accurate. To save on power, operating systems started caching location readings and sharing them amongst apps, and apps learned to sample location data only when necessary, for instance remaining idle if the user were stationary.
These improvements have allowed companies to utilize location data to build reliable products and roll them out to a vast number of users. For instance, Google Maps now uses this data to offer real-time and historical foot traffic data for millions of businesses. Foursquare notifies you to automatically check in after detecting the venue you’re visiting. Now, luckily for data scientists like us, these improvements are being rolled out as APIs for outside developers, as well, taking the burden of reinventing the wheel off our shoulders.

Google Maps, for instance, has for a while, provided access to its Geocoding and Places APIs, which allow developers to easily turn coordinates into addresses and business names, and vice versa. Its Roads API enables developers to turn a series of coordinates into a path by snapping them onto existing roads. The APIs are used in many popular apps.

Google’s Roads API determines a user’s driving path from a series of coordinates

A much more interesting development is the launch of Foursquare’s Pilgrim SDK last year as part of Foursquare’s pivot towards B2B products. The mobile SDK is built on top of Foursquare’s years of experience in building its own apps that rely heavily on location understanding. When integrated into an 3rd party app, the SDK gives it a chance to react each time the app user enters or exits their home or workplace–which are inferred by the SDK–or visits any of the 105M+ public places in Foursquare’s database.

3rd party apps using the Foursquare Pilgrim SDK can use visit data to segment users. Image from Foursquare.com

New GPS Chip Redefining Accuracy

Despite these remarkable tools, there are still significant shortcomings in current hardware that make it hard to pinpoint users reliably, especially in dense urban settings and indoors. In these situations, which of course make up a large portion of a user’s day, even Foursquare and Google’s data-wrangling often don’t help. Apps may notify us to rate our experience at a Nike store while we’re actually grabbing ice-cream next-door. They most likely won’t be able to tell if we’re at a hotel bar or just checking in at the lobby nearby.
But that’s also about to change. In the coming years, we should brace ourselves for being pushed into an exciting future where those errors will virtually disappear and our apps will become extremely context-aware.
Let’s start with developments in satellite geolocation. For years, improvements in smartphone geolocation accuracy have been elusive, especially in urban settings where it’s possible for satellite signals to reflect off tall buildings and confuse our phones. Custom receivers that use ground base-stations for geolocation have been able to achieve very high accuracy numbers, but according to the government, the accuracy of GPS with current smartphone chips has remained at around 5 meters under open sky. However, in 2017, a new chip was announced by Broadcom that does much better. It uses a new radio frequency named L5, in addition to the L1 frequency that our phones currently use, allowing it to achieve an accuracy of 30cm–a 15x improvement. The chip is reported to have been included in the designs of several phones slated for release in 2018.

The additional signal also allows for significantly better accuracy in cities’ concrete canyons, due to its ability to distinguish between signals arriving directly from satellites and those bounced off of buildings around the device. And this is only the latest development in a global push for centimeter-level navigation accuracy. You can read more about the technical details of how the chip works and why using the L5 frequency helps here.

Has the Time for Beacons Finally Come?

Of course, improvements in satellite-aided navigation can only help when the user’s phone can see enough satellites in the sky, not when the device is carried indoors. The most common way for smartphones to reliably locate themselves indoors is through sensing Bluetooth Low Energy signals emitted by bluetooth beacons–inexpensive chips installed in public spaces. This is typically achieved using the so-called iBeacon protocol. The protocol was introduced by Apple in 2013 and allows for devices to constantly listen for beacons while minimally impacting battery life. The beacons repeatedly advertise a unique iBeacon identifier and the device uses them to detect the beacons in its vicinity, and by knowing their locations, approximately locate itself.
In 2013, the launch of iBeacon caused a wave of excitement and speculation about the possibilities of indoor location. Many retailers, airports, and museums, experimented with the technology, typically using it to push simple coupons or notifications to their users. Despite this initial traction, beacons never really took off as anticipated. Multiple shortcomings led to their limited adoption. Firstly, earlier beacon hardware had to be maintained after installation: their firmware had to be updated, their batteries had to be monitored and replaced, and since they didn’t connect to wifi, all of this had to be done by a person approaching each beacon phone in hand–an unreasonably costly process. They also simply didn’t offer many of the features that brought businesses real value: They could only be used if a business had convinced their customers to install their native mobile app, background tracking was limited (only a limited number of beacons could wake an application), and they had to densely cover a building to be used for navigation, which meant footing a large bill to install and maintain them.
Many of these problems have by now been solved. Several large beacon manufacturers, most notably the Polish company Estimote, have created beacons that are reasonably priced, yet reliable, and allow for remote monitoring and firmware updates through a bluetooth mesh network. They also allow for much longer replacement cycles, and precise location tracking using far fewer beacons. Their offerings drastically reduce maintenance costs and offer much anticipated features like background tracking and automatic mapping of indoor spaces.
In the past few years, there have also been signs of a decisive push for BLE beacons by Google, which has introduced its own Physical Web and EddyStone projects, and is leading the way in developing Web Bluetooth, all with the aim of allowing interaction with BLE devices without the burden of downloading an app. The new capabilities and lower costs may finally tip the balance in favor of beacons and we may see them enter businesses like hotels, airports, museums, and especially retail stores. The new capabilities might, for instance, help retailers in their fight to stay relevant: It could enable them to provide better recommendations to customers by analyzing browsing patterns in stores. They could use it to optimize product shelf placement, staff management, or to provide wayfinding solution to items in their stores.

Estimote’s iOS and Android SDKs allow for precise background monitoring for the first time.
Image from estimote.com

Show the App Where You Are

Only time will tell if iBeacon will finally deliver on its initial promise. But now, there’s an even more interesting and promising technology on the block: indoor location powered by visual signals. For a moment, let’s think about how we humans figure out where we are. Suppose you worked in a large office building for a long while. Now, imagine you were blind-folded and brought to a random location in your workplace. After looking around, you’d probably be able to tell where you are in a matter of seconds. Enter VPS. In a developer’s conference last May, Google revealed its Virtual Positioning Service, a successor to its Tango 3D project, which aims to enable devices to precisely locate themselves indoors down to a few centimeters. The tech was described as being “kind of like GPS”, but using “distinct visual features in the environment” instead of satellites. It’s likely that this type of location sensing would improve rapidly, because it can be used not only for smartphones, but also by autonomous robots or drones that want to know their exact position within the confines of concrete walls.

Google tweeted about its new VPS project during Google IO

Regardless of which technology ends up being the most useful and widely adopted, it’s clear that data scientists are seeing an explosion in the availability and usefulness of location data, and like always, it’s up to us to take advantage of it to benefit end-users while doing all we can to shield them from the danger of sacrificing their privacy.

Keshav Dhandhania is a cofounder of Compose Labs (commonlounge.com) and has spoken on GANs at international conferences including DataSciCon.Tech, Atlanta and DataHack Summit, Bangaluru, India. He did his masters in Artificial Intelligence from MIT, and his research focused on natural language processing, and before that, computer vision and recommendation systems.

Posts Filed Under: