Sep 29, 2015 · post

Machine Learning in Retail: Consumer Privacy Implications


In our last retail post, we explained how emerging sensor technologies are changing data science for brick and mortar. Many companies are working to retro-fit physical stores with capabilities originally developed for ecommerce. Without adding any new sensors or tags, image recognition company Blippar provides APIs that digitize physical products to create an in-app experience around them. Farfetch, which initially brought an exclusive network of luxury brick & mortar shops online, recently announced that it will soon provide in-store analytics.

In parallel, digital companies continue to seek out physical spaces to augment their customer experience. Warby Parker and Rent the Runway have physical stores, and Gilt and Etsy are exploring new models to engage consumers offline (Gilt has a private retail space in their corporate headquarters and Etsy launched an app to inform shoppers of sellers’ nearby products). These digital companies were built upon testing every detail affecting their website performance. As tracking migrates to physical retail, it’s important that retailers consider the consumer experience risks associated with new technologies as they explore the benefits.

“When you look at something on the web, you get ads that pop up and follow you around - companies like that have much better advantage over brick-and-mortar retailers, and so they’re under more pressure to equalise the playing field.” Kevin Kearns, ShopperTrak

Reasonable Expectations for Privacy

Before retailers invest resources to model consumer data, they should ask whether users would respond better to less personalized advertising experiences. There’s a difference between personal preferences and personal traits. A personalized experience that tends to an individual’s preferences can feel bespoke, high-service, and attentive to the consumer’s needs. A personalized experience generated by traits or demographics can feel more like differential treatment, where the experience is triggered by a collection of “contexts” sensed about the consumer.

Adobe and Page Fair point out that there are nearly 200 million active users of ad blocking software, costing advertisers about $22 billion dollars a year, with 48% growth in the US over last year. In addition, in a 2000-person survey from Altimeter Group, all age groups expressed concern about data collected in public, with specific discomfort around use of body data. Retailers should keep consumer tolerance in mind as new sensors and software makes it easier to capture extremely detailed profiles of in-store shoppers.

Technology always outpaces policy and law, and to date retailers lack precedent for providing consumers with privacy policies articulating passive data collection practices. Some tech companies, like Prism Skylabs sidestep privacy concerns by anonymizing personally identifiable data, giving “a sense of what’s happening in a store without having to be there physically counting customers or examining shelves.” Neiman Marcus, alternatively, asks for explicit consent to record gender, age, and body type data when testing their Memomi smart mirrors.

After a decade of digital ecommerce, many people’s expectations regarding online privacy have shifted to expect monitoring by retailers and advertisers. But the same does not hold in physical retail, where past methods of sales tracking amounted to conversations with sales associates or, at most, wifi tracking using mobile phones. But emerging technologies can do far more than track location: they can add sophisticated “metadata” about a consumer’s age, body measurements, and browsing habits.

To think through privacy expectations, researchers Soultani and Bankston developed a model that relates individual expectations of privacy to technology costs. “If the cost of surveillance using the new technique is an order of magnitude (ten times) less than the cost of the surveillance without using the new technique,” they claim, “the new technique violates a reasonable expectation of privacy.” So what are the economics of the new tracking tools? Nomi’s Brickstream sensor costs as low as $20/month for a plan of 25 in-store sensors, amounting to at least $500/month (pricing may vary over time). If tracking consumer preferences - or bodies - were left to a sales associate making $28K/year, monthly costs would amount to at least $2333, notwithstanding the fact that a single person’s attention cannot scale across the customer base. The cost to track a single person quickly falls near $0 for a sensor that scales across the customer base. If we trust Soultani and Bankston’s model, the prices of these new tools trigger a re-evaluation of best practices for policy around privacy.

Technology Limitations Beyond Privacy

Even if consumers tolerate passive data collection, algorithmic pitfalls also impact user experience. Many machine learning systems learn using training datasets, which impact the predictions they make about new data in the wild. As boyd and Crawford point out, “a data set may have millions of pieces of data, but this does not mean it’s random or representative.” Imagine generating a body-specific ad, or even matching existing inventory to a consumer’s body measurements. What happens if there is not enough data to generate a realistic representation of an individual? If nothing in a store’s inventory fits a customer’s body? If a 15-year-old girl’s body best matches 40-year-old women in your database? Would automated suggestions pull up inventory more appropriate for older customers, alienating her in front of friends with bodies more like the average 15-year-old?

A careful, opt-in experience does have the power to create experiences that were much harder to scale in the past. Companies like Slyce, Cortexica, Deepomatic, and Donde Fashion are harnessing deep learning to augment the browsing experience. During New York Fashion Week this month, Cortexica pointed out the challenge of finding the right item when a shopper brings a photo of what he/she wants that might not exactly match inventory. Without requiring an item tag, their software analyzes photo data to suggest items with similar colors and textures, within or across detected categories. Here, software augments customer preferences rather than prescribing preferences from passive data collection about multiple consumers. Shopping is all the more fun when a photo of a sunflower is game for inspiration like a street style star.


Read more

Oct 6, 2015 · post
Sep 24, 2015 · post

Latest posts

Nov 15, 2022 · newsletter

CFFL November Newsletter

November 2022 Perhaps November conjures thoughts of holiday feasts and festivities, but for us, it’s the perfect time to chew the fat about machine learning! Make room on your plate for a peek behind the scenes into our current research on harnessing synthetic image generation to improve classification tasks. And, as usual, we reflect on our favorite reads of the month. New Research! In the first half of this year, we focused on natural language processing with our Text Style Transfer blog series. more
Nov 14, 2022 · post

Implementing CycleGAN

by Michael Gallaspy · Introduction This post documents the first part of a research effort to quantify the impact of synthetic data augmentation in training a deep learning model for detecting manufacturing defects on steel surfaces. We chose to generate synthetic data using CycleGAN,1 an architecture involving several networks that jointly learn a mapping between two image domains from unpaired examples (I’ll elaborate below). Research from recent years has demonstrated improvement on tasks like defect detection2 and image segmentation3 by augmenting real image data sets with synthetic data, since deep learning algorithms require massive amounts of data, and data collection can easily become a bottleneck. more
Oct 20, 2022 · newsletter

CFFL October Newsletter

October 2022 We’ve got another action-packed newsletter for October! Highlights this month include the re-release of a classic CFFL research report, an example-heavy tutorial on Dask for distributed ML, and our picks for the best reads of the month. Open Data Science Conference Cloudera Fast Forward Labs will be at ODSC West near San Fransisco on November 1st-3rd, 2022! If you’ll be in the Bay Area, don’t miss Andrew and Melanie who will be presenting our recent research on Neutralizing Subjectivity Bias with HuggingFace Transformers. more
Sep 21, 2022 · newsletter

CFFL September Newsletter

September 2022 Welcome to the September edition of the Cloudera Fast Forward Labs newsletter. This month we’re talking about ethics and we have all kinds of goodies to share including the final installment of our Text Style Transfer series and a couple of offerings from our newest research engineer. Throw in some choice must-reads and an ASR demo, and you’ve got yourself an action-packed newsletter! New Research! Ethical Considerations When Designing an NLG System In the final post of our blog series on Text Style Transfer, we discuss some ethical considerations when working with natural language generation systems, and describe the design of our prototype application: Exploring Intelligent Writing Assistance. more
Sep 8, 2022 · post

Thought experiment: Human-centric machine learning for comic book creation

by Michael Gallaspy · This post has a companion piece: Ethics Sheet for AI-assisted Comic Book Art Generation I want to make a comic book. Actually, I want to make tools for making comic books. See, the problem is, I can’t draw too good. I mean, I’m working on it. Check out these self portraits drawn 6 months apart: Left: “Sad Face”. February 2022. Right: “Eyyyy”. August 2022. But I have a long way to go until my illustrations would be considered professional quality, notwithstanding the time it would take me to develop the many other skills needed for making comic books. more
Aug 18, 2022 · newsletter

CFFL August Newsletter

August 2022 Welcome to the August edition of the Cloudera Fast Forward Labs newsletter. This month we’re thrilled to introduce a new member of the FFL team, share TWO new applied machine learning prototypes we’ve built, and, as always, offer up some intriguing reads. New Research Engineer! If you’re a regular reader of our newsletter, you likely noticed that we’ve been searching for new research engineers to join the Cloudera Fast Forward Labs team. more

Popular posts

Oct 30, 2019 · newsletter
Exciting Applications of Graph Neural Networks
Nov 14, 2018 · post
Federated learning: distributed machine learning with data locality and privacy
Apr 10, 2018 · post
PyTorch for Recommenders 101
Oct 4, 2017 · post
First Look: Using Three.js for 2D Data Visualization
Aug 22, 2016 · whitepaper
Under the Hood of the Variational Autoencoder (in Prose and Code)
Feb 24, 2016 · post
"Hello world" in Keras (or, Scikit-learn versus Keras)


In-depth guides to specific machine learning capabilities


Machine learning prototypes and interactive notebooks

ASR with Whisper

Explore the capabilities of OpenAI's Whisper for automatic speech recognition by creating your own voice recordings!


A usable library for question answering on large datasets.

Explain BERT for Question Answering Models

Tensorflow 2.0 notebook to explain and visualize a HuggingFace BERT for Question Answering model.

NLP for Question Answering

Ongoing posts and code documenting the process of building a question answering model.

Cloudera Fast Forward Labs

Making the recently possible useful.

Cloudera Fast Forward Labs is an applied machine learning research group. Our mission is to empower enterprise data science practitioners to apply emergent academic research to production machine learning use cases in practical and socially responsible ways, while also driving innovation through the Cloudera ecosystem. Our team brings thoughtful, creative, and diverse perspectives to deeply researched work. In this way, we strive to help organizations make the most of their ML investment as well as educate and inspire the broader machine learning and data science community.

Cloudera   Blog   Twitter

©2022 Cloudera, Inc. All rights reserved.