3PM Solutions: Using data to detect, remove, and protect against online marketplace fraud
About 3PM
3PM Solutions is an eCommerce brand analytics and protection company focused on marketplace websites like Amazon, eBay, and Walmart.com. 3PM Solutions' cloud-based platform offers solutions that help manufacturers, retailers, and brands to solve complex problems impacting revenues. The company’s unique B2B2C Chrome extension, ReconBob, improves and protects consumers’ online shopping experience.
Tell us your challenge. We're here to help.
Contact us3PM Solutions empowers global brands to manage, protect, and grow revenue on Amazon, eBay, and Walmart.com by leveraging Google Cloud services such as Cloud Bigtable and Dataproc.
Google Cloud results
- Reduces infrastructure maintenance from 5 hours/week to 0 hours
- Analyzes 160 million customer reviews of more than 2 million sellers in less than 4 hours
- Scales to meet 600% increase in new customers within 1 year without adding headcount
- Retrieves all relevant brand data from more than 1 billion listings in a matter of seconds
Launches new customers in minutes versus days
During the first quarter of the year, third-party sellers accounted for 52% of items sold on Amazon, according to Statista—a percentage that has slowly increased over the past five years. While many of these sellers are legitimate, too many of them sell counterfeited or misrepresented products to unsuspecting consumers.
The practice has caught the eye of federal investigators. Recently, CBS News reported that investigators purchased 47 products from third-party sellers on Amazon, Walmart.com, eBay, and other sites. They found that 20 of the products, including cosmetics, shoes, and electronics, were counterfeit. All the items were advertised as new. And the sellers had average customer ratings above 90%.
Marketplace business model growth may be fueled by limited seller oversight. But, by providing unregulated third-party sellers direct access to consumers, marketplaces have created a potentially massive problem for consumers and businesses alike. Consumers lose out when they receive a substandard branded product that looks legitimate and may even be harmful. Counterfeited or misrepresented products can damage the brand reputation of manufacturers, retailers, and other companies, and cause millions of dollars in lost sales.
The marketplace websites have vowed to address the problem—even creating a place for companies to register their brands. However, Rob Dunkel, President and Founder of 3PM Solutions says, “relying on the marketplace operator to protect your brand’s revenue is like letting the fox guard the hen house.” So how do you effectively stop the bad actors?
3PM Solutions is helping to lead the charge with actionable data and intelligent solutions by leveraging Google Cloud services.
The company developed its cloud-based software platform to aggregate, organize, and analyze unstructured data at scale related to third-party sellers on Amazon, eBay, and Walmart.com. The goal is to give its customers—many of them global brands—transparency and actionable insights into third parties that are selling counterfeits, grey market, or misrepresented versions of their products. 3PM Solutions also developed a free Chrome browser extension, ReconBob, that provides consumers with greater transparency into third-party sellers on Amazon.
3PM Solutions’ mission to help brands gain control over their products on marketplace websites requires making sense of large, unstructured datasets. Initially, the company had a hosted cloud environment with Hbase clusters running on top of Hadoop to process and analyze large datasets. But its engineering team spent too much time manually deploying and maintaining Hadoop clusters.
“With our previous infrastructure, it took about three weeks to set up our Hadoop cluster, and we spent five hours a week on maintenance. It took only a few minutes to get up and running on Google Cloud Bigtable, and we don’t spend any time maintaining it. It just runs.”
—Sergey Belov, Director of Technology, 3PM SolutionsIn hopes of dramatically reducing the maintenance time, 3PM Solutions moved its infrastructure to Google Cloud, with Google Cloud Bigtable serving as its fully managed NoSQL database service.
When Shane Ghiotto, Director of Data Science for 3PM Solutions joined the team, the Python client for Cloud Bigtable provided easy-to-use APIs that helped him evolve from having no knowledge of Google Cloud to reading and writing data efficiently to Cloud Bigtable in less than a day. Cloud Bigtable provides easy access to the data so Shane can focus on the analytics.
3PM Solutions’ previous HBase cluster was compatible with the Cloud Bigtable API, which allowed the company to migrate all its data with no major changes or hassle.
“With our previous infrastructure, it took about three weeks to set up our Hadoop cluster, and we spent five hours a week on maintenance” says Sergey Belov, Director of Technology for 3PM Solutions. “It took only a few minutes to get up and running on Google Cloud Bigtable, and we don’t spend any time maintaining it. It just runs.”
3PM Solutions began using Cloud Bigtable when it was still in beta, and it played a foundational role in helping the company structure its data tables. “We’re saving about 1 billion data points per day, and because of the speed and stability of Cloud Bigtable, we’re planning to increase the volume of data points by 4x or 5x in the next 6 months,” says Sergey.
Identifying and removing frauds
In a recent count, there were approximately 562 million products for sale on Amazon.com. Each product falls into one of a huge variety of categories, and each category has different key attributes that describe products. 3PM Solutions aggregates all this data from Amazon, eBay, and Walmart.com, so that its software can help its customers make sense of it—and more importantly, to get fraudulent product listings and sellers removed.
For example, from the data it aggregates and analyzes, 3PM Solutions has created a database of seller histories that helps it identify frauds. Based on a seller’s track record, customer reviews, and other data it aggregates, 3PM Solutions’ software can identify a seller as fraudulent and notify the online marketplace to have the seller removed.
“We create risk profiles of third-party sellers, so our customers know who is representing their products, how their products and IP are being represented, and their customer’s experience,” says Rob. “When a customer has a poor experience from a counterfeit branded product, the brand may lose a customer, but the marketplace operator may not.”
“With the feedback classification model we’ve built, and by leveraging Google Compute Engine and Cloud Dataproc, we can perform an analysis of more than 160 million customer reviews of more than 2 million sellers in just under 4 hours, which is phenomenal.”
—Shane Ghiotto, Director of Data Science, 3PM SolutionsAnalyzing millions of comments
“Inherently, the data we gather doesn’t come with a natural, easy-to-follow structure,” says Shane. “We aggregate what is essentially a big blob of text data, and we need to determine for each product which category attributes are the most important for our clients.”
3PM Solutions spins up virtual machine clusters as needed in Google Compute Engine and uses Google Cloud Dataproc to index its listing tables, which helps ensure that unstructured text data can be searched easily. Reviews of third-party sellers that customers leave are also imported from marketplace websites and classified into various categories. The categories indicate if a seller has previously sold used products listed as new, counterfeit, or expired products, for instance.
“With the feedback classification model we’ve built, and by leveraging Google Compute Engine and Cloud Dataproc, we can perform an analysis of more than 160 million customer reviews of more than 2 million sellers in just under 4 hours, which is phenomenal,” says Shane.
Protecting brands via images
3PM Solutions uses Cloud Vision API to match images that third-party sellers post, which are compared against licensed images from brands. When there’s a match, 3PM Solutions can report the misuse of the images to the marketplace. Identifying unauthorized use of images is important because some third-party sellers create multiple duplicates of authentic product listings to fool consumers.
3PM Solutions plans to further bolster its image recognition by integrating the open-source machine learning framework TensorFlow, which Google developed, and Cloud TPU, the Google machine learning accelerator for TensorFlow. Deep learning that TensorFlow makes possible will also enable 3PM Solutions to help drive consumers to brand-approved sellers on Amazon or refer them to a brand-approved site off Amazon with a coupon offering.
Scaling in seconds
Demand for 3PM Solutions’ solution is rapidly growing. The company experienced a 600% increase in new customers over the last full fiscal year and was up 50% for 2018 as of June that year. “Because Google Cloud requires so little attention to run, we’ve added all these new customers without having to add headcount,” says Rob.
“Before the move to Google Cloud, launching a demo or a new customer up could take one or two days. Because of Google Compute Engine and Cloud Dataproc, we can quickly pull in all potentially relevant data and organize it in a useful way in a few minutes, which is phenomenal."
—Shane Ghiotto, Director of Data Science, 3PM SolutionsAdditionally, Google Cloud enables 3PM Solutions to quickly onboard new customers. “Before the move to Google Cloud, launching a demo or a new customer up could take one or two days,” says Shane. “Because of Google Compute Engine and Cloud Dataproc, we can quickly pull in all potentially relevant data and organize it in a useful way in a few minutes, which is phenomenal."
The ability to scale quickly to grow its customer base and services is critical for 3PM Solutions. The company is currently leveraging Google App Engine to scale applications in a serverless environment, along with Google Kubernetes Engine to automatically manage cloud resources and operations.
Using App Engine as a backend for ReconBob allows 3PM Solutions to scale its system to tens of millions of users without spending any time on improvements. App Engine also enables 3PM Solutions to concentrate on helping consumers shop safely on Amazon.com. Cloud Bigtable, which is at the heart of its system, is also extremely easy to scale. “We can automatically scale up to 20 Cloud Bigtable cluster nodes from 3 for a specific customer, and then automatically scale down as needed, without our customer seeing any delay,” says Sergey.
Getting peace of mind
The security controls and features integrated into Google Cloud give 3PM Solutions engineers peace of mind and free them up to focus on other priorities. "We don’t really spend any time on security, because we know that the work we do on Google Cloud is highly secure,” says Sergey.
Developing a partnership
The support 3PM Solutions receives from Google has been “extraordinary,” Rob says. “Google engineers check in with our team periodically, and if we have a question, they find the right person to answer it. Our relationship with Google feels like real teamwork.”
Support from Google, coupled with the minimal maintenance requirements of Google Cloud and the advanced technologies that Google offers, all combine to position 3PM Solutions engineers for the future.
“Because we don’t have to spend time on administrative tasks, we can experiment with Google services like natural language processing and machine learning that will help us grow our features,” Rob says. “With Google Cloud, we’ve built and can continue to build a really powerful solution that goes head-to-head with, and exceeds, the competition.”
Tell us your challenge. We're here to help.
Contact us
About 3PM
3PM Solutions is an eCommerce brand analytics and protection company focused on marketplace websites like Amazon, eBay, and Walmart.com. 3PM Solutions' cloud-based platform offers solutions that help manufacturers, retailers, and brands to solve complex problems impacting revenues. The company’s unique B2B2C Chrome extension, ReconBob, improves and protects consumers’ online shopping experience.