Multinational food and retail delivery e-commerce platform Glovo had a big problem: existing users were creating new accounts to abuse promotions or hide their histories of seeking refunds. These users made it hard to distinguish between true new customers and fraudsters, and they cost the company money it wanted to spend on new customers. 

Duplicate accounts aren’t the only problem Glovo faces — with operations in 23 countries across Europe, Asia, and Africa, Glovo deals with myriad fraud issues that require sophisticated tools to combat. 

Fingerprint’s Senior Developer Evangelist Keshia Rose hosted a webinar with Glovo’s Fraud Manager Bruno Pizzani to talk about how Glovo uses Fingerprint’s device intelligence to help their analysts and train their machine learning models fend off bot attacks, friendly fraud, account takeovers, and more. 

Here are some key takeaways from that webinar. 

Types of fraud Glovo faces

Glovo connects customers with a network of couriers to order food, groceries, medications, and other goods from nearby local businesses. Per Pizzani, the main type of fraud Glovo contends with is policy abuse, where users exploit loopholes in refund and promotions policies to avoid paying for delivery fees. Their 12-person fraud analysis team also handles bots, friendly fraud, and other types of deceptive practices on a regular basis. 

Some of the types of fraud Glovo sees most commonly include:

• Policy abuse:
  • Incentives fraud: Existing customers make new accounts to take advantage of coupons or promotions for first-time users (such as free delivery on a first order).  
  • Refund fraud: Users create duplicate accounts to obscure a pattern of taking advantage of refund policies to return items after using them.
• Credit card fraud: Fraudsters use stolen credit cards to purchase high-value retail goods, such as electronics, that can be resold. 
• Bots: 
  • SMS pumping schemes: Fraudsters pocket fees that Glovo pays to send verification texts to premium-rate phone numbers in certain countries. 
  • Review fraud: Local businesses that sell through Glovo’s platform create fake reviews to boost their placement in the app.
• Account takeover: Fraudsters gain unauthorized access to a user account, sometimes through a phishing attack, and place orders with it.
• Friendly fraud: Users file chargebacks for legitimate purchases after experiencing buyer’s remorse or simply because they’re acting in bad faith. 
• Courier fraud: Customers and couriers collude to place fake orders to manipulate the rates a courier is paid. 

The impacts of fraud at Glovo go beyond the direct financial costs. For example, customer acquisition costs are driven up by people who abuse new-user promotions and bots that spam the user sign-up system as part of SMS pumping schemes. In addition to draining marketing budgets, these types of fraud create inaccurate data about who the company’s new users are.

Glovo’s challenges when it comes to fraud

While the fraud areas Glovo deals with aren’t unique to a multinational e-commerce business, what’s distinctive is the volume and variety. It’s tough to have eyes everywhere, Pizzani says, which has made Fingerprint valuable because it can be used in all of Glovo’s fraud domains. Here are some of the major challenges he and his team deal with at Glovo.  

Global complexities 

Because Glovo operates in 23 countries, it deals with many unique regional risks related to payment infrastructure — in some countries, for example, Glovo accepts cash payments and mobile payments. This makes particular market expertise crucial for handling fraud risks. 

Data gaps and observability

Glovo doesn’t always have enough data to “one-size a problem,” as Pizzani puts it, or to detect trends in real time. The more data they can gather, the more informed their decisions can be. Device intelligence and behavioral data are two areas where Glovo has gotten a lot of value from Fingerprint to gain information about user history, Pizzani says.

Evolving fraud tactics

The falling cost of many fraud-enabling technologies can make it hard to keep up with malicious actors. Getting a new SIM card or a new phone number has become so easy, for example, it’s cheap for users to sign up for new accounts. Fraudsters are using AI to create fake IDs for opening new digital credit cards, and new software is making it easier than ever to bypass CAPTCHAs and other security measures.

How Glovo uses Fingerprint for fraud prevention

Fingerprint’s device intelligence enables Glovo to identify past offenders, suss out bots, and smooth the user experience for legitimate customers. Glovo also uses Fingerprint data to improve machine learning models looking for fraud patterns in network data. Here are some specific areas where Fingerprint is helping Glovo. 

Spotting past offenders

Fingerprint’s device intelligence helps the company spot users who are creating duplicate accounts or fake accounts, which is useful in fighting incentive fraud, refund fraud, credit card fraud, and customer-courier collusion. Glovo uses Fingerprint to detect whether a courier and customer are using the same device, for example. 

Glovo also uses Fingerprint to aggregate data about unique devices, such as whether they have placed many orders in a short time period. Fingerprint’s IP Geolocation Smart Signal enables Glovo to spot suspicious checkouts, such as when a user in one country inputs a credit card from another country. 

Bot detection 

Fingerprint’s Bot Detection Smart Signal helped identify specific frameworks being used for SMS pumping schemes. This particular feature has reduced Glovo’s SMS costs by 20-30%, and has helped identify bots that fraudsters were using to create fake reviews for businesses that sell on Glovo. 

Recognizing trusted users 

Glovo’s issues with duplicate accounts hampered the marketing team’s understanding of who the company’s new customers were. Fingerprint helped Glovo figure out whether one user account was linked to another and whether that user had placed orders in the past. 

Glovo also uses device intelligence to improve the user experience for trusted customers, which Pizzani says has been one of the most valuable parts of using Fingerprint. If Glovo recognizes a returning user’s device, they can make the buying process smoother by removing MFA requirements. On the other end, they can more effectively prevent account takeovers by using additional security measures for unrecognized devices. 

Key takeaways: Device intelligence key part of fraud prevention for Glovo

Fingerprint’s device intelligence has given Glovo crucial insights into customer behavior. Glovo is using that data to make it harder for past offenders to repeat and streamline the user experience for legitimate customers.  

If you missed the webinar and want to watch the full video, you can find it here.