Not what I ordered: Harassment, fraud, and fake reviews in food delivery services

In 2022, a total of 1.85 billion people used meal delivery services worldwide, a number which is expected to reach 2.64 billion by 2027: adding up to a 3.5X increase in just one decade. Accordingly, the market size for food delivery has also ballooned, reaching $150 billion by 2021. However, as with any successful e-commerce or marketplace business, that success comes at a cost, with a high potential for platform misuse and abuse that can impact user safety – both on- and offline, and negatively impact business outcomes.

For food delivery services, there is no time like the present to understand the broad range of threats to the business and define proactive strategies to avoid harming users, sellers, drivers, and themselves.

The Trust & Safety Challenges of Food Delivery Services

Like many innovative services before them, food delivery applications are finding that intentional platform misuse is an inevitable operational risk. Threat actors are highly creative and skilled at detecting and capitalizing on gaps in policy and platform features to cause harm. In food delivery, platform risks center around three focal points: reviews, in-app chats, and transactions.

1-Star: Review violations in food delivery services

Like any transactional service online, reviews are a powerful tool for vendors and customers. In fact, reviews influence 88% of consumers in discovering a local business, and only 9% would consider engaging with a lower-ranked business. While this is true for e-commerce, similar stats can be found in the food industry, where 94% of US consumers make dining decisions based on reviews.

But what happens when those reviews are misused? A significant cause of concern among all marketplaces and food delivery apps is the intentional misuse of the feature to promote fake reviews or otherwise harmful and violative reviews.

Malicious Reviews

Like any type of user-generated content field, the review box is exposed to a wide range of abusive activity. Ranging from relatively simple profanity to more complex violations like the promotion of nudity and hate speech – violative reviews can leave users with a sour taste and a loss of trust, which may deter them from using the service in the future.

Detecting these violations poses a unique challenge for trust & safety teams, as the content may be shared in multiple languages (depending on the country in which the platform operates or the user’s language), may contain violative emojis or nuanced language that makes it hard to assess whether or not the item is violative, and, specifically when dealing with cultural variations and unique dish names, it may be hard to differentiate between legitimate and offensive language. Take, for example, the popular British steamed custard dish – the Spotted Dick: distinguishing its name from adult content presents a challenge for teams moderating across cultures.

Fake Reviews

In any marketplace or e-commerce business, fake reviews, used to boost one’s business or harm a competitor, are a cause of concern. Both positive and negative fake reviews harm the user experience: 81% of consumers will not buy from a business after discovering that its reviews are fake. Generally, fake reviews fall under two categories:

1. Fake reviews by fake accounts:
   These are usually acquired through dedicated vendors, these are relatively easy to identify as they may contain copied content, and the accounts promoting them may display classic indicators of bot activity. However, the influx of new generative AI tools to the market now means that these will be harder to detect as variations of text can easily be created in nearly-human English.
2. Fake reviews by real accounts:
   Trickier to identify, these may be posted by real individuals who are either associated with a business (employees, family members) or not (acquired reviews via influencers or otherwise). These are harder to identify because they involve real text, produced by real people.

The customer is always right? Harassment in food delivery chat

A unique challenge of food delivery services is the existence and management of multiple communication channels involving distinct stakeholders. The service manages the communication between end-users (customers) and drivers, customers and businesses, businesses and drivers, and each entity with support agents. This four-way communication stream can create opportunities for friction and harassment – ultimately creating a trust & safety concern for food delivery businesses.

When communication between drivers and customers becomes heated, the potential risk is high on all fronts. Given the online-offline nature of food delivery, the risk can go beyond the uncomfortable engagement and likely deterrence from using the app and into actual physical risk for either party. Even when benign, these situations are likely to develop into a PR nightmare for the food delivery business, as these stories are repeatedly featured in the media.

Like violative reviews, the challenge of in-app harassment is complicated for many reasons. Multiple languages are a given, as apps that expand into new geographic regions are exposed across the languages used in the new country (including variations and slang). Similarly, the existence of nuanced language and in-group terminology means that unskilled moderators – even those fluent in the moderated language, may not catch violations – especially if the keyword lists they moderate are not up-to-date. Finally, these linguistic and keyword challenges are further compounded by the fact that in-app communications are live, meaning that moderation must be swift, close to real-time, to stop the situation from escalating.

Cancel my order: Fraud in food delivery

The classic case for trust & safety in online marketplaces and other e-commerce platforms, which is most closely associated with the business’s financial outcomes, is fraud. Like other marketplaces, food delivery platforms generally deal with two types of risk: on the buyer and seller sides.

Seller-side fraud

Seller-side fraud consists of any fraudulent activities involving the sellers – or those impersonating sellers. In food delivery, these could include:

1. Merchant impersonation: creating a fake (or misleading) merchant account to conduct any kind of fraud, including:
2. False advertising: listing and selling items that are misrepresented or exaggerated – leading to customer dissatisfaction, complaints, and refund requests.
3. Fake listings: taking orders for items the merchant does not intend to fulfill.
4. Collusion: where fraudulent sellers and customers coordinate a fake transaction to capitalize on promotional offers, boost seller rating, or make purchases using stolen credit cards.
5. Driver fraud: unique to delivery services, this includes situations where drivers claim to have delivered an order when in fact, they did not.

Buyer-side fraud

Fraud is also prevalent on the buyer side and includes:

1. Account takeovers: when user account details are leaked and accessed by fraudulent actors, who then gain access to the user’s financial information to make transactions on their behalf.
2. Fake accounts: used to take advantage of new customer promotions or to write fake reviews.
3. Payment fraud: using stolen credit card information to make a purchase using the account.
4. False claims (transaction or friendly fraud): this happens when users complain about a delivered item – either claiming it did not arrive or that it was not up to standards – to receive a refund. This can implicate all other parties – including the driver, restaurant, and delivery service.

The solution: Food delivery T&S

This article has outlined three significant areas of (trust & safety) concern for food delivery businesses: violative reviews, in-app harassment, and fraud. These complex issues carry risks for all involved parties: legitimate merchants who may lose business to fraud, drivers who may be implicated by fraud and harassment, users who are exposed to fraud, harmful engagements, and potential real-life dangers, and finally – the platform itself which could lose credibility, resulting in decreased usage and loss of business.

To tackle these harms, food delivery businesses should consider trust & safety as another risk prevention strategy: defining their policies for in-app communication and investing in dedicated solutions.

ActiveFence’s solution for food delivery services covers all of these abuses: our content moderation platform automatically surfaces harmful content in reviews and in-app communications across a variety of abuse areas and over 100 languages, allowing teams to quickly identify and take action against threats – enabling action in near-real-time. Additionally, our threat intelligence investigations and red team solutions provide deep insights into new forms of fraud, allowing food delivery businesses to take proactive measures to stop this abuse from impacting their business.

Click here to learn more about our offering for food delivery services.