BNP Paribas Cardif: Cultivating a safer online environment for kids
Giacomo Losa and Roberta Avanzini, from the Research & Development team at BNP Paribas Cardif (Italy) describe how their Bodyguard tool is protecting children from the dangers of online life.
Interviewees
What led to the creation of Bodyguard for Families?
As we are positioned as a “preventive insurer”, the idea is to prevent cyberbullying on social media, removing comments even before the victim can see it and informing parents that something bad happened. It is perfectly in line with our mission and we conceived of this innovation on the surge of online exposure. Benefitting from a partnership with Plug&Play , we selected Bodyguard during one of their annual batches in order to develop a pilot project.
How does the app work? How do you adapt it to new threats and dangerous content? And in terms of child protection, how do you evaluate the app's results? How much potentially harmful content does it filter out?
The parent subscribes to the service on Bodyguard’s website. The parent then receives an email with the link to download a free app and connects his/her child/children’s social media profiles to Bodyguard (it can be on the parent’s phone or on the children’s phone). Bodyguard works in the background; the parent will never see any content from the children’s dashboard. When someone publishes offensive content on the social media dashboard, the algorithm deletes the comment in less than one second, this works even if the comment is made by the children, not only received. At the same time an email is sent to the parent alerting them that something bad happened (if the children are below the age of 13, the parent will also see the offensive content).
The algorithm is continuously updated with new terms, abbreviations, etc. and there is also a “human” check. It also works the other way around, alerting parents when their children are bullying someone.
The results are based on KPI’s such as the number of comments analyzed (more than 20k), detection rate (97%), avg rate on stores and reviews (5 stars avg), etc.
We talk a lot about successes but we also know that failures are very rich in lessons. What advice would you give to someone who is just starting to design a similar product? What mistakes have you made or almost made? What are the pitfalls to avoid?
Our suggestion is to start small, test often and fast, and have a clear and measurable objective to reach.
In our case we stumbled upon an unexpected delay. Willing to test the proposition in order to understand the potential reach, we were faster than insurance product readiness, with distribution channels requesting an all-in product. So, we are now trying to bridge this gap with pure marketing initiatives, which, on the other side, gives us the worthy chance to train the algorithm for a more and more powerful moderation tool. Heart of the solution is language and once organized, Cardif and Bodyguard would have a very distinctive solution.
What are your objectives or KPIs? And in terms of downloads and customer satisfaction, what are your results? And what is the average subscription duration?
Our objective is to include this solution into a full cyberinsurance product. Customer satisfaction is very high and it is measurable thanks to the reviews on the app stores. Downloads are also good considering that we’ve not made any bold communication on the project until now. There are only two ways to subscribe: 1 month or 6 months.
Do you offer or plan to offer legal or tech support for kids who are victims of digital harassment? And what other products do you offer for kids/families?
This will be part of the full insurance service.
The app also exists for individuals and for businesses. What are the differences between the three versions?
The algorithm behind works in the same way. It is able to recognize offensive content organized in many categories of insults and bad words, including also emojis and abbreviations. The main differences are that the individual solution is completely free and is downloaded by a person for him/herself in order to protect his/her own social media channels from insults and offensive content but without the alert to the parent/tutor.
For the second option that is “for families”, you have to pay a small monthly fee. While for the business solution there is a subscription paid by the company to Bodyguard and they’re able to protect all the social media accounts of the company, the accounts of the key figures in the company, or for example in sports companies they protect all the players’ accounts. The company also has access to a backend dashboard to monitor all the profiles protected.
The app protects kids in real time from hate comments received on Twitter, Instagram, YouTube and Twitch.
Leverage community expertise to redefine finance
Our communities cover diverse topics such as digital transformation, SME finance, or Embedded insurance, providing a platform to learn from industry experts and peers.