2. STOCK
PHOTOGRAPHY
Connecting Global PartnersâŚ
2
Š 2018 Payoneer, Proprietary & Confidential
ROYALTIES
FREELANCING:
TRANSLATION,
DESIGN, DEVELOPMENT
VACATION RENTALS
& TRAVEL
DIGITAL
MARKETING
ECOMMERCE
MARKETPLACES
3. Payoneer Payment Platform â the numbersâŚ
3
Banks & Payment
providers
45+
Million active users4M
Support tickets/ mont190K
Registrations/ month200K
Payments/ month2.8M
Countries200+
Currencies150+
$2B+ in payments/month
4. Description of services
Account
5
Local Bank Transfer
Use Payoneer-issued EUR,
GBP, USD, JPY, AUD, CAD
bank accounts to receive local
payments from buyers
Credit Card Payments
Receive V/MC payments
Wire
International wires in multiple
currencies
In-Network Payments
Receive funds directly from
other Payoneer Account
Holders
Mass Payout Clients
Upwork, Airbnb, and others
instruct that payeeâs Payoneer
Account be funded
Local Bank
Funds delivered to the sellerâs local
bank account
in local currency
Payoneer Disbursement Card
Load funds on to Payoneer
Disbursement Card enabled for
withdrawing cash at local ATM or
use online or at POS
In-Network Payments
Pay other Payoneer Account
holders
Payments to Third Parties
Payments to suppliersâ and other
third partiesâ (eg, tax authorities)
bank accounts
Pay with Payoneer
Enables marketplaces and other
third parties to debit a Payoneer
Account for fees due
Payoneer
Account
Inflows Outflows
21. 22
⢠Higher percentage of fraud detection
⢠Saves time in investigations and query processing
⢠Generates business opportunities
Conclusion â Main benefits
Payoneer is a financial services company that provides online money transfer and digital payment services. Account holders can send and receive funds into their bank account, Payoneer e-wallet, or onto a re-loadable prepaid debit card that can be used online or at points-of-sale.the company specializes in facilitating cross-border B2B payments. It serves customers in various business verticals
You can see the variety of services we provide that allow different inflows in different currencies as well as different outflows
Lets explain it with an example Airbnb allows renters to rent their appartments collects money from tenants and in the end of the month needs to pay the renters however it is not their expertise since renters come from different countries and Airbnb need to face compliance so they use companies like us at first weâd collect moey and transfer it directly to tenants but when we allowed virtual balances created eco systembut also allowed fraud and money laundering and still we need to obey the compliance .. Compliance us and china
At first we used rdbms but it wasnât sufficient, you canât know the source of funds since it can make dozens of hops from one payoneer user to another.
So we started to use neo4jWeâve enabled our risk investigators to quickly trace the funds to their source, how they initially entered a payoneer eco system and their entire lifecycle in itmore than that in rdbms we are used to look at funds as just a simple decimal number however in neo4j we can look at it as a sum of its elementsthese 300$ can be ⌠but also 7cts from a and 5 cents from b much more complicated structurehaving earned the better latency and ability to read complicated graphs on the fly allowed us to make much more complicated decisions.For instance if 10% of the funds went through specific node we can raise a flag and let risk decide how the choose to deal with that specific transactionwithout a graph such rules or actions would involve tedious human work of our analysts with graph it became automated and happen in real time
Once we achieved that complicated structure we could achieve more goals one of the main beneficiaries of the graph is our risk department that is required to detect fraud and money laundering attempts. Weâve built a visual tool that allowed the risk team to make easily make very complicated investigations and reach fast conclusions they basically were able to answer 2 simple questionsfirst is what inflow and outflow transactions a specific person made and their past and futures until the distance of x hops and when did they happen in addition to that we could trace the funds all the path they made in payoneer with a simple explode of a node. You can see that some suns and just hops and we could find entire money laundering networksin the past it would have been almost impossible with rdbms .The second question is whether 2 people are connected and have they ever made transactions from one another regardless how they tried to hide it, by just using simple graph algorithms.
In addition we could stop fraudulent behavior early by recognizing patterns and applying some more complicated algorithms such as ktails or other sophisticated pattern recognition algorithms thus we could predict possible fraudists
Thank you Arthur for interesting use case.
Hello everyone,
Iâm Alex Mirkin, and I work in Data department of Payoneer (Data = BI).
In which I manage companyâs DWH and work on various data solutions and integrations.
Today I will present how we use Neo4j in our department.
We are developing tools for Risk analysts that help detect suspicious activities such as: Fraud, abuse and money laundering.
Today, fraudsters are more and more sophisticated
Once it was a single man trying to find ways around the system
Today we are dealing with multiple fraudsters which form organizations with departments.
One common way of their activity is to register multiple accounts with different account details.
But sometimes each account has the one same detail as the other and that forms a fraud ring.
It is almost impossible to detect that kind of activity with relational databases.
If an analysts at risk wanted to understand how one account is related to another he had to run several reports, where each report output will serve the next report input.
With Neo4j we are able to run this kind of analysis far more quickly by a simple cypher query.
Another example â when analyzing report data in tabular form it is very hard to understand the connections.
But when visualazing the data as a graph we can immediately see the connections and the structure.
This is an actual case that we had.
Refer a friend reward abuse.
Suddenly we started to see too many rewards.
To analyze this I have loaded all the refer a friend data to a graph.
And then wrote a cypher query that returns this pattern.
And we saw right away the problem.
We use neo4j in two different ways:
Visual Investigation
Automation
API â pre defined checks of Risk level of a transaction
or check if two accounts are related
Alerts â suspicious transactions or registrations
We use Linkurious as our client tool for the analysts.
They can save cases and shave between others.
Refer a friend nodes displayed randomly
Now the same nodes after using left to right layout
The data make much more sense in that way.
A common case where funds are transferred from multiple account to the one in the center.
Which may indicate a fraud.
This is a Gephi viz that presents accounts that transferred money inside payoneer.
The groups in the center had the most activity and we can use this data to understand common patterns and maybe new business opportunities.
To conclude:
The main benefits that we get with Neo4j are:
Catch cases of fraud that we could not detect before
Perform same analysis much faster
By understanding the common patterns we can propose better terms for customers.