Microsoft D365 Fraud Protection - Tech Community Announcement
Dear Microsoft Dynamics 365 Fraud Protection Customers, We are excited to share with you that the Microsoft Tech Community Discussion Forum for Dynamics 365 Fraud Protection (DFP) customers is ready and available for you to explore. This collaborative space is designed for DFP customers to connect, ask questions, find answers quickly, and enhance their journey with other customers, partners, and the product team. Getting started is easy! Click this link to register and create a member profile: Home - Microsoft Community Hub To register for the tech community discussion forum, click the link above, and select “Register” in the middle of the home page, then follow the prompts to create a user profile and you’re ready to start engaging. 2.To start a new post, go to the Security, Compliance, and Identity discussion space and click ‘Start a New Discussion’: Security, Compliance, and Identity - Microsoft Community Hub. 3. Before posting a new discussion thread, make sure to select the label ‘Fraud Protection’ by scrolling to the bottom of ‘Choose a Label’ section. This helps others find your posts easily and stay notified about 'Fraud Protection' discussions. 4.To ‘Reply’ to a discussion, Use the ‘Fraud Protection’ label to search for DFP-related posts. To reply, simply use this link: Discussions with Label: Fraud Protection - Microsoft Community Hub We’re also excited to share that each week we'll spotlight a particular topic in the community called ‘Customer Handbook’ where we’ll share commonly asked questions and answers to help you get the most out of our product.Here are some of the DFP product topics we intend to cover in the coming weeks: rules, device fingerprinting, bot protection, scoring, reason codes, search, external calls, case management. We are thrilled to have you join us in the Tech Community and look forward to your active participation! If you have any questions, please reach out to us at: dfppmca@microsoft.com Thanks, The DFP Product GroupReminder to join our Fraud Protection On-Camera AMA tomorrow at 9AM PST!
Learn more about improving your lines of defense with the Fraud Protection team! We will talk about some of the new assets that our customers can leverage, see some updates on the engagement model (Community Discussion Space, other ways to engage, etc.) and our experts are available to answer any other questions you might have! So tune in and get your fraud juices flowing. The event page is here to watch and ask questions: Fraud Protection Tech Community Live! | Microsoft Community Hub See you there!Additional commonly asked Q&A related to ‘Device Fingerprinting’ in DFP continued
We're excited to keep our weekly spotlight series going on various topics within our Microsoft Fraud Protection Tech Community to help you maximize the benefits of Microsoft Dynamics 365 Fraud Protection (DFP). This week, we're continuing our focus on commonly asked questions about DFP 'Device Fingerprinting' which you can check out the Q&A details here: If you have any questions, please feel free to reach out in the Fraud Protection Tech Community. Your feedback is incredibly valuable to us. Best wishes, DFP Product Team ------------------ 1. Is device fingerprinting necessary? For DFP to provide the most accurate scores, device fingerprinting is highly recommended as it provides hundreds of device attributes. These critical attributes are used by DFP's machine learning to constantly improve the accuracy of your system. For more information, see the DFP Documentation site: Overview of device fingerprinting - Dynamics 365 Fraud Protection | Microsoft Learn 2. What is DFP Device Fingerprinting and how does it work? For a description of DFP Device Fingerprinting and how it works, please refer to the following DFP documentation: Overview of device fingerprinting - Dynamics 365 Fraud Protection | Microsoft Learn 3. What data isretained by DFP Device Fingerprinting and for how long? The data collected by the device fingerprinting feature is stored in a Microsoft designated data center closest to the location of the transaction source for up to 28 days. The data could also be stored along with the transaction that was sent against this profiling session in the customer’s selected geography, if the customer has opted in to storing data with DFP. (Note – for legacy Purchase assessment, data storage is not optional) 4. How can I tell if device fingerprinting has stopped for some reason? In Microsoft Dynamics 365 Fraud Protection, you can tell if device fingerprinting has stopped by monitoring the SSL certificate status and ensuring it is up to date. If the SSL certificate used for device fingerprinting is not renewed before its expiration date, device fingerprinting will stop collecting information. You should receive notifications regarding the SSL certificate for renewal status, as it is a critical component for the device fingerprinting service. Additionally, you can monitor the health and status of device fingerprinting through the Dynamics 365 Fraud Protection portal, which provides metrics that refresh near real-time. These monitors are designed to assist in detecting unusual transaction patterns or anomalies in observation events, such as fraud attacks and faulty rule releases. References: Overview of device fingerprinting - Dynamics 365 Fraud Protection | Microsoft Learn Web setup of device fingerprinting - Dynamics 365 Fraud Protection | Microsoft Learn Monitoring - Dynamics 365 Fraud Protection | Microsoft Learn 5. Outline the device profiling capabilities you support, if any. D365 Fraud Protection (DFP) supports probabilistic device identification, which involves returning an assigned device ID to the client along with device enrichment information. 6. What kind of device metadata can be gathered from the device being used? Data categories collected for web include: UserAgent information Canvas/WebGL data HTTP data Within and across session anomaly information IP, network, VPN and geo intelligence TCP Signature SSL/TLS Signature Client hints Javascript collected information like OS, processor, screen resolution, round trip time, etc. Data categories collected for iOS and Android include: Accelerometer and gyroscope data Location data Emulator and rooted information SIM card information Device specification data like advertising ID, screen size, total memory, screen refresh rate, build ID, etc. User preference data like is closed captioning enabled, is speak screen enabled, is haptic feedback enabled, etc. For a full list of attributes we collect across web, Android, and iOS, see Attributes in device fingerprinting - Dynamics 365 Fraud Protection | Microsoft Learn. 7. How is the metadata evaluated to identify anomalies and create sticky identifiers for device recognition? D365 Fraud Protection (DFP) enriches the attributes collected from the device and runs these attributes through an embedding model, creating a vector representation of a device that remains sticky over time. DFP then checks similarity to determine device ID assignment. With device vectors, we can consistently identify returning devices. 8. What kind of challenges (e.g., CAPTCHAs) are invoked if suspicious activity is detected? D365 Fraud Protection (DFP) doesn't provide challenge capabilities in the product, however, clients can invoke different kinds of challenges that suit their own business needs (CAPTCHA, RECAPTCHA or MFA, for example), through a “challenge” decision based on the bot score rules they configure in our rule engine. 9. What if clients are using a device fingerprinting of their own and they would like to complement with MS DFP, could they use both? Yes, they could use both services. The client can integrate with DFP and their other device fingerprinting and use the data from both on their end. 10. In the portal UX for classic PP, can attributes returned by device fingerprinting only be used in the "Post Risk Scoring" clause section? No, you can reference @"deviceAttributes.trueIp" (for example; gets returned from Device Fingerprinting) in both types of rule clauses – Prior to Scoring, Post Risk Scoring – as this is different than generating a risk score.Commonly asked Q&A related to ‘Rules’ in DFP
Hello Microsoft DFP Customers, We're excited to share some answers to commonly asked questions about D365 Fraud Protection (DFP)! Each week, we intend to spotlight a particular topic to help you maximize the benefit of our product and post the answers to questions here. This week, we're diving into DFP 'Rules'. Should you have any questions regarding the commonly asked Q&A provided, please do not hesitate to reach out here in the Fraud Protection Tech Community. Your feedback is incredibly valuable to us, and we genuinely appreciate your ongoing collaboration. Best regards, DFP Product Team ------------------ 1. What are the different inputs that can be passed into rules? In Microsoft Dynamics 365 Fraud Protection, you can create rules that utilize various inputs to convert an assessment into a decision, such as Approve, Reject, Review, or Challenge. The inputs for these rules can include: Attributes sent in the API request for the assessment, including custom data which can be accessed with the @ operator. For example, @"user.userId". Scores generated from Fraud Protection's artificial intelligence models, such as @"riskscore". Lists that you have uploaded to Fraud Protection. You can reference these lists in your rules after uploading them. Velocities that you have defined in Fraud Protection to perform velocity checks. External calls that you have created in Fraud Protection. Functions that you have created within Fraud Protection. References: Manage rules - Dynamics 365 Fraud Protection | Microsoft Learn 2. Why did a particular transaction not hit rule ‘X’? There could be several reasons why a transaction did not trigger a specific rule (Rule X) in Microsoft Dynamics 365 Fraud Protection. Here are some common factors to consider: Rule Configuration: Ensure that Rule X is correctly configured with the appropriate conditions and logic. If the conditions are not met, the rule will not trigger. Rule Order: The order of rules matters. If Rule X is lower in the order and a previous rule has already made a decision on the transaction, Rule X may not be evaluated. Rule Scope: Check if Rule X is scoped correctly to apply to the transaction in question. It might be limited to certain types of transactions or channels. Data Availability: The necessary data to evaluate Rule X must be present in the transaction. If the required data is missing or incorrect, the rule may not trigger. Rule Status: Verify that Rule X is active and not disabled or in 'observe' mode, which would prevent it from taking action on transactions. For a specific transaction, you can review the Rule analyst reports and Summary report in Dynamics 365 Fraud Protection, which provide insights into the transaction volume, rule decision distributions, and the impact of rules that you've enabled [1][2]. These reports can help you understand why Rule X did not trigger for a particular transaction. If you're still unable to determine why Rule X did not hit, you may need to consult with your Dynamics 365 Fraud Protection support team or review the service logs for more detailed information. There might have been a recent update or an issue escalated that could be related to the rule's behavior. References: [1] Rule analyst reports - Dynamics 365 Fraud Protection | Microsoft Learn [2] Summary report - Dynamics 365 Fraud Protection | Microsoft Learn 3. Why do we need to set up rules if the score can help evaluate risk? In Microsoft Dynamics 365 Fraud Protection, while the score generated by the AI model provides a valuable assessment of risk, setting up rules is crucial for several reasons: Customization: Rules allow you to tailor the fraud protection system to your specific business needs and risk appetite. You can create rules that threshold the score to make decisions that suit your business, such as approving transactions below a certain score and challenging or rejecting those above it. Complex Scenarios: Scores alone may not capture the complexity of certain fraud scenarios. Rules can incorporate additional parameters from the transaction payload, enabling you to detect business policy violations or emerging fraud patterns specific to your business. Control: Rules give you control over the decision-making process. You can define what actions to take based on the score and other attributes, such as triggering MFA or reviewing transactions from certain geographies. Adaptability: Fraud patterns evolve, and rules can be quickly adjusted to respond to new threats, whereas model retraining for scores might take longer. Segmentation: You can segment your traffic and set custom score cutoffs for different segments, optimizing fraud control for various product lines or transaction types . For a more detailed understanding of the role of rules in fraud protection, you can refer to the official documentation on Manage rules - Dynamics 365 Fraud Protection | Microsoft Learn which provides comprehensive guidance on rule management within the system. 4. What rule can help catch more fraud based on past data? In Microsoft Dynamics 365 Fraud Protection, transactions with the highest risk scores are those that are most likely to be fraudulent. The common rules applied to these transactions are designed to identify and prevent high-risk activities. Here are some of the rules that are commonly used: Threshold rules: These rules reject transactions that exceed a certain risk score. For example, transactions for gift cards might be rejected if the risk score is above 400. Velocity rules: These rules identify and block rapid, repeated transactions from the same entity, which could indicate fraudulent behavior. List checks: These rules compare transaction data against lists of known fraud indicators, such as device fingerprints or IP addresses. Anomaly detection: These rules look for patterns of behavior that are unusual and deviate from the norm, which could indicate fraud. For a more detailed understanding of the common rules applied to high-scoring transactions, you may want to review the "Score analyst reports" in the Dynamics 365 Fraud Protection portal, which can provide insights into the relationship between Fraud Protection scores and the rules that were executed. If you need further assistance or have specific questions you can also contact Microsoft support or your Microsoft authorized partner for additional assistance. References: Score analyst reports - Dynamics 365 Fraud Protection | Microsoft Learn How does inheritance work for rules? 5. How does inheritance work for rules? In Microsoft Dynamics 365 Fraud Protection, rule inheritance works within a multi-environment hierarchy. If your Fraud Protection instance has multiple environments, you can manage rules in a specific environment using the environment switcher. Rules in the top-level parent environment are evaluated first. If the rule settings for the top-level parent environment are set to "Run all matching rules until a decision is made," the rules in the second-level parent environment are evaluated next. This process continues unless the rule settings for an environment are set to "Run only the first matching rule," or until all the rules for the parent environment and the current environment are evaluated [1]. However, it's important to note that all resources, such as velocities, external calls, lists, and external assessments, are always local to an environment. Even in a hierarchy, resources defined in a parent environment are not inherited for use in rules in child environments. They are inherited for aggregation purposes but not for use in rules. For example, a velocity defined in a parent environment would increment based on transactions to a child environment, but if you wanted to reference that velocity in a rule, the rule would have to be in the same (parent) environment [2]. For functions, you can create them in any environment in the multi-hierarchy stack. When a function references resources available in the environment, the lower environments that invoke the function also inherit the resources that the function references For a more detailed understanding of how inheritance works for rules in Microsoft Dynamics 365 Fraud Protection, you can refer to the official documentation on Manage rules - Dynamics 365 Fraud Protection | Microsoft Learn References [1] Manage rules - Dynamics 365 Fraud Protection | Microsoft Learn [2] Functions - Dynamics 365 Fraud Protection | Microsoft Learn 6. How often should we revisit the rule and make adjustment? In Microsoft Dynamics 365 Fraud Protection, it's important to regularly revisit and adjust rules to ensure they remain effective against evolving fraud patterns. While there is no one-size-fits-all answer, here are some best practices: Regular Review: Rules should be reviewed on a regular basis, such as monthly or quarterly, to ensure they align with current fraud trends and business strategies. Performance Analysis: Utilize the Rule analyst reports to monitor the performance and impact of your rules. Adjustments may be necessary if you notice changes in fraud patterns or false positive rates. After Major Events: Review and potentially adjust rules after major events such as product launches, holiday seasons, or known fraud attacks, as these can change the fraud landscape significantly. Feedback Loop: Incorporate feedback from customer service and fraud investigation teams into your rule adjustments to address any new types of fraud they are encountering. It's also beneficial to stay informed about updates to Dynamics 365 Fraud Protection features and capabilities, as new functionalities may offer additional ways to enhance your rules and fraud protection strategies. References: Manage rules - Dynamics 365 Fraud Protection | Microsoft Learn 7. How do I create a rule based on ASN attribute To create a rule based on the ASN (Autonomous System Number) attribute in Microsoft Dynamics 365 Fraud Protection, you would typically use the ASN as part of the condition in a WHEN statement within the rule definition. Here's a quick guide on how to do it: Identify the ASN attribute: Determine the ASN attribute from the transaction data that you want to use in your rule. This could be part of the device information or network data. Access the Rules Editor: Go to the Dynamics 365 Fraud Protection portal and navigate to the rules editor section. Create a New Rule or Edit an Existing One: You can either create a new rule or edit an existing one to include the ASN attribute in the conditions. Define the Rule: Use the RETURN and WHEN keywords to define your rule. The basic structure of a rule is as follows: RETURN <decision> WHEN <condition> For example, if you want to flag transactions from a specific ASN for review, your rule might look like this: RETURN Review("Suspicious ASN") WHEN @"network.asn" == "12345" Replace "12345" with the actual ASN you want to monitor. Test the Rule: Before activating the rule, test it to ensure it works as expected and does not impact legitimate transactions. Activate the Rule: Once you are satisfied with the rule's performance, activate it to start using it for real-time transaction assessments. You can also use the visual mode in the rules editor for an easier rule creation experience, where you can select attributes from a drop-down menu and add multiple filters to a clause. For more detailed instructions and best practices on rule creation, you can refer to the official documentation on Manage rules - Dynamics 365 Fraud Protection | Microsoft Learn. 8. How do I create a velocity-based rule? Creating a velocity-based rule in Microsoft Dynamics 365 Fraud Protection involves defining velocities that monitor the frequency of events from a user or entity, which can indicate suspicious activity and potential fraud. Here's a step-by-step guide on how to create a velocity-based rule: Define a Velocity: Velocities are defined using the SELECT, FROM, WHEN, and GROUPBY keywords. Here's the structure you would use: SELECT <aggregation method> AS <velocity name> FROM <event type> WHEN <condition> GROUPBY <attribute name> For example, to define a velocity that counts the number of purchases from a specific IP address, you might use: SELECT Count() AS numPurchases FROM Purchase WHEN @"device.ipAddress" == "192.168.1.1" GROUPBY @"device.ipAddress" Create a Velocity Set: In the Fraud Protection portal, navigate to the Velocities section and select 'New velocity set'. Define your velocities within this set. Publish the Velocity: After defining your velocity, you need to publish it so it can be used in rules. Create a Rule Using the Velocity: Now that you have a defined velocity, you can create a rule that uses this velocity. In the rules editor, you would reference the velocity in a WHEN clause of a rule. For example: RETURN Review("High number of purchases") WHEN numPurchases > 5 Test and Activate the Rule: Before activating the rule, test it to ensure it works as expected. Once you're satisfied, activate the rule for it to take effect on real-time transaction assessments. For more detailed instructions, you can refer to the official documentation on Perform velocity checks - Dynamics 365 Fraud Protection | Microsoft Learn. 9. How do I create an IP-based rule? Creating an IP-based rule in Microsoft Dynamics 365 Fraud Protection involves using the IP address as a condition within the rule's logic. Here's a general guide on how to create an IP-based rule: Access the Rules Editor: Navigate to the Dynamics 365 Fraud Protection portal and open the rules editor. Define the Rule: Use the RETURN and WHEN keywords to define your rule. The basic structure of a rule is as follows: RETURN <decision> WHEN <condition> For an IP-based rule, your condition will involve the IP address attribute. For example: RETURN Reject("Suspicious IP") WHEN @"device.ipAddress" == "192.168.1.1" Replace "192.168.1.1" with the actual IP address you want to monitor. Test the Rule: Before activating the rule, test it to ensure it correctly identifies transactions based on the IP address without impacting legitimate transactions. Activate the Rule: Once you're satisfied with the rule's performance, activate it to start using it for real-time transaction assessments. For more detailed instructions, you can refer to the official documentation on Manage rules - Dynamics 365 Fraud Protection | Microsoft Learn. 10. Can you recommend the rule structure for MFA flow? In Microsoft Dynamics 365 Fraud Protection, setting up a rule structure for Multi-Factor Authentication (MFA) flow would typically involve creating rules that trigger MFA challenges based on certain conditions. Here's a recommended structure for such a rule: Define the Condition: Identify the conditions under which you want to trigger MFA. This could be based on risk scores, user behavior, transaction details, or other attributes. Create the Rule: Use the RETURN and WHEN keywords to define your rule. The basic structure of a rule is: RETURN <decision> WHEN <condition> For example, if you want to challenge a login attempt when the risk score is high, your rule might look like this: RETURN Challenge("MFA Required") WHEN @"riskscore" > 800 Test the Rule: Before activating the rule, test it to ensure it correctly identifies scenarios for MFA without impacting legitimate users. Activate the Rule: Once you're satisfied with the rule's performance, activate it to start using it for real-time assessments. For more detailed instructions, you can refer to the official documentation on Manage rules - Dynamics 365 Fraud Protection | Microsoft Learn.Fraud Protection Release Notes
Dear Microsoft DFP Customer, We are excited to share with you several new features/capabilities we have recently added this July to D365 Fraud Protection: Device anomalous attributes Assessment Reports Warm up calls in External Call New attributes for Purchase Protection Check references Updates to Agent Name field for Case Management All the product features above are available in your DFP environment today, and further details about each are below. If you have questions regarding any of the updates provided, please don’t hesitate to ask. Your feedback is extremely valuable to us, and we really appreciate your ongoing support and collaboration as we continue to improve the product! Best Regards, D365 Fraud Protection Team ----------- Device anomalous attributes We’ve added some additional attributes in the Device Fingerprinting Assessment template to help identify anomalous sessions. The new attributes help detect anomalies both within and across sessions. For cross session data, the sessions are grouped together using the cookie id. New attributes are listed below and their descriptions are available in our documentation. i. Screen resolution anomaly ii. Session anomaly iii. User agent browser anomaly iv. User agent language anomaly v. Time zone mismatch vi. OS mismatch vii. Session count viii. IP address history Note: These attributes are only available for Assessments created using Device Fingerprinting template. We’ve also added an additional device attribute function called Device.GetSpeedOfTravel(), which returns the maximum travel speed of a device in miles per hour. The maximum speed is determined by looking at the last five consecutive fingerprinting sessions, calculating the speed of the device from session to session, and returning the maximum. The device is identified over sessions using the cookie id. More details about this function can be found here. Assessment Reports Enhanced Assessment Reports & Updates: We're excited to announce that Virtual Fraud Analyst (VFA) reports are now available for our Assessment customers, allowing these customers to get valuable insights previously available only for Purchase Protection and Account Protection. Additionally, Assessment customers can now tailor the data filters to segment the data according to their needs, configure the observation events attributes to view distributions, and aggregate figures by selecting different amount attributes. Report Relocation & Renaming: Based on feedback, we've relocated these reports to the assessment level and renamed them for clarity and ease of use: i. Virtual fraud analyst is now Reporting ii. Summary is now Summary report iii. Rule analyst is now Rule report iv. Score analyst is now Score report v. Threat analyst is now Threat report Updates to Role permissions based on feedback: Manual Review Fraud Manager (for non-PSP customers): This role has been updated to include access to all the reports, ensuring comprehensive oversight Reporting (for PSP customers): This role has been updated to include access to both reports and monitoring dashboards Warm up calls in External Call If the traffic to your External call endpoint is too low, the connection can go cold and can increase the latency of response from the external service. To mitigate this, you can now choose to enable warm-up calls from the external call setup page. To enable warm-up calls, you will be required to provide a valid keep-alive URL. If you opt in to this functionality, whenever your traffic volume dips too low DFP will automatically make anonymous warmup calls to the keep-alive endpoint using GET method only. No parameters, configurations, or configurable headers can be used in warm-up calls. New attributes for Purchase Protection We are adding the following new fields to the Purchase Protection and Assessments schemas in this release: Attribute Name Schema Section Data Type daysSinceEmailFirstSeen user Int departureDistanceFromBillingAddress travelOverview Int latitude address Double longitude address Double phoneAddressMatchStatus paymentInstrumentList/paymentInstruments String These fields will be added to the Account Creation and Account Login schemas in an upcoming release. Check references We’ve made it easy to validate if a resource (such as velocity, external call, custom list, etc.) is being referenced elsewhere in DFP. With the new ‘Check references’ feature, you can quickly and efficiently do any of the following checks: If a custom list is referenced in decision rule, post decision action, velocity set, or function If a velocity set is referenced in decision rule, post decision action, or function If an external call is referenced in decision rule, post decision action, velocity set, or function If an external assessment is referenced in a decision rule, post decision action, velocity set, or function If a case management queue is referenced in a routing rule You can click the ellipsis and find ‘Check references’ Example: If a velocity is referenced in a rule, or functions, it will be shown in the pop up. Updates to Agent Name field for Case Management To comply with Privacy requirements, going forward, Fraud Protection would not store Friendly Name of Agents who make decisions on cases using Case Management. We would only store the user Id and the conversion to name would be done at run time using Microsoft Entra Active Directory. Searching for cases decisioned by a specific agent or reviewing Case Management reporting would still be possible, with a few changes in the user experience described below The filter attribute “Agent name” in Search has been renamed to “Agent name (deprecated)”. This attribute can be used to search for cases decisioned before July 29 th 2024 but it would be blank for cases decisioned after this date to comply with the Privacy requirement. Just as before, to search using this attribute, provide the full name of the agent in the value field (e.g. first filter in the screenshot below) A new filter attribute “Agent name (new)” has been added to Search to replace old field (Agent name). This new attribute can be used to search cases decisioned after July 29 th 2024. This filter also supports autosuggest so you need not type the full name, once you type a few characters of the desired agent name, autosuggest results should help you pick the right agent from a list of all users in the Microsoft Entra Active Directory matching the characters you typed. You can use an ‘Or’ query (see screenshot below) to search for cases that were decisioned before and after July 29, 2024 Note: After August 29, 2025, any data with “Agent name (deprecated)” field populated would have aged out because of the 13-month data retention policy for Fraud Protection Search. At that point, the “Agent name (deprecated)” filter would be removed to avoid confusion, as all the data in Search would be searchable via the new field. Case Management Report has been modified as well. The filter for “Agent name” now includes autosuggest experience. Just like before, multiple selections are allowed and you can easily include or exclude cases auto-decisioned by the System (due to queue timeout). TrevorRusher
