Overview
ScanX offers two AI-powered capabilities to automatically identify and set MCC values and provide underwriters with recommendations to find the best fit. Powered by KYC SiteScan, these capabilities use publicly available business information, such as a merchant’s website and business details, to recommend the most accurate MCC.
Rapid MCC Auto-Detection and auto replacement immediately benefits partners who are using our MATCH Pro integration which, as of January 1st, 2026, requires MCC in the data payload to run successfully. Without an MCC value when the Scorecard is started, MATCH will fail. This results in more manual work for underwriters who need to enter the MCC and rerun MATCH. It also means no more auto approvals due to the integration failure.
In addition to rapid MCC replacement, the MCC Detection scoring rule spends more time analyzing the business to make 3 highly accurate MCC recommendations. Underwriters can select from any of the results based on the model's reasoning or determine a different value of their choosing.
By automating MCC detection and providing confidence scores and clear reasoning, ScanX reduces manual review time, improves consistency, and helps underwriters make faster, more informed decisions.
Prerequisites
Before using the MCC Detection Scoring Rule, ensure the following:
- Contracted for KYC SiteScan
If you don't currently have a KYC SiteScan contract in place, contact your Customer Success Manager for more details; including limiting usage to just the rapid detection and replacement.
How does MCC Detection work?
The MCC Detection Scoring Rule uses business data collected during the application process to perform an AI-powered analysis of available data to make an MCC recommendation. The input data includes:
- Business legal name and DBA name
- Physical address (city, state, ZIP)
- Phone number
- Website
Based on this information, the rule generates three MCC recommendations. Each recommendation includes:
- 4-digit MCC
- Description of the MCC
- Confidence score indicating how likely the MCC accurately represents the merchant’s business
- Reasoning for the recommendation with excerpts from their website or other content to support it
Not all of the input fields are required, but the more data provided, the more accurate the results should be. For the rapid lookup, the model spends less time analyzing the data for a quick recommendation (15-20 seconds) that is suitable to use for the MATCH Pro requirement without slowing down the Scorecard running the other integrations.
The full MCC Detection scoring rule spends more time analyzing available data such as additional web pages for broader business context. The resulting recommendations should be similar to the rapid lookup, but the ranking and confidence score may be different. For example, restaurant may be the initial recommendation from rapid lookup, however, after deeper analysis of a web page about their catering service the top result end up being catering restaurant.
Confidence scores above 80% are generally considered accurate. Scores below 80% should be reviewed manually to confirm the best MCC selection. As AI is being leveraged to help make the recommendation, users should be sure to review the data for accuracy and report any inconsistencies in order to improve the tool.
In order to solve the MATCH Pro MCC requirement, Rapid MCC Auto-Detection is its own scoring rule that runs separately from the full KYC SiteScan report. When a Scorecard is run, Negative file runs first as usual and then we run the Rapid MCC Auto-Detection rule. The rapid rule autofills the top recommended MCC, then proceeds to trigger the rest of the integrations as usual with KYC SiteScan being the final integration to run and include the more in-depth MCC Detection analysis that is included in the report from KYC SiteScan.
If you are using both Rapid MCC Auto-Detection and MCC Detection rules on your Scorecard, you will only be charged for one KYC SiteScan report as you currently do. On the backend we are tying the rapid lookup request to the subsequent request for the full report so no additional usage charges are incurred.
Displayed information in the Scorecard
A summary of the MCC recommendations is included on the Summary page of the Scorecard for easy access to underwriters who can also make updates to the selected MCC from there. This includes visual indicators on the MCC field if an MCC was provided and it matches the top recommendation or not. It will also indicate to the underwriter if the MCC Detection top recommendation was used as the field value automatically so users can validate the reasoning if needed.
The MCC Detection rule itself displays several helpful fields:
- MCC Code: The currently active MCC in the scorecard. This may be the originally submitted value, an auto-selected recommendation, or a manually selected MCC.
- Original MCC: The MCC initially submitted. This appears only if the value has been overwritten, allowing easy comparison or reversion.
- View Details: Expands the rule to show input data and MCC recommendations.
MCC Recommendations
When you select View Details, the MCC Recommendations section displays the top three AI-recommended MCCs. For each recommendation, you’ll see:
- MCC: The recommended 4-digit Merchant Category Code
- MCC Description: A summary of the industry or services associated with the MCC
- Confidence Score: The model’s confidence that this MCC is correct
- Selected / Select This MCC: Indicates whether the MCC is currently applied, or allows you to manually apply it
Underwriters can manually select any recommendation to overwrite the current MCC directly from the scorecard or summary page.
Recommendation reasoning
Expanding an MCC recommendation reveals the AI’s reasoning. This includes excerpts from the merchant’s website and highlights the language or services that influenced the recommendation. Reasoning is especially useful when confidence scores between recommendations are close.
Enabling MCC auto-replacement
To add MCC Detection to your Scorecards, navigate to the Scorecards management page and select the Scorecard you want to add the rules to. Once in the Scorecard Editor, click on the Scoring Rules & Weights tab. Click the + in the page where you want to add the rule and search for the rule; searching "MCC" works well. The rules are under the KYC SiteScan section. Select the rules and click Next until it's added to your Scorecard. Be sure to save and publish your Scorecard for the changes to take effect.
Partners can configure how MCC Detection behaves in the scorecard:
- Auto-replace MCC in scorecard: Automatically applies the top MCC recommendation if it meets a minimum confidence threshold (default is 80%).
- Manual review only: Displays recommendations without automatically updating the MCC.
Even when auto-replacement is enabled, users can still change the MCC or revert to the original value at any time.
If you are looking to use Rapid MCC Auto-Detection to automatically add the top recommended MCC, you need to have the Auto-Replace MCC in Scorecard setting enabled. You may want to consider lowering the minimum confidence score threshold to reduce or avoid instances of MATCH failing due to missing MCC if the top recommendation doesn't end up being automatically inserted.
Credential Configuration for Rapid MCC Auto-Detection
In order for the Rapid MCC Auto-Detection rule to work, there is an additional credential that needs to be added under the Scorecard's Credential Manager. The new credential title is KYC Quick Check. Here you will enter the same information as your existing KYC credentials. Since the quick check is a separate API call, it needs a separate configuration, but ultimately ties to the same KYC account.
Thresholds and flags
For the standard MCC Detection scoring rule, you can optionally configure thresholds that trigger scorecard flags based on the top MCC confidence score:
- Review (yellow)
- Decline (red)
- Prohibited (black)
Thresholds can be set to trigger when the confidence score is greater than or less than a specific value. Because MCC Detection uses AI-generated confidence scores, best practice is to use Review flags only, prompting manual verification rather than automatic declines.
The most likely use case here is to flag to underwriters when the confidence score is below a certain threshold where you're concerned about the accuracy and want to make sure they manually review to ensure the most accurate MCC is selected. Our recommendation would be to start by using a threshold of < 80 to trigger a review as your team tests the results of the tool and gets comfortable with its recommendations, especially if you're using the auto-replace feature.
If any flag is triggered, the scorecard cannot be auto-approved.
Reports
The KYC SiteScan PDF report includes a Merchant Category Code section showing the same top three AI-recommended MCCs, confidence scores, and reasoning displayed in ScanX. This report can be used for audit, documentation, and underwriting review.
Frequently Asked Questions
What confidence score should I trust?
Confidence scores of 80% or higher are generally accurate. Scores below 80% should be reviewed manually.
Can I change the MCC after it’s auto-replaced?
Yes. You can select a different recommendation or manually enter an MCC at any time. The original MCC is always stored for reference.
Does MCC Detection block approvals?
Only if a configured flag is triggered. Without flags, MCC Detection does not prevent approval.
Where does the data come from?
The model uses business information provided in the application and data gathered from the merchant’s website during the KYC SiteScan web crawl.
Do I have to have KYC SiteScan to use MCC Detection?
Yes, you need to have credentials configured for KYC SiteScan to use the MCC Detection scoring rules. Contact your Customer Success Manager for more information if you want to enable it.
Do I have to use both the Rapid MCC Auto-Detection and standard MCC Detection rules on a Scorecard?
No, you can use either or based on your preference. Rapid MCC Auto-Detection is most useful for partners who don't collect an MCC during their merchant application process and are also running MATCH on the Scorecard. As of January 1, 2026, MATCH will fail to run if no MCC value is submitted in the payload. Using both allows a more accurate MCC to be provided afterwards to help with proper categorization and consistency across your portfolio.