RecordPoint includes a library of predefined standard signals that automatically detect sensitive or important content within your organisation’s data. This article describes each available signal, what it detects, how detection works, and how context words affect detection behaviour.
How signals detect content
Standard signals use one or more of the following detection methods:
- Pattern matching: the signal scans content for text that matches the structural format of the target identifier.
- Validation logic: where the identifier follows a published validity rule, the signal applies that rule to confirm a candidate match is plausible.
- Named Entity Recognition (NER): a trained model identifies entities such as person names based on linguistic context. Used by HasPerson.
- Format-aware validation: some signals (for example, HasPhone) use additional logic that understands international and national format conventions and filters out values that are not plausibly real.
Any of these detection methods can be combined with optional context words to further filter results. See How context words affect signal detection below.
Global signals
The following signals apply globally and are not specific to any country or region.
| Signal | What it detects | Detection method | Region |
| HasPII | Umbrella signal — set to true when any other PII or PCI signal fires. Useful for filtering records that contain any form of sensitive information. | Derived from all other signals (no direct detection logic). | Global |
| HasPCI | Payment card numbers (credit and debit cards) from all major card schemes — Visa, Mastercard, Amex, Discover, JCB, Diners, and Maestro. | Pattern matching with validation logic that confirms the candidate is a plausible card number. | Global |
| HasPhone | Telephone numbers in a wide range of international and national formats. | Format-aware validation. Multiple confidence tiers — fully formatted international numbers score higher than bare local numbers. | Global |
| HasEmail | Email addresses. | Pattern matching with format validation. | Global |
| HasIBAN | International Bank Account Numbers used for cross-border bank transfers. | Pattern matching with validation logic. | Global |
| HasCrypto | Cryptocurrency wallet addresses. Currently supports Bitcoin only. | Pattern matching with validation logic. | Global |
| HasPerson | Full personal names, including first, middle, and last names or initials. | Named Entity Recognition (NER) using a trained model. Not pattern-based — relies on linguistic context. | Global (English-optimised) |
USA signals
The following signals are specific to the United States.
| Signal | What it detects | Detection method |
| HasUSSSN | US Social Security Numbers — the identifier issued to US citizens and residents. | Pattern matching. |
| HasUSBankNo | US bank account numbers. | Pattern matching. |
| HasUSDriverLicense | US driver’s licence numbers. Each US state has its own format. | Pattern matching tailored to each US state. |
| HasUSITIN | US Individual Taxpayer Identification Numbers — issued to non-citizens who cannot obtain an SSN. | Pattern matching tailored to the ITIN format. |
| HasUSPassport | US passport numbers. | Pattern matching. |
| HasMedicalLicense | US Drug Enforcement Administration (DEA) registration numbers — the identifier issued to practitioners authorised to prescribe controlled substances. | Pattern matching with validation logic specific to the DEA registration format. |
United Kingdom signals
The following signal is specific to the United Kingdom.
| Signal | What it detects | Detection method |
| HasUKNHS | UK National Health Service (NHS) numbers — the unique patient identifier used across the NHS. | Pattern matching with validation logic specific to the NHS number format. |
Australia signals
The following signals are specific to Australia.
| Signal | What it detects | Detection method |
| HasAUTFN | Australian Tax File Numbers — the ATO’s unique identifier for taxpayers (individuals and organisations). | Pattern matching with validation logic specific to the TFN format. |
| HasAUMedicare | Australian Medicare card numbers — the identifier for Australia’s public health rebate system. | Pattern matching with validation logic. |
| HasAUABN | Australian Business Numbers — the identifier for businesses registered in the Australian Business Register. | Pattern matching with validation logic specific to the ABN format. |
| HasAUACN | Australian Company Numbers — the ASIC identifier for registered companies under the Corporations Act 2001. | Pattern matching with validation logic specific to the ACN format. |
Over 170 additional out-of-the-box recognisers are available. To add them to the list of signals processed on ingested content, click Add on the Standard Signals page.
Enabling, processing, and disabling signals
Enabling a signal applies it to newly ingested content, not existing content. To process enabled signals against existing content, select the signals you want and click Process Signals. Selected signals are reprocessed and overwrite previous results on existing content. Unselected signals remain unchanged.
When you disable a signal, results that were already detected remain on existing records until those records are reingested or replaced by a new version.
Process status:
Process Required — a signal has been enabled but has not yet been processed.
Processed — a signal has been processed after being enabled, either through ingestion or manually.
Disabled — a signal that is no longer running on newly ingested content or was never enabled.
If you have reached the signals limit, you cannot enable additional signals.
Customising signals
Administrators can add context words to tune out-of-the-box signal detection to their needs, ensuring detected patterns appear in a meaningful context. To add or edit context words, select the relevant signal from the list, then click Edit button.
The signal’s configuration dialog opens, where you can add or edit context words.
How context words affect signal detection
Context words are an optional configuration available in RecordPoint. They let you narrow detection so that a signal only fires when specific keywords appear near the matched content.
- No context words configured (default): the signal detects matches using its built-in pattern, validation logic, and any model-based recognition. All matches are reported as normal.
- Context words configured: a match will only be reported when one or more of your configured context words appear near the detected pattern in the same content. Context words act as a filter — they reduce false positives, but they may also reduce recall if your context words are not present in every relevant document.
- How multi-word entries are interpreted:
- Each line is a separate search term. Enter one context word or phrase per line.
- Multiple words on the same line are treated as a phrase. All the words must appear together, in that order, near the matched content.
- Multiple lines act as alternatives. A match is reported if any of the lines appears near the detected pattern.
Example: For HasUSDriverLicense, entering drivers license on a single line looks for the full phrase "drivers license" near the match. Entering social security on one line and ssn on the next looks for either the phrase "social security" or the term "ssn" near the match
Context words are available on every standard signal listed in this article and on custom signals you create. The only exception is HasPII, which is a derived rollup with no detection logic of its own.
Example: If HasPhone is configured with context words ‘mobile’ and ‘contact’, a phone number buried in unrelated numerical data will not be detected — only phone numbers that appear near those words will be flagged.
You can test standard signals and their context words before enabling them. Select Test recogniser to highlight any matches found in the test content:
To apply the definition, select Use recogniser. This closes the dialog and updates your signal.
Using signals in search
To find records marked with specific signals, use Advanced Search, which has distinct search properties for each signal.
To search for content containing a standard signal:
- Sign in to RecordPoint.
- Select Advanced Search.
- In the property field, type “has” to display the list of available signals.
- Set the search to PCI/PII Signals — Signal name — Equals — Yes.
- Select Search.
- The results will display all records which match your query.
Only enabled signals that have been processed are searchable.
If a signal is disabled after being enabled and processed, it still appears in the Advanced Search dropdown menu, although no new records will match it.