VigiDoc a product by SSHack

Guide

What is DLP software? How it works and how to choose one

DLP software spots sensitive data leaving your company, whether by mistake or with malicious intent. This guide explains what it does, how it works, the main types, its limits and the criteria for choosing one. It is written by SSHack, an offensive security firm and the maker of the VigiDoc DLP software.

Published · about 10 min read

In short

DLP (data loss prevention) software monitors the channels through which data leaves a company: USB drives, email, web browsing, cloud. It identifies sensitive content using rules (keywords, patterns, labels), then logs, alerts or blocks depending on its configuration, and keeps the records needed for an investigation.

  • DLP protects against insider leaks, intentional or not, which antivirus software and firewalls don't see.
  • There are four main types: endpoint, network, email and cloud.
  • DLP can stop at detecting and alerting, or go as far as blocking: each approach has its advantages.
  • The right choice depends on the channels to cover, where the data is hosted and how much operational workload you can take on.

What is DLP software?

DLP stands for data loss prevention, sometimes also called data leak prevention. DLP software is a security tool that tracks a company's sensitive data (customer files, contracts, designs, health or financial data) and flags, or even prevents, its transfer to unauthorized people or destinations.

Data exists in three states, which DLP tools cover to varying degrees: data at rest (stored on a computer, a server or in the cloud), data in motion (traveling across the network, by email or through a web upload) and data in use (opened, copied or printed on a computer).

Despite the name, DLP does not always "prevent": some tools block transfers, others only detect and log them. This is one of the first selection criteria.

Why does DLP software matter?

Antivirus software, firewalls and EDR mainly protect against attacks from outside. A data leak, on the other hand, often happens through ordinary actions taken from inside: copying files to a USB drive before leaving the company, sending an attachment to the wrong recipient, uploading a folder to a personal file-sharing service. Without a dedicated tool, these transfers leave no usable trace.

Insider leaks, intentional or not

An insider leak can be a mistake (wrong recipient, wrong file) or a malicious act (a departing employee, espionage). Either way, the company first needs to know it happened, then be able to establish what left, when and through which channel.

Departures are a critical moment. According to Proofpoint's 2026 Voice of the CISO report, 36 of the 100 French CISOs surveyed had suffered a significant data loss, and all 36 said departing employees had played a part in it. A study by the vendor Cyberhaven covering the activity of 1.4 million employees (2022) found that nearly one in ten employees exfiltrates sensitive data within six months, with an 83% increase in incidents in the two weeks before a resignation. More figures, each with its source, edition and scope, are collected on our page Insider data leaks: the 2026 figures (in French).

Security obligations

None of these texts requires DLP software as such, but DLP is one way to show that concrete measures protect sensitive data.

A financial risk

According to the Cost of a Data Breach study by IBM and the Ponemon Institute (2026 edition, 35 French organizations), the average cost of a data breach for a company in France is €3.49M. This figure covers all breaches, not only insider leaks, but it gives a sense of the stakes.

How does DLP software work?

Whatever the vendor, DLP software follows the same main steps.

  1. Define what is sensitive. The company describes the data to protect: keywords ("confidential", a project name), regular expressions (credit card numbers, IBANs, social security numbers), fingerprints of reference documents or classification labels applied to files.
  2. Monitor the exit channels. Agents installed on endpoints, network sensors or cloud connectors watch outgoing transfers: copies to removable media, sent emails, web browsing and uploads, cloud sync, printing.
  3. Analyze content and context. The DLP checks each transfer against the rules: file name and content, recipient, channel, user, time.
  4. Respond. Depending on the policy, it logs the event, alerts an analyst, warns the user, blocks the transfer, quarantines it or encrypts it.
  5. Investigate and prove. The logs, and sometimes a copy of the files involved, let you reconstruct the incident: who, what, when, through which channel.

Everything depends on the quality of the rules: too broad, they bury the team in false positives; too narrow, they let leaks through.

The different types of DLP

DLP tools differ first in where they observe the data.

The four types of DLP
TypeWhat it monitorsStrengthsBlind spots
Endpoint DLPActivity on each computer: USB drives, printing, applications, uploadsSees data in use, even outside the corporate networkAn agent to deploy and maintain on every computer
Network DLPTraffic crossing the corporate network: web, email, file transfersNo agent to install on computersBlind to encrypted traffic without decryption, to computers off the network and to USB drives
Email DLPMessages and attachments sent through the email systemCovers the most common channel for accidental leaks: the wrong recipientOnly sees email
Cloud DLP (CASB)Data stored and shared in online applicationsControls sharing in the cloudDepends on each application's connectors; data passes through a third-party service

Full suites combine several types. A small business often starts with endpoint and email, which cover the most common actions: copying to a USB drive and sending an attachment.

Detect or block?

DLP in blocking mode stops the transfer as soon as a rule is triggered. DLP in detection mode lets it through, but logs it, raises an alert and keeps the evidence.

Detection or blocking: what changes
CriterionDetect and alertBlock
Impact on workNone: nothing is interruptedA false positive blocks a legitimate action
RolloutFast: rules are refined based on the alertsSlower: each rule must be tested before it is enabled
Accidental leakSpotted right away, then corrected (reminder, procedure)Prevented if a rule covers it
Malicious leakLogged, with the evidence needed to investigateSlowed down, but the person can look for another route
Evidence for the investigationAt the core of the tool: logs and copiesVaries by tool

From an insider attacker's point of view, a block reveals exactly which rule exists and invites a workaround: an encrypted archive, a screenshot, another channel. Discreet detection, on the other hand, leaves a trace the person does not suspect. A common approach is to start by detecting, measure the real data flows for a few weeks, then block only the most critical cases. We expand on this view, with a case from our engagements, in Detect rather than block: a Red Team perspective.

On-premises or cloud DLP?

By definition, DLP handles a company's most sensitive data: its logs and file copies must be protected at least as well as the originals.

On-premises vs. cloud DLP
CriterionOn-premises DLPCloud DLP (SaaS)
Logs and captured filesOn your serversWith the vendor or its hosting provider
Internet accessOptional, compatible with an air-gapped networkRequired
Control over the dataFullDepends on the contract and where the data is located
OperationsA server to host and maintainInfrastructure managed by the vendor

Sending captured files to a third party raises an obvious question: the DLP itself becomes a way for data to leave. This is why organizations bound by confidentiality (defense, healthcare, law firms) or working on air-gapped networks may prefer on-premises hosting.

The limits of DLP software

DLP reduces risk and provides evidence; it does not replace access rights management or staff awareness training.

How to choose DLP software: 8 criteria

  1. Channels covered. USB drives, email, web, cloud, printing: start from your real leak scenarios.
  2. Mode of action. Detection only, blocking, or both depending on the rule.
  3. Hosting. On-premises or cloud, and support for air-gapped networks if needed.
  4. File formats analyzed. Text, PDF, Office documents, archives, and above all what the tool cannot read.
  5. Tamper resistance. Encrypted logs, alerts when an agent stops, protection against uninstallation.
  6. Operational workload. Deployment time, rule tuning, number of alerts to review each week.
  7. Legal framework. Proportionate use and employees informed beforehand, as French law requires; check the rules that apply in your country.
  8. License and trial. Price per endpoint, commitment period, ability to test on your own fleet.

We apply these criteria to nine tools, VigiDoc included, in our DLP software comparison for small businesses.

Where does VigiDoc fit?

VigiDoc is endpoint DLP software designed by SSHack for small and midsize businesses and public institutions that run Windows. It takes the detection approach and never blocks a transfer. It logs copies to USB drives, Outlook emails and their attachments, attachments added in Gmail and Outlook on the web, and visited web pages. It raises an alert as soon as a sensitive keyword appears.

The logs and an encrypted copy of the files sent out via USB drive or email stay on your own server, including on an air-gapped network. On the web, VigiDoc records the address, title and time of the pages visited: it does not read the content the browser sends.

Frequently asked questions

What does DLP stand for?

DLP stands for data loss prevention, also called data leak prevention. The term covers the tools and practices that spot sensitive data leaving an organization, then alert or block according to defined rules.

What is the difference between DLP and EDR?

EDR (endpoint detection and response) monitors how programs behave in order to detect malware and intrusions. DLP follows the data: it spots sensitive files leaving the company. The two tools complement each other.

Is DLP software mandatory?

No. There is no law that requires DLP as such. However, the GDPR (article 32), the NIS 2 directive and trade secret protection all require appropriate security measures, and DLP is one way to implement them and to show that they are in place.

Can DLP read encrypted files?

Generally, no: a password-protected archive or an encrypted file cannot be analyzed. DLP can still log the file leaving (name, size, channel, recipient), which remains useful information for an investigation.

Is DLP software suitable for a small business?

Yes, as long as you choose a tool that is simple to run. For a small business, the deciding factors are the number of alerts to handle, the deployment time and the ability to test the tool on your own fleet before buying.

Can you monitor employees' computers with DLP?

Yes, within a defined framework. In France, workstation monitoring must be proportionate to its purpose and employees must be informed beforehand, under the French Labor Code and the recommendations of the CNIL (France's data protection authority). Other countries have their own rules: check your local employment and data protection law.

Sources