7 of 18 DLP tools watch email, laptop and cloud; none documents a PAN detector

01Seven of eighteen watch all three exits the 2024 post promised

The guide records, for each product, the channels its vendor documents: endpoint, email, web, cloud applications, removable media, print and AI prompts. Endpoint is the most common at 14 of 18. Email follows at 13, cloud at 11. Only 7 products document all three of the exits our 2024 post described.¹

The gaps follow the architecture. Inline platforms such as Zscaler and Netskope see web, cloud and email traffic that passes through them, and nothing on the device: no USB, no print. Endpoint-first products such as Trellix DLP Endpoint and Seqrite DLP are strong on the USB stick and blind to the cloud copy. The AI-prompt channel, the newest way data leaves, is documented by just two: Cyera Omni DLP and Seclore AI-DLP. A buyer who writes “DLP” in a requirement without naming the exits will get a working product watching the wrong door.

The 2024 post also missed a second question: not what left, but who is behaving differently. Content rules catch a file with a card number in it. They do not catch the employee who copies a little more each week in the month before resigning. Only 5 of the 18 products have what the guide rates as a full behavioural insider-risk engine; 4 have none, and 9 rely on rules and keywords rather than machine-learning classification, which means someone maintains the patterns by hand.¹ If the incident that started your project was a leaver, shortlist from the five, not from the channel chart.

02Six of eighteen are modules you may already own

Twelve of the 18 are products you can buy on their own. The other six are modules of platforms carded in other categories: Zscaler and Netskope in SASE, CrowdStrike on the endpoint, Proofpoint and Mimecast in email, and Coro’s small-business platform. In each case the data control sits inside a subscription the buyer may already hold.¹ The cheapest finding in any DLP project is that it was already paid for.

Two other constraints cut the list faster than any feature comparison. Fourteen of the 18 need their own endpoint agent, which has to coexist with the EDR, device-management and SASE agents already on the laptop; only four need none. And nine can run on your own infrastructure, which matters to the Indian estates whose classified content and incident evidence cannot sit in a vendor’s cloud.¹

03Two of eighteen print a price, and the licence is the small number

Sixteen of the 18 are quote-only or sold inside a platform. The two that publish are Forcepoint DLP, from about US$52 a user a year at entry and rising for the full channel stack, and Safetica, at US$72 to US$144 a user a year across its tiers.¹ At our guide’s conversion of ₹83 to the dollar, that is about ₹4,316 a user a year for Forcepoint and ₹5,976 to ₹11,952 for Safetica.

For a 200-person company that is roughly ₹8.6 lakh a year for Forcepoint at entry and ₹12 lakh to ₹23.9 lakh for Safetica, before GST. Neither figure is the real budget. The guide’s assessment is that classification and policy tuning commonly cost more than the licence in the first year. Someone has to decide what “confidential” means in your organisation, write it down, and cut the false positives until blocking is safe to switch on.

04None of the eighteen documents an Indian ID detector

A DLP classifier is only as useful as the patterns it knows. US products ship detectors for social security numbers and US card formats. We looked for documented PAN, Aadhaar or GSTIN detection across all 18 products and found it for none.¹ That does not prove they cannot be configured to find them, since most allow custom patterns, but it means the work and the testing fall to you.

Being built in India does not change the classifier result: Data Resolve, Seqrite and Seclore do not document Indian data-type detection either. The product that does is outside our 18. Microsoft Purview publishes a built-in sensitive-information type for the India Permanent Account Number, matched on the ten-character format plus nearby keywords.² For an estate on Microsoft 365, that is one more reason to test what is already licensed before buying. For everyone else, make PAN and Aadhaar detection a proof-of-concept test on your own records, not a datasheet question.