Communications Surveillance Software Pricing for a 250-Person Investment Firm
We price surveillance for a hypothetical 250-person firm with offices in London, New York and Hong Kong. Licenses look like the big cost, but alert-review staff can cost as much as the software.
Monitored users, not headcount, set communications surveillance software pricing
The first number a surveillance vendor asks for isn't your headcount. It's how many people you monitor, on how many channels, and how many hours of recorded voice you produce. Those three inputs drive most communications surveillance software pricing. The fourth big cost is the analysts who clear alerts, and it rarely appears in a vendor proposal.
To keep the numbers concrete, we use one firm throughout. Hypothetical example: Meridian Ridge Capital is a 250-person investment firm with an asset manager and a small broker-dealer arm, and offices in London, New York and Hong Kong. We assume 110 staff are in scope for electronic communications (eComms) monitoring: portfolio managers, traders, sales, research and some senior management. Of those, 40 traders and sales staff use recorded voice on turrets, desk lines or mobile. The firm also needs trade surveillance for its order flow.
All prices below are typical 2026 market ranges in USD. They vary by contract, region and deal size, so confirm them with vendors. Quotes in the Gulf, India or Southeast Asia can differ a lot from London or New York.
How EU MAR, FCA, SEC and SFC obligations decide scope
Regulation sets your floor. Vendors will happily sell you more than that, so know which rules apply to each desk and entity before you count users. Rules change, so treat the following as a general orientation, not a scoping document.
- EU and UK: The Market Abuse Regulation (EU MAR and the UK's onshored version) requires firms that arrange or execute transactions to detect and report suspicious orders and transactions. MiFID II and the FCA's recording rules require firms to record calls and electronic communications relating to transactions. Retention periods are typically several years, and regulators can ask for longer.
- United States: SEC books-and-records rules require broker-dealers and investment advisers to keep business communications, and FINRA rules require supervision. The SEC and CFTC have brought many off-channel messaging cases, so WhatsApp, iMessage and similar apps should be treated as in scope.
- Hong Kong: The SFC's Code of Conduct requires firms to record client order instructions and keep the records for a minimum period. The SFC also expects firms to supervise staff use of external messaging apps.
- Elsewhere: MAS in Singapore, SEBI in India, and the DFSA and ADGM regulators in the Gulf all set their own rules on recording, retention and market-conduct monitoring. Some bite on data residency, which limits the cloud regions a vendor can offer you.
For Meridian Ridge, voice capture is needed in London and Hong Kong, and messaging-app capture everywhere. Retention runs at the longest applicable period, commonly in the range of five to seven years, held in tamper-proof (WORM) storage. Have a lawyer licensed in each jurisdiction confirm your obligations.
eComms surveillance cost and voice surveillance licensing
eComms is usually priced per monitored user per month, in two layers. Capture and archiving pulls in email, Teams, Bloomberg chat, WhatsApp, WeChat and similar channels. Expect roughly $15–$30 per user per month across four or five channels, with each extra messaging-app connector adding perhaps $3–$10. Surveillance analytics runs on the archive: about $30–$55 per user per month for lexicon-based tiers and $60–$110 for machine-learning tiers.
Voice surveillance licensing has more moving parts. You pay for recording (roughly $20–$45 per user per month if the vendor provides it), for transcription by the audio hour, and for analytics on the transcripts.
Say Meridian Ridge's 40 voice users record about two hours each per trading day. That's roughly 20,000 audio hours a year. Transcription built for finance, with trading vocabulary, speaker separation and Cantonese and Mandarin support, might run $1–$3 per hour. Generic speech-to-text costs less, but on trading floors it tends to garble tickers and slang, and the surveillance then runs on bad transcripts. Insist on a test with your own recordings.
Watch the volume bands. Many contracts bundle a fixed number of hours, and overage rates can run well above the base rate, sometimes 1.5 to 2 times.
Lexicon versus machine learning: the alert-review math
The lexicon-versus-ML question is really a staffing question. Lexicon rules flag phrases such as "keep this between us" or "take it offline." They're transparent and easy to explain to an examiner. They also throw many false positives: newsletters, disclaimers, sports chat.
Assume Meridian Ridge's lexicon setup produces about 180 eComms alerts per trading day, around 45,000 a year. At three minutes per alert, eComms review alone takes roughly 2,250 hours. Add voice alerts, trade surveillance alerts, quality checks and case write-ups, and the firm needs about 2.5 full-time reviewers.
ML models score context and cut noise. Vendors often claim large alert reductions, commonly 40–70%. Treat any such figure as something to prove in a pilot on your own data, not a promise. If Meridian Ridge achieves 50%, it needs about 1.25 reviewers.
Fully loaded reviewer costs vary widely. As a rough guide, budget $90,000–$130,000 in London or New York, $70,000–$110,000 in Hong Kong, and $30,000–$55,000 in offshore centers such as Poland, India or the Philippines. We use a blended $70,000–$110,000.
Our position: above roughly 80 monitored users, ML usually costs less in total, provided you keep a lexicon layer for clear-cut terms and can document how the model works. Below that, a well-tuned lexicon is often enough.
Worked example: Meridian Ridge's annual surveillance budget
| Line item (hypothetical, USD/year) | Lexicon stack | ML stack |
|---|---|---|
| eComms capture and archive (110 users) | $20k–$40k | $20k–$40k |
| eComms surveillance analytics | $40k–$73k | $79k–$145k |
| Voice recording (40 users) | $10k–$22k | $10k–$22k |
| Voice transcription (~20,000 hours) | $20k–$60k | $20k–$60k |
| Voice analytics | $14k–$34k | $14k–$34k |
| Trade surveillance software | $60k–$160k | $60k–$160k |
| WORM retention storage | $5k–$20k | $5k–$20k |
| Software subtotal | $169k–$409k | $208k–$481k |
| Alert-review staffing | $175k–$275k (2.5 FTE) | $88k–$138k (1.25 FTE) |
| Annual run cost | $344k–$684k | $296k–$619k |
| One-off implementation | $60k–$180k | $80k–$220k |
Three things stand out. First, staffing costs about as much as the software, so a cheaper license can be the more expensive choice overall. Second, trade surveillance has the widest range. It depends on asset classes, order volume and the market abuse scenarios you need, such as spoofing, layering, insider dealing and front-running. Platforms that combine trade and comms surveillance can save money when they link a suspicious trade to the chat behind it. Third, the full stack lands at roughly $1,200–$2,700 per employee per year, a useful sanity check when a quote falls far outside that range.
Contract red flags and a buying sequence
Push back on these before you sign:
- Counting every licensed employee as a monitored user. Negotiate on in-scope users only, and agree how users are added and removed mid-term.
- Exit fees on archived data. Retrieving years of WORM data at contract end can cost more than a year's license. Fix the export format and price now.
- Uncapped model tuning billed as professional services. ML tuning continues for the life of the contract. It shouldn't arrive as a surprise invoice.
- Per-connector pricing that grows quietly. Every new messaging app becomes a new line item.
- No documentation of how the model works. If you can't explain why an alert fired or didn't, you'll struggle in a regulatory exam.
A sensible buying sequence: map your obligations by entity, count in-scope users and voice hours, shortlist three vendors, and run a four-to-eight-week pilot on a few months of your own historical data. Score each vendor on alerts per user, true-positive rate and analyst minutes per alert, then compare three-year total cost, not year-one license. If one stack costs less once reviewer hours are counted, pick it, even if its license is higher.
Quick answers
How much does communications surveillance cost per monitored user?
In typical 2026 markets, eComms capture plus surveillance runs about $45–$140 per monitored user per month. Lexicon tiers sit at the lower end and ML tiers at the higher end. Voice adds recording and analytics fees plus transcription by the hour. Prices vary by contract and region, so confirm with vendors.
Is voice transcription usually included in voice surveillance licensing?
Sometimes. Many contracts include a fixed number of audio hours and charge an overage rate above it, often well above the base rate. Ask for your expected annual hours to be priced explicitly, along with language support. Cantonese, Mandarin and Arabic sometimes cost more.
Do regulators require machine-learning surveillance?
No major regulator we're aware of mandates a particular technology. EU MAR, the FCA, the SEC and the SFC expect monitoring that works, suits your business, and that you can explain and evidence. ML is acceptable if you can document how it works and show it doesn't miss risks a lexicon would catch. A licensed compliance adviser in each jurisdiction should confirm your approach.
Should trade surveillance and communications surveillance come from one vendor?
One platform makes it easier to link trades with the chats behind them, and it can cut integration and review time. Specialist vendors may still beat it on depth, for example for complex fixed income or derivatives. Run both options through the same pilot and compare three-year total cost, including analyst time.
Written by the Smartnode Suite editorial team. Prices, features and rules change often - confirm details with the vendor or a licensed professional before you decide.