Short answer: Edesy number masking is ₹1.50 per minute, pay as you go, with two-way masking, call recording, real-time analytics, API access, webhook events and the self-serve portal included rather than charged separately. Enterprise pricing is available for high volume. Current detail is on the pricing page.
The per-minute rate is the easy part to compare. The parts that actually move a bill are below.
What Gets Charged Separately Elsewhere
When comparing providers, these are the line items that turn a cheap headline rate into a surprising invoice:
| Charge | Why it appears | What to ask |
|---|---|---|
| Number rental | A monthly fee per number in your pool | Is pool rental separate from per-minute? |
| Recording storage | Free to record, charged to retain | How long is retention, and what after? |
| API request fees | Charged per session create or per call | Is the API metered separately? |
| Minimum commitment | A floor regardless of usage | Is there a monthly minimum? |
| Two-way as an upgrade | One-way in base, two-way costs more | Is bidirectional included? |
| Webhook delivery | Charged per event | Are events included? |
The pattern to watch for is a low per-minute rate paired with per-number monthly rental. Masking is number-heavy by nature — see pool sizing — so rental scales with your concurrency, and on a large pool it can exceed the call charges entirely.
Model Cost Per Order, Not Per Minute
Per-minute pricing is not the number that tells you whether this is affordable. Your business cares about cost per order:
Cost per order = average masked minutes per order × rate
Worked example. A delivery platform where a typical order generates two masked calls averaging 90 seconds each:
- Masked minutes per order: 3
- Rate: ₹1.50/min
- Cost per order: ₹4.50
Compare that against your take rate per order, not against a competitor's per-minute rate. For most marketplaces the number lands in single-digit rupees against a take rate measured in tens or hundreds — which is why the decision is usually about capability rather than price.
Then check the sensitivity: what happens if average call duration doubles? At 6 minutes an order it is ₹9. Still small. If your model breaks at that point, the problem is the margin, not masking.
Levers That Genuinely Reduce Cost
End sessions when the order completes. This is the biggest one and it is free. Sessions left to expire on a timer hold numbers that could be serving other orders, inflating the pool you need.
Right-size the pool. Over-provisioning is quiet, continuous waste. Instrument utilisation and size to your real peak rather than a comfortable guess.
Be deliberate about recording. Recording every call is the default choice, not necessarily the correct one. If you only need recordings for disputes, consider whether every call needs to be retained — and for how long.
Check what your calls actually cost you in duration. A confusing IVR or a bad routing rule adds seconds to every masked call. At scale, trimming ten seconds off an average call is a real line item.
Where Cutting Costs Backfires
- Under-sizing the pool. Saving on numbers by running at 95% utilisation means allocation failures at peak — orders with no contact path, which cost far more than the numbers would have.
- Turning off recording entirely. Fine until the first serious dispute, when the absence of a recording is the expensive part.
- Choosing one-way to save money. It does not save money — a one-way session consumes the same number — and it drops the callback. See one-way vs two-way.
What "Cheapest" Usually Hides
A materially lower headline rate normally means one of: two-way is an upgrade, numbers are rented separately, recording retention is short or charged, there is a monthly minimum, or support is limited. None of those are dishonest — they are just a different packaging, and you only find out which by asking the six questions in the table above.
Getting a Real Number for Your Case
The fastest way to a credible estimate is to run a small pilot rather than model it. Create a few sessions on the self-serve portal, run a week of real orders through them, and measure your actual average masked minutes per order. That figure is the one your whole model rests on, and it is almost never the one people guess.
See number masking ROI for the fuller business case.