Short answer: in last-mile delivery, number masking puts a temporary virtual number between the rider and the customer for the life of the delivery. Both can call each other while the order is out, and when it completes neither retains the other's number. On Edesy this runs through the Session API at ₹1.50 per minute — roughly ₹3-5 per order for a typical two-call delivery.
Delivery is the use case masking was effectively invented for, and it has operational details that generic privacy guidance misses.
The Three Problems It Solves
Rider safety and customer safety, in both directions. A customer who has a rider's personal number can call them at midnight; a rider who has a customer's number can contact them after the order. Both happen, both generate complaints, and both end up as the platform's reputational problem regardless of fault.
Attribution on failed deliveries. "The rider never called" versus "the customer did not pick up" is the single most common last-mile dispute. Masked calls produce call logs tied to the order — with attempt times and outcomes — so the question is answerable instead of adjudicated by goodwill.
Rider churn does not leak your customer base. Fleet attrition in Indian last-mile is high. If riders accumulate customer numbers, that data walks out with them. A masked number that expires with the order means there is nothing to accumulate.
Delivery-Specific Details That Matter
Both directions are mandatory
The customer calling the rider is not a nice-to-have in delivery — "I'm at the gate, which building?" is the call that completes the order. A one-way setup where only the rider can initiate drops exactly that call. Use two-way masking.
Session lifetime should match the delivery, not the day
Ending the session when the order is marked delivered — rather than letting it expire on a fixed timer — does two things. It shrinks the number pool you need, because numbers return to circulation sooner. And it closes the window in which a rider can call a customer about an order that is already finished.
Reassignment and returning customers
In dense delivery areas, the same customer orders repeatedly. If a number is recycled quickly and a customer calls back the number from their previous order, they can reach an unrelated rider. A cooldown before reuse mitigates this — at the cost of a slightly larger pool. See pool sizing.
Peak concurrency, not daily volume
Delivery volume is extremely peaked — lunch and dinner for food, evenings for e-commerce. Size the pool to the worst hour. A fleet doing 5,000 deliveries a day may need only a few hundred numbers, but it needs them all simultaneously at 8pm.
Riders need it to work with no training
The rider taps call in your app; masking should be invisible to them. If a rider has to understand anything about how it works, the design has gone wrong. The same applies to the customer, who should simply see a number that rings the person delivering their order.
What Happens When Nobody Answers
Worth designing explicitly, because delivery has a specific pattern: the customer is unreachable and the rider is standing outside.
- A missed masked call must be returnable. If the customer sees a missed call from the virtual number and calls back within the session, it should reach the rider. This is two-way working as intended.
- After the session ends, it should not. The returned call should land somewhere sensible — a support line — rather than silently failing or reaching a stranger.
- Attempt records drive the failed-delivery policy. Whether a rider genuinely attempted contact before marking an order undeliverable is a question your call logs answer.
Cost Per Delivery
At ₹1.50/min, a typical delivery generating two calls averaging 90 seconds costs about ₹4.50.
Model it against your per-order economics rather than against another provider's per-minute rate. The comparison that matters is the cost of one avoided failed delivery — re-attempt cost, refund, or a lost customer — against a few rupees of masked calling. See what number masking costs.
What It Does Not Fix
- It does not improve pickup rates. A customer who ignores unknown numbers will ignore this one too.
- It does not cover SMS or WhatsApp. Edesy number masking is voice.
- It does not prevent a number being read aloud on the call.
- It does not fix address quality. If your addresses are bad, riders will keep calling — masking makes those calls private, not unnecessary.
Related Pages
Edesy has dedicated pages for delivery-adjacent verticals: grocery delivery, pharmacy delivery, moving and relocation and car rental.
Start a session on the self-serve portal and test the full flow — including a callback after the session ends — before wiring it into your dispatch system.