Connect Rate Benchmark for Outbound Calling in India 2026: What Actually Gets Answered, by Hour, Circle, Operator and Number Type
A D2C brand in Gurugram was three months into a voice AI deployment and convinced the agent was the problem. Order confirmation completion was 31%, well under the number they had modelled. The team had rewritten the script twice, changed the voice once, and were preparing to switch platforms.
The agent was fine. Of every 100 numbers dialled, 68 never connected at all. The completion rate on connected calls was 82%, which is a good number. The campaign was dying at the dial, and every optimisation effort had been aimed at the 32% of calls where the customer was already listening.
Connect rate is the largest single multiplier in Indian outbound calling and the one most teams treat as a fixed constant. It is not fixed. Across the data in this post it ranges from 19% to 71% depending on when you dial, what circle the number is in, which operator carries it, what number you dial from, and how many times you have tried before. Those are all controllable.
This is the benchmark, segmented every way that turned out to matter, and the dialling strategy the numbers support.
What counts as a connect
Definitions first, because vendors and dialers report this inconsistently and the inconsistency is usually flattering.
Connect rate here means the called party answered and audio was established, as a percentage of numbers dialled. It excludes calls that reached voicemail, calls answered by an IVR or call-blocking service, and calls where the network returned a ringing state that never completed.
Three numbers frequently reported as "connect rate" that are not: dial rate (calls placed over numbers attempted, which is close to 100% and meaningless), answer-seizure ratio (a network metric that counts any answer supervision including operator messages), and contact rate over unique contacts rather than over dials, which counts a person reached on the fourth attempt as one connect out of one contact rather than one out of four dials.
Insist on connects over dials. It is the only version that predicts cost.
The dataset
18.4 million outbound call attempts placed between January and July 2026 across collections, delivery confirmation, appointment reminders and lead follow-up workflows. All India, all mobile-terminated except where landline is broken out. Numbers were DLT-scrubbed at dial time and all campaigns were consent-based and TRAI-compliant, which matters for comparability: unscrubbed campaigns show different and better-looking connect rates because they are dialling numbers that should not be dialled.
Connect rate by hour of day
All-India, weekdays, mobile-terminated:
| Time block (IST) | Connect rate |
|---|---|
| 08:00 to 09:30 | 22.4% |
| 09:30 to 11:00 | 31.7% |
| 11:00 to 13:00 | 43.8% |
| 13:00 to 14:30 | 34.2% |
| 14:30 to 16:30 | 38.6% |
| 16:30 to 18:00 | 45.1% |
| 18:00 to 20:00 | 47.3% |
| 20:00 to 21:00 | 39.4% |
Two peaks, late morning and early evening, with a lunch trough between them. The evening block is the strongest and it is also the block most compressed by regulation and by customer tolerance, so the effective window is narrower than the table suggests.
The early-morning number is the one worth acting on. A large number of Indian campaigns start dialling at 09:00 because that is when the operations team arrives. Connect rate in that first ninety minutes is roughly half the evening rate, which means a meaningful share of many campaigns' list is being burned at the worst hour of the day. Shifting the first hour's volume to the 11:00 block is free.
Segment variation is large. Collections on Hindi-belt borrowers peaked later, 18:30 to 20:00 at 51.2%, and performed notably worse before 10:30 at 17.9%. Appointment reminders to urban professionals peaked in the 11:00 to 13:00 block. Delivery confirmation was flattest across the day, because the recipient is expecting a call.
Connect rate by telecom circle
Top and bottom circles, weekday aggregate:
| Circle | Connect rate |
|---|---|
| Bihar and Jharkhand | 52.7% |
| Uttar Pradesh East | 49.8% |
| Madhya Pradesh and Chhattisgarh | 48.1% |
| Rajasthan | 46.9% |
| West Bengal | 44.2% |
| Andhra Pradesh and Telangana | 41.6% |
| Tamil Nadu | 38.4% |
| Karnataka | 34.7% |
| Maharashtra and Goa | 33.9% |
| Mumbai | 29.6% |
| Delhi NCR | 27.3% |
The spread is 25 points and the pattern is consistent: metro circles answer least, and the circles usually described as Tier-2 and Tier-3 answer most. Delhi and Mumbai subscribers receive far more unsolicited calls, screen more aggressively, and have higher adoption of call-blocking apps.
The practical implication runs against most campaign planning. Teams routinely prioritise metro segments because those customers have higher order values, then find the campaign economics do not work. On a cost-per-connect basis a Bihar number is roughly 1.9× cheaper to reach than a Delhi number, and that ratio frequently outweighs the order-value difference.
Connect rate by operator
| Operator | Connect rate |
|---|---|
| BSNL | 46.3% |
| Vodafone Idea | 41.8% |
| Airtel | 38.2% |
| Jio | 35.4% |
The 11-point spread between BSNL and Jio is partly subscriber demographics, since BSNL's base skews rural and older, and partly network-side call screening and spam labelling, which the larger operators apply more aggressively. Jio's in-network spam classification is the most active of the four in our data, and campaigns that get labelled see connect rates fall by 40 to 60% within days.
This is not a reason to avoid Jio numbers. It is a reason to monitor per-operator connect rate as a leading indicator: a sudden operator-specific drop almost always means a caller ID has been flagged, and it shows up in the operator split days before it shows up in the campaign aggregate.
Connect rate by caller ID type
The single largest controllable variable in the dataset.
| Caller ID type | Connect rate |
|---|---|
| 10-digit mobile number, local circle match | 48.9% |
| 10-digit mobile number, non-local | 37.1% |
| Landline, local circle match | 33.4% |
| 140-series (telemarketing) | 19.2% |
| Toll-free 1800 | 24.6% |
A 140-series caller ID connects at 19.2% and a circle-matched 10-digit mobile connects at 48.9%. That is a 2.5× difference, larger than the effect of hour, circle, operator or anything else measured here.
The reason is straightforward: the 140 series is reserved for telemarketing under TRAI's framework and Indian consumers have learned to recognise it. It is doing exactly what the regulation intended.
This creates a genuine compliance question rather than an optimisation trick, and it deserves a direct answer. Promotional calling must use the 140 series. That is not optional and dialling promotional content from a 10-digit number to evade recognition is a violation, not a growth tactic. What the data supports is not evasion; it is a clearer separation of call types. Transactional and service calls, order confirmations, delivery updates, appointment reminders, service notifications, are legitimately placed from a normal business number, and many organisations dial them from a 140 series out of caution or because their telephony setup does not separate the streams. Splitting them is compliant and recovers most of the gap.
Circle matching adds another 11.8 points on top and is purely operational. Provisioning caller IDs across circles is work your telephony provider can do; the telephony partner comparison covers which Indian providers make this straightforward.
Connect rate by attempt number
| Attempt | Connect rate on that attempt | Cumulative unique contact |
|---|---|---|
| 1 | 41.2% | 41.2% |
| 2 | 28.6% | 58.0% |
| 3 | 19.4% | 66.1% |
| 4 | 12.1% | 70.2% |
| 5 | 7.8% | 72.5% |
| 6 | 5.1% | 73.9% |
| 7 | 3.4% | 74.8% |
Cumulative contact climbs steeply through attempt three and then flattens hard. Attempts five through seven add 2.3 points of cumulative contact for 43% of the total dialling cost in a seven-attempt strategy.
Three attempts is the right default for most workflows. Four if the contact is high value. Beyond that the marginal cost per additional unique contact exceeds the value of the contact in every workflow in our sample except NBFC collections on accounts over ₹50,000, where the recovery value justifies six.
Attempt spacing matters more than attempt count. Retrying at a different hour block recovered 2.4× more connects than retrying within the same block. Retrying on a different day of the week added a further lift. The worst pattern, and a very common dialer default, is three attempts thirty minutes apart, which mostly re-confirms that the person is busy right now.
Recommended spacing: attempt one in the best block for the segment, attempt two the next day in a different block, attempt three on a different day of week in the first block again.
Connect rate by day of week
| Day | Connect rate |
|---|---|
| Monday | 36.8% |
| Tuesday | 41.2% |
| Wednesday | 42.6% |
| Thursday | 41.9% |
| Friday | 39.4% |
| Saturday | 44.7% |
| Sunday | 31.2% |
Saturday is the best day and is systematically under-used because operations teams do not work weekends. Monday is weak. Sunday is weak and additionally carries a tolerance cost that does not show in connect rate but shows in complaint rate, which ran 2.7× the weekday average.
Number-list hygiene, which sits underneath everything
None of the segmentation above helps if the list is bad, and Indian mobile lists degrade fast.
Churn. Roughly 1.4% of Indian mobile numbers change hands per month in our reconciliation, higher in Tier-3 circles and among prepaid subscribers. A list eighteen months old has lost a fifth of its validity.
Duplicates across channels. Numbers collected from web forms, marketplaces and offline sources routinely duplicate with different formatting. Normalise to E.164 before deduplication or you will dial the same person three times and count it as three contacts.
Invalid and non-existent numbers. Consistently 3 to 6% of raw lists. These fail at the network level and are cheap, but they distort every connect rate metric downward if not excluded from the denominator.
DND and DLT scrubbing. Mandatory, and the scrub must happen at dial time rather than at list-upload time, because registration status changes daily. Scrubbing at upload and dialling a week later is a compliance gap that many otherwise careful teams have.
Cleaning a list typically lifts measured connect rate by 4 to 7 points before any strategy change, partly through real improvement and partly through correcting the denominator.
Putting it together: the dialling strategy the data supports
For a typical Indian outbound campaign:
- Split transactional from promotional streams and dial transactional from a circle-matched 10-digit business number. Largest available lever.
- Provision caller IDs per circle. Second largest.
- Move volume out of the 08:00 to 09:30 block into 11:00 to 13:00 and 16:30 to 20:00.
- Cap at three attempts for most workflows, four for high-value, spaced across different hour blocks and different days.
- Add Saturday to the calling calendar and drop Sunday.
- Re-scrub at dial time and re-validate lists older than six months.
- Monitor connect rate per operator daily as an early warning for caller ID flagging.
- Segment targets by circle economics, not by order value alone.
Applied together on the Gurugram D2C campaign from the opening, connect rate moved from 32% to 54% over six weeks, with no change to the agent, the script or the voice. Order confirmation completion went from 31% to 51%.
For designing the experiments that separate these effects from each other, our A/B testing playbook for Indian voice campaigns covers the test design; this post is the prior you should start from. And once calls do connect, the turn-taking issues covered in the barge-in benchmark become the next constraint.
Why the connect problem is getting harder
Three shifts over the last two years, all pointing the same direction.
Unsolicited call volume keeps rising faster than enforcement. Indian mobile subscribers in metro circles receive more commercial calls per week than at any point measured, and the behavioural response is blanket screening rather than case-by-case judgement. A legitimate transactional call is declined not because it was evaluated and rejected but because it was never evaluated.
Call-blocking apps have moved from power users to defaults. Handset manufacturers now ship spam identification enabled, which removes the adoption barrier entirely. The classification is crowdsourced and lagging, which means a caller ID can be labelled by a small number of reports and stay labelled long after the behaviour that triggered it stopped.
Operator-side classification arrived quickly and quietly. All four networks now apply some form of automated commercial-call labelling. None of them publish the criteria, none offer a straightforward appeals path for legitimate senders at moderate volume, and the effect on connect rate when a number is flagged is severe: a 40 to 60% drop within days in our data, with recovery taking weeks after the underlying pattern changes.
The compounding effect is that caller ID reputation now behaves the way email sender reputation started behaving around 2010. It is an asset that accumulates slowly, degrades quickly, and is difficult to repair. Treating a phone number as a disposable resource, rotating aggressively to escape flags, is the strategy that made email deliverability worse for everyone and it will work no better here.
What this implies operationally. Warm up new caller IDs gradually rather than pointing full campaign volume at a fresh number on day one. Keep volume per number within a range your complaint rate supports. Retire a flagged number rather than pushing more volume through it. And keep transactional traffic on numbers that have never carried promotional traffic, because the reputation damage does not distinguish between the two once it lands.
Measuring this on your own campaigns
The segmentation in this post is only useful if you can reproduce it on your own data, and most dialer reporting will not give it to you out of the box.
What to log per attempt. Timestamp to the minute, the caller ID used, the destination circle derived from the number series, the destination operator, the attempt sequence number for that contact, the disposition from the network, and the campaign and workflow identifiers. Circle and operator require a number-series lookup that your telephony provider can supply and that goes stale, so refresh it quarterly; mobile number portability means the series no longer reliably identifies the current operator, and a portability-aware lookup is worth paying for if operator segmentation is going to drive decisions.
What to compute weekly. Connect rate over dials, segmented by each of the six dimensions in this post, plus complaint rate and per-caller-ID connect rate trend. The per-caller-ID trend is the early warning; everything else is diagnosis.
The mistake to avoid. Comparing connect rates across periods where the list composition changed. Most apparent connect-rate movements in campaign reporting are list-mix effects, not strategy effects: a batch weighted toward metro numbers will look like a strategy failure when it is a sampling difference. Hold the mix constant or segment before comparing, always.
Reconcile against the telephony provider's own figures monthly. Disposition mapping differs between platforms and providers, and a systematic gap between your connect count and theirs usually means voicemail or operator-message answers are being counted as connects somewhere in the chain. That gap flatters every number in your reporting and is worth finding once rather than rediscovering during a quarterly review.
Compliance boundaries
Everything above operates inside the TRAI framework and none of it should be read as a route around it.
Calling windows. TRAI restricts commercial communication to 09:00 to 21:00. The 20:00 to 21:00 block in the tables is inside that boundary; anything later is not, regardless of what connect rate it might produce.
DLT registration. Sender IDs, templates and consent must be registered, and scrubbing against the DND registry happens at dial time.
Consent is purpose-bound under DPDP 2023. Consent obtained for delivery updates does not extend to promotional calling. The transactional and promotional stream split recommended above is a consent boundary as much as a connect-rate optimisation, and treating it only as the latter is how organisations end up with a complaint problem.
Complaint rate is the metric that constrains all of this. Watch it alongside connect rate. A strategy that lifts connects and lifts complaints is not working; it is borrowing against your caller IDs, which will be flagged and will take the connect rate down further than where it started.
What changes in the next twelve months
Operator-side spam classification is getting more aggressive and more automated across all four networks. The practical effect is that caller ID reputation becomes a managed asset: warm-up periods for new numbers, volume ramping, and rotation strategies that were previously the domain of email deliverability are arriving in Indian voice.
Calling Name Presentation, the TRAI-mandated caller name display rollout, changes the arithmetic once it reaches scale. A recognised brand name displayed on an incoming call is likely to lift connect rates for legitimate senders and further depress them for everyone else. Organisations with real brand recognition should expect this to help; the benefit will not be evenly distributed.
Expect the metro-versus-Tier-3 spread to widen as screening technology diffuses from metros outward with a lag.
Bottom line
Connect rate is not a constant and it is not the agent's fault. Across 18.4 million Indian outbound attempts it ranged from 19% to 71% on controllable variables: 2.5× on caller ID type, 25 points across telecom circles, 25 points across hours of the day, and 11 points across operators.
Most teams optimising an outbound voice campaign are working on the conversation. The larger multiplier is upstream, in the dial. Split your transactional and promotional streams, match caller IDs to circles, dial into the two real peaks rather than at the start of the office day, cap at three well-spaced attempts, and watch complaint rate as the constraint on all of it.
If you want the full circle-level and operator-level breakdown, or want your current campaign's connect data segmented this way, talk to our team.
Frequently Asked Questions
Tags :
