Cold calling statistics are unusually contaminated. Most of the numbers circulating in 2026 come from vendor blogs quoting other vendor blogs, and a large share trace back to a single study from before 2015. Where two credible sources disagree, this page shows both and says why.
Every figure below names its source, its sample size where the source discloses one, and whether we reached the primary study or only a secondary report of it. Three dial-volume figures are our own editorial estimates and are labelled as such; they are not measurements.
Updated 2 October 2026: three figures this page credited to Saleshandy (1.55 attempts, the 2.7 percent success rate and the call length) are Cognism's, Cognism's best day is Thursday, and the voicemail figures now come from Gong. The details are on /corrections.
Connect rates
| Metric | Figure | Source |
|---|---|---|
| Connect rate per dial, direct numbers | 7.9% (12.73 dials per connect) | Vorsight, quoted by Gartner (2019); no year or sample given |
| Connect rate per dial, switchboard numbers | 5.3% (18.83 dials per connect) | Vorsight, same source |
| Connect rate per dial, 175,000+ B2B dials | 9.9% | Belkins, dials logged in 2025 |
| Prospects reached at least once, about 3 attempts each | 24.5% | Belkins, same dataset |
| Connect rate, verified numbers | 44.63% | Saleshandy, 7,699 calls, July to August 2026 |
| Connect rate, older aggregate figures | under 5% | Cognism aggregate, 2025 |
The spread here is the most useful fact on this page. A connect rate is mostly a property of your data, and only marginally of your technique. Saleshandy's 44.63 percent on verified numbers against Belkins' 9.9 percent across an agency's lists is a fourfold gap, and no opener closes it. The cleanest number-type comparison, Vorsight's, is smaller: direct numbers connect about 1.5 times as often as switchboards.
Until 4 October 2026 this table also listed 8 to 12 percent on generic data and 18 to 22 percent on verified mobile direct dials, credited to SkipCall. SkipCall, a dialer vendor, credits those figures to Cognism's 200,000-call report, which contains no connect-rate figure, so we removed them. The change is on /corrections.
Belkins' two figures are the same calls counted two ways. Per dial, about one in ten connects. Per prospect, about one in four eventually picks up. A rep who stops after one dial gives up most of the gap between those two numbers.
If your connect rate is under 3 percent, buy better data before you buy a better dialer.
Success and conversion rates
| Metric | Figure | Source |
|---|---|---|
| Meeting-booked rate per dial, 2025 | 2.3% | Cognism, State of Cold Calling 2025, 204,000+ calls |
| Meeting-booked rate per dial, 2026 | 2.7% | Cognism, State of Cold Calling 2026, 200,000+ calls |
| Healthy conversation-to-meeting rate | 10 to 20% | MRR Nerds, from operator benchmarks |
| Meetings held vs booked, healthy | over 70% | MRR Nerds, from operator benchmarks |
The distinction between per-dial and per-conversation conversion is where most teams misdiagnose themselves. A rep at 2 percent dial-to-meeting with a 6 percent connect rate is converting one in three conversations, which is very good. The same 2 percent with a 20 percent connect rate is converting one in ten, which is a script problem.
Measure per conversation. Per dial mostly measures your list.
Dial volume
| Metric | Figure | Source |
|---|---|---|
| Average SDR dials per day | ~52 | Bridge Group, via SkipCall |
| Share of reps making 50+ dials/day | ~30% | Cognism, 2025 |
| Dials per hour, manual | 15 to 25 | SkipCall, 2026 |
| Dials per hour, power dialer | 30 to 50 | SkipCall / Belsmart, 2026 |
| Dials per hour, parallel dialer | 150 to 300 | Vendisys / Belsmart, 2026 |
| Dials per day, manual from a spreadsheet | 35 to 50 | MRR Nerds editorial estimate, not measured |
| Dials per day, single-line power dialer | 120 to 200 | MRR Nerds editorial estimate, not measured |
| Dials per day, parallel dialer, 3 to 4 lines | 300 to 600 | MRR Nerds editorial estimate, not measured |
Vendor figures for parallel dialing run higher than ours because they measure a clean list under ideal conditions. Mojo, for example, publishes 85 calls an hour on single line against 300 on triple line, which is a threefold multiplier that assumes no dead numbers and no research between calls.
The number that matters is conversations. At 150 dials and a 6 percent connect rate a rep has nine conversations. At 450 dials on four lines with the same connect rate, less the ones lost to connect delay, they have roughly 22.
Attempts and persistence
| Metric | Figure | Source |
|---|---|---|
| Average attempts to reach a prospect, 2025 data | 2.9 (Cognism's own page rounds it to 3) | Cognism, 204,000+ calls |
| Share of conversations happening by attempt 3 | 93% | Cognism, 2025 report |
| Average attempts to reach a prospect, 2026 report | 1.55 | Cognism, 200,000+ calls |
| Attempts per prospect before a connection | 3.51 average, median 2 | Saleshandy, 7,699 calls, 2026 |
| Attempts per prospect | about 3 | Belkins, 175,000+ dials, 2025 |
| Touches across all channels before first meeting | 8 | RAIN Group |
| Attempts most reps make before stopping | 2 | Brevet Group, secondary, sample not disclosed |
The 2026 drop from 2.9 to 1.55 is the largest year-on-year move in any figure on this page. Cognism does not publish how it counts an attempt or whether the same teams are in both years, so read it as one vendor's change in its own data until a second dataset shows the same thing. Two independent 2026 datasets sit nearer three: Saleshandy at 3.51 and Belkins at about three.
Saleshandy's median of 2 against an average of 3.51 says the distribution has a long tail. Half the prospects who connect do so by the second dial, and a minority take many more, which pulls the average up.
RAIN Group's eight touches counts email and LinkedIn as well as calls, so it measures the road to a meeting. The call figures measure the road to a conversation.
The actionable fact is Cognism's 93 percent. Most reps stop at two, and the third call is where the remaining conversations happen. By the same figure, every call from the fourth onward shares the last 7 percent of conversations.
Buyer receptivity
| Metric | Figure | Source |
|---|---|---|
| Buyers who accept meetings with sellers who reach out | 82% | RAIN Group |
| Buyers who accepted a call from a new provider in the past year | 69% | RAIN Group |
| C-level and VP buyers who prefer phone contact | 57% | RAIN Group |
The seniority finding is counterintuitive and consistently replicated: the more senior the buyer, the more they prefer the phone. Junior buyers hide behind email. Executives take calls, because their time is scheduled by other people and a phone call is a faster way to dispose of a question than a thread.
Timing
| Metric | Figure | Source |
|---|---|---|
| Best hour | 4 to 5pm local | CallHippo, 50,000+ calls |
| Second-best hour | 8 to 9am local | CallHippo |
| Best hour | 10 to 11am, then 2 to 3pm | Cognism, 2025 and 2026 reports |
| Best day | Wednesday, 48% better than Monday | CallHippo |
| Best day | Thursday, then Tuesday | Cognism, 2026 report (Wednesday second in 2025) |
| Best days | Wednesday and Thursday | Gong |
The studies disagree on the day and broadly agree on the shape: late morning and late afternoon beat the rest, and the middle of the week beats the ends.
The variable that matters more than either is time zone. A rep in New York dialing California at 9am is calling people who are asleep, and no study corrects for how often that happens.
Voicemail
| Metric | Figure | Source |
|---|---|---|
| Connect rate on later dials, no voicemail left | 7.18% | Gong Labs, 2024, stated 300M+ calls |
| Connect rate on later dials, after a voicemail | 5.17% (28% lower) | Gong Labs, 2024 |
| Email reply rate, no voicemail | 2.73% | Gong Labs, 2024 |
| Email reply rate, after one or two voicemails | 5.87% | Gong Labs, 2024 |
| Email reply rate, after three or more voicemails | 2.2% | Gong Labs, 2024 |
| Optimal voicemail length | under 20 seconds | MRR Nerds, from operator benchmarks |
| Average cold call length | 82 seconds (2026), 93 seconds (2025) | Cognism |
Gong's figures make a voicemail an email play. One or two more than double the reply rate on the email that follows, from 2.73 to 5.87 percent, and the connect rate on later dials falls. The widely repeated 4.8 percent callback figure attributed to Marchex is left out because we could not find the study it comes from. Scripts built on Gong's numbers are in the voicemail guide.
Cost
| Metric | Figure | Source |
|---|---|---|
| US SDR base salary, 2026 | $55,000 to $60,000 | Visdum / Trellus, 2026 |
| US SDR on-target earnings, 2026 | $83,000 to $85,000 | Winsabove, 2026 |
| Enterprise SDR OTE | ~$95,000 | Visdum, 2026 |
| SMB SDR OTE | ~$75,000 | Visdum, 2026 |
| Average quota attainment | 47% | Forrester |
| Twilio US outbound, per minute | $0.0140 | Twilio pricing, verified Sept 2026 |
| Twilio US local number, per month | $1.15 | Twilio pricing, verified Sept 2026 |
The last two rows are the ones that reframe every software decision on this site. A fully loaded SDR costs roughly $7,000 a month. The telephony to keep them dialing 150 numbers a day costs about $16. Every dollar between those two numbers is software margin, and it is worth knowing which vendors take how much of it.
Compliance exposure
| Metric | Figure | Source |
|---|---|---|
| TCPA statutory damages per call | $500 | 47 U.S.C. § 227 |
| Trebled damages for willful violations | $1,500 per call | 47 U.S.C. § 227 |
| Share of TCPA suits that are class actions | ~80% | Leadgen Economy analysis |
| Predictive dialer abandonment cap | 3% | FCC TCPA rules |
| STIR/SHAKEN implementation deadline | 30 June 2021 | FCC, TRACED Act |
The absence of a damages cap is the fact that should govern any decision about consumer dialing at volume. Ten thousand improperly placed calls is a five-million-dollar theoretical exposure. This is orientation, not legal advice.
How to use these numbers
Benchmark against the right cohort. B2B office-hours dialing connects far better than consumer lists. Comparing your insurance desk to a SaaS SDR's numbers will tell you nothing.
Measure per conversation, not per dial. Per-dial metrics mostly measure your data.
Fix the list before the script. The connect rate spread between generic and verified data is larger than any technique effect on this page.
Do not trust a statistic without a date. A large share of the cold calling numbers circulating in 2026 originate in studies from 2012 to 2015 and have been re-quoted so many times the origin is lost. If a source will not date its figure, treat it as decoration.
Frequently asked questions
- What is the average cold call connect rate?
- Per dial, about 10 percent: Belkins' 175,000+ B2B dials logged in 2025 connected on 9.9 percent. Vorsight, quoted by Gartner in 2019, put direct numbers at 7.9 percent per dial and switchboards at 5.3 percent. Saleshandy's 7,699 calls against verified numbers in July and August 2026 reached 44.63 percent, which makes data quality the single largest variable in connect rate. Per prospect the number is higher: Belkins reached 24.5 percent of prospects at least once across about three attempts.
- What are good cold call connect rate benchmarks per prospect?
- Belkins' 2025 data is the clearest published benchmark: 9.9 percent of 175,000+ dials connected, and 24.5 percent of unique prospects were reached at least once across about three attempts each. So a list where fewer than one prospect in five ever picks up is underperforming that benchmark, even if the per-dial connect rate looks normal. Per-dial rates mostly measure the list; per-prospect rates also measure how many attempts a rep made.
- What is a good cold calling success rate?
- Cognism's State of Cold Calling reports put the industry average for booking a meeting at 2.7 percent of dials in 2026, up from 2.3 percent in 2025, across more than 200,000 calls each year. Measured against conversations rather than dials, a healthy conversation-to-meeting rate is 10 to 20 percent. The distinction matters: a rep with a 2 percent dial-to-meeting rate and a 6 percent connect rate is converting a third of their conversations, which is excellent.
- How many cold calls should an SDR make per day?
- The average SDR makes about 52 dials a day, and roughly 30 percent of reps make 50 or more. What is achievable depends on tooling: a rep dialing manually from a spreadsheet completes 35 to 50, a single-line power dialer takes the same rep to 120 to 200, and a parallel dialer on three or four lines reaches 300 to 600 against a clean list.
- How many attempts does it take to reach a prospect by phone?
- Between 1.5 and 3.5 attempts, depending on whose data you read. Cognism's 2025 report of 204,000+ calls found 2.9 attempts on average, with 93 percent of conversations happening by the third call, and its 2026 report puts it at 1.55. Saleshandy's 7,699 calls in 2026 averaged 3.51 attempts per prospect with a median of 2, and Belkins' 175,000+ dials about three. A first meeting takes about eight touches across all channels, per RAIN Group.
- What is the best time of day to cold call?
- The two windows with the strongest evidence are 10 to 11am and 4 to 5pm in the prospect's own time zone. CallHippo's analysis of over 50,000 calls found 4 to 5pm the best hour and Wednesday the best day, with 48 percent higher first-attempt connect rates than Monday. Cognism's 2026 report of 200,000+ calls points to 10 to 11am and Thursday. The studies disagree on the day and broadly agree that late afternoon and late morning beat everything else.
- How much does an SDR cost?
- In the US in 2026, an SDR's base salary averages $55,000 to $60,000 with on-target earnings around $83,000 to $85,000, and a median split of roughly $60,000 base plus $25,000 variable. Enterprise SDRs average around $95,000 OTE and SMB SDRs around $75,000. Forrester puts average quota attainment at 47 percent, which is the number that should frame any tooling purchase.
How to cite this page
Abhi Chawla. "Cold calling statistics for 2026, with every source named." MRR Nerds, 2026-10-04. https://www.mrrnerds.com/guides/cold-calling-statistics. Accessed 2026-10-16.
The underlying vendor data is published at /data under CC BY 4.0. Cite the dataset for numbers, this page for the analysis.