← Back to insights
INSIGHTS • 10SPRING

Does Cold Outbound Still Work?

Cold outbound still works for some B2B teams, but not as a volume play. Here is what reply-rate data, sender rules, and buyer research show, and how to test whether your outbound produces pipeline.

Illustration showing cold outbound focused on right-fit accounts and relevant outreach, leading toward qualified B2B pipeline.

Cold outbound has been declared dead many times. Each year brings new data that seems to support both sides. Reply rates look low, inboxes look crowded, and buyers say they would rather research on their own. Yet sales teams still book meetings from calls and emails to people who never asked to hear from them.

The short answer is yes, with conditions. Cold outbound still works for teams that reach a defined set of accounts with a relevant reason, keep their sending infrastructure clean, and follow every reply through to a qualified opportunity. It works poorly as a volume play. The research describes a narrower channel that is less forgiving than it was before the 2024 and 2025 sender rule changes, not an extinct one.

This article covers how to define "works," what the benchmark studies show and where they mislead, how mailbox providers and privacy rules shape outbound, and what buyer research says about being contacted. It then looks at calling, AI-written outreach, and how to diagnose and measure your own motion. It draws on studies from Instantly, Smartlead, Belkins, Hunter, Gong, Digital Applied, and the Bridge Group. It also uses buyer research from Gartner, 6sense, RAIN Group, and G2, plus guidance from Google, Microsoft, the FTC, and the UK Information Commissioner's Office. It notes where each source has limits, including which ones sell tools or services that benefit from a particular answer.

How to define whether outbound works

Most debates about outbound stall on definitions. "It works" can mean five different things.

  • A reply. Any response, which in some studies includes out-of-office notices, objections, and unsubscribe requests.
  • A positive reply. A response that shows interest.
  • A booked meeting. A calendar event, which the prospect may never attend.
  • A held meeting. A conversation that actually took place.
  • A qualified opportunity. A held meeting that sales accepts as a real buying opportunity, plus the pipeline and revenue that follow.

The benchmark studies mostly report the first item. Digital Applied counted any inbound message within 14 days as a reply, including out-of-office notes, unsubscribes, and objections. Belkins divides replies by total emails sent. Smartlead does not say whether negative replies count. These are different measurements, so the numbers do not sit on one scale.

Sales development benchmarks use a later point in the funnel. The Bridge Group's 2025 research covered 351 B2B companies. It reported that 60% of sales development reps were at quota, which it called the lowest on record. It also reported that the median monthly quota for meetings held was 10, down 40% since 2018. Quota design differs from company to company, so that does not prove outbound is failing. It does show that the industry's own benchmark counts meetings held, not meetings booked.

A revenue-first answer to "does it work" sits at the last item on the list. That means qualified opportunities and the pipeline they create, attributed only where reliable data supports it. Everything earlier is a diagnostic signal. It tells you where to look, not whether the motion pays off.

What the cold email benchmarks show

Four large datasets give four very different average reply rates.

The spread, from under 0.5% to over 4%, is the first lesson. It does not mean one study is wrong. The datasets differ in who is counted, what counts as a reply, and what the denominator is. Belkins says it changed its method. It used to calculate reply rates against recipients who opened an email, and it now divides by total sends, so its new figures cannot be compared with its older reports. Smartlead leaves out senders with fewer than 500 emails in the period. Each dataset comes from one vendor's users or one agency's own campaigns, and each publisher sells something that a healthy cold email market supports.

You will also see claims that average reply rates have fallen by half or more in two or three years. Those trend figures usually compare datasets with different definitions. This article does not rely on them.

The datasets agree on more than their averages suggest.

  • Most sends get no reply. At every average above, more than 95% of cold emails go unanswered.
  • Results are uneven. Smartlead's top 10% of senders earned about 3.5 times the typical sender's reply rate. Instantly's top 10% earned roughly three times its average.
  • Short messages and follow-ups travel with better results. Instantly reports that elite senders average under 80 words in a first email and that 42% of replies come from follow-ups. Smartlead found that first emails under 50 words earned a reply every 68 contacts, against every 97 for emails of 100 to 149 words. Hunter found that campaigns with follow-ups averaged 4.9% against 3% for those without, and that unsubscribe rates rise after the third follow-up.
  • Company size and seniority change the baseline. Belkins found reply rates falling as companies grew: 0.72% for companies with up to 10 employees, 0.49% for 11 to 50, and 0.22% for enterprises of 10,000 or more. Founders and owners replied at 0.57%, C-level executives at 0.42%, and vice presidents at 0.32%.

Treat the follow-up and word-count findings as correlations. Smartlead describes its own patterns as not controlled tests, and sender skill, list quality, and offer all vary along with the tactic. The Belkins findings matter most for teams that sell high-ACV products into large accounts. A platform-wide average is the wrong target for them. A team that measures 0.6% against a 3.4% headline may be doing well or badly, and the benchmark cannot say which.

How mailbox rules changed the work

Outbound now runs inside rules written by the companies that operate inboxes.

Google's sender guidelines took effect for Gmail in 2024. They require every sender to authenticate mail with SPF or DKIM. They require senders of 5,000 or more messages a day to Gmail to use SPF, DKIM, and DMARC together. Bulk senders must also support one-click unsubscribe on marketing and subscribed messages. Google's guidance asks senders to keep the spam rate reported in Postmaster Tools below 0.10% and to avoid ever reaching 0.30%. It lists that guidance among the requirements for all senders, not only bulk senders.

Microsoft followed. On April 2, 2025, it announced authentication requirements for domains sending more than 5,000 emails a day to its consumer Outlook.com, Hotmail.com, and Live.com addresses. Enforcement began May 5, 2025. Microsoft said non-compliant messages would first go to Junk and might eventually be rejected, and it updated the post on April 29 to change the action taken. Read the current policy page before relying on any summary.

Postmaster Tools reports user-reported spam on mail sent to personal Gmail accounts. A team that mostly writes to corporate mailboxes sees only part of its real complaint picture. A clean dashboard is therefore not complete evidence of a clean reputation.

Four points follow for B2B outbound.

  • The rules do not ban cold email. They make poor targeting and poor list hygiene visible and costly.
  • Volume thresholds are not safe harbors. Many teams spread sending across several mailboxes so that no domain nears 5,000 messages a day. Smartlead's data shows the typical mailbox sends about 14 cold emails a day. The spam-rate guidance applies to every sender, so complaints matter at any volume.
  • List quality shows up in bounce rates. Smartlead treats a 2% bounce rate as a danger line and reports that a quarter of senders sit at 3.21% or higher. That suggests list quality, not copy, is a common failure.
  • Authentication shows a modest link to results. Smartlead reports that senders with SPF, DKIM, and DMARC needed 108 contacts per reply, against 120 for senders without them. That is a small difference from a vendor's own data, not a controlled test. It is a reminder that deliverability is a precondition. It is not a growth lever on its own.

Legal and platform limits to design around

Cold outreach is lawful in many places under conditions. This section is general information, not legal advice. Rules differ by country and by channel, so involve counsel before building a program.

  • United States. The CAN-SPAM Act works on an opt-out basis. The FTC says it covers more than bulk email and applies to business email as well. Every commercial message needs a working opt-out. Requests must be honored within 10 business days, and each violating email can draw penalties of up to $53,088.
  • United Kingdom. Under the Privacy and Electronic Communications Regulations, the consent rule for marketing email does not apply to corporate subscribers such as companies and limited liability partnerships. It does apply to sole traders and some partnerships, which are treated as individuals. Live marketing calls are covered for business recipients too. Data protection law can also apply when an address identifies a named person.
  • Canada. The anti-spam law is stricter, and a sender cannot treat an address as fair game just because it is public. In a 2016 decision, the CRTC rejected that argument and imposed a $50,000 penalty. A 2020 Federal Court of Appeal ruling said the conspicuous publication rule does not permit collecting addresses from third-party directory websites.
  • LinkedIn. LinkedIn says it does not allow third-party software or browser extensions that scrape, modify, or automate activity on its site. Automated connection and messaging tools work against that policy.

The compliance work belongs inside the outbound system, not beside it. That means suppression lists that sync across tools, opt-outs that reach every sequence, and a record of where each contact's data came from. It is a revenue operations task as much as a legal one.

What buyers say about being contacted

Buyer research points in two directions at once.

Buyers say they want fewer seller interactions. Gartner's June 2025 release reported that 61% of 632 B2B buyers, surveyed in 2024, prefer a rep-free experience. It also reported that 73% actively avoid suppliers that send irrelevant outreach. Its March 2026 release put the rep-free figure at 67%, among 646 buyers surveyed in August and September 2025. These are stated preferences from press releases. Neither release publishes question wording or a margin of error, and Gartner sells research on the topic.

Many buyers still accept outreach. RAIN Group's study of 488 buyers and 489 sellers found that 82% of buyers accept meetings at least sometimes with sellers who reach out. It found that 71% of buyers who accept meetings want to hear from sellers early in the buying process. The study is older, with its press release dated January 2018, and RAIN Group sells sales training.

Relevance is the common complaint. Hunter's survey of 217 decision-makers found that 61% preferred cold email as an outreach channel, ahead of LinkedIn at 29% and cold calls at 10%. It also found that 71% named lack of relevance as the main reason they do not reply. The sample is small, and Hunter sells cold email tools.

These findings do not contradict each other as much as they appear to. Buyers dislike irrelevant contact and accept relevant contact. Gartner's 2025 release says buyers favor self-service for general information but want seller input on tasks such as judging whether a product fits. A separate Gartner survey, summarized by Demand Gen Report, found that 69% of buyers prefer to validate AI-generated insights with sales reps. A rep-free preference describes how buyers want to research. It does not say they will refuse a message that helps.

Shortlists often form before outreach arrives. The 6sense 2025 Buyer Experience Report drew on just under 4,000 buyer responses. It found that the winning vendor was on the buyer's Day One shortlist 95% of the time and that four in five deals went to the buyer's favorite before contact. It also found that buyers had prior personal experience with at least one shortlisted vendor 97% of the time. On average, buyers did not engage sellers until about two-thirds of the way through their journeys. 6sense sells intent and sales intelligence products, the data is self-reported, and the page gives no fieldwork dates.

The implication is not that outbound is pointless. It is that outbound competes against preferences that are already forming. A message that arrives while a buyer builds a shortlist, from a company the buyer has already seen, has a different job than a message that arrives after a favorite is set.

Buyers also check you out before they reply. G2, a review marketplace, reports that 51% of the 1,076 B2B software buyers it surveyed say they start with an AI chatbot more often than with Google. RAIN Group reported that 82% of buyers look up a provider on LinkedIn before replying to outreach. Gartner reported that 69% of buyers see inconsistencies between a seller's website and what sellers tell them. A prospect who researches you in Google Search or in AI Search Engines will see the same pages as everyone else. If those pages contradict the email, the email loses.

Where cold calling fits

Calling data is thinner and mostly held by vendors. Gong's analysis of more than 300 million cold calls, published in July 2024, found an average connect rate of 5.4% and a top-quartile rate of 13.3%. On connected conversations, the average rep set a meeting 4.6% of the time, against 16.7% for top-quartile reps. Gong also reported email reply rates of 3.44% for prospects who were also called, against 1.81% for those who were not. That is an association, not proof that the call caused the replies, because the same targeting may drive both. Gong sells conversation analytics software and promotes a book on cold calling. The page does not state the collection period or sample details.

Other data points the same way. The Bridge Group's 2025 report, as compiled by CloudTalk, found that the median sales development rep logs 112 activities a day: 44 on the phone, 41 by email, and 19 on LinkedIn. Outreach, a sales engagement vendor, reports from its own data a 7% response rate for multichannel sequences against 5% for email-only ones. These sources suggest outbound works better as one coordinated motion than as separate channels. That is a hypothesis to test, not a rule.

What AI-written outreach changes

AI tools made it cheap to send more. The evidence on whether they make outreach better is thin and mixed.

Digital Applied, a marketing agency, published a 100,000-email analysis in April 2026. It paired 50,000 AI-written emails with 50,000 human-written ones, drawn from Smartlead, Instantly, Apollo, and its own sources. It reported reply rates of 4.1% for AI and 5.2% for human, and positive replies of 1.4% and 2.1%. It reported meetings booked at 0.7% and 1.1%, and spam flags at 8% against 3%. The agency sells AI outreach services. It gives no confidence intervals, and the dataset is not public, so no one can check the findings independently.

Hunter's survey points to a perception gap. Its AI cold email guide draws on 200 decision-makers and 300 B2B professionals. In it, 47% of B2B professionals said they would be less likely to reply if they thought an email was AI-written. Yet 67% of decision-makers said they do not mind AI use. Hunter's data suggests people cannot reliably tell the difference and that recipients judge messages on relevance and credibility.

Taken together, these sources do not show that AI-written outreach is better or worse. They suggest the risks sit in volume, deliverability, and sameness. A reasonable working hypothesis is that AI helps most with research and preparation and least with the final voice. Test that inside your own motion, because the strongest figures come from sellers of AI outreach and from agencies.

How to diagnose your own outbound

Before changing copy or buying more tools, find where the motion breaks. The order matters, because each layer limits the next.

  1. Infrastructure. Confirm that SPF, DKIM, and DMARC pass and align for every sending domain. Read bounce rates and spam complaints. If messages do not reach inboxes, no copy test will tell you anything.
  2. List and account selection. Compare the accounts you contact with the accounts that have closed. Ask which segments produced qualified opportunities, not just replies. Remove contacts that do not match.
  3. Message and offer. Check that each message gives a specific reason for contact tied to the recipient's role. Gong's cold call statistics report that callers who state their reason for calling book about 2.1 times more often. That is also a correlation, but it fits what buyers say about relevance.
  4. Sequence and channel design. Decide how email, calls, and LinkedIn work together within each platform's rules. Set a stop rule for contacts who do not respond. Sources disagree on how many follow-ups help, so test the number.
  5. Handoff and follow-through. Track the time from reply to meeting, the show rate, and whether held meetings become qualified opportunities. Count meetings held, as the Bridge Group does.
  6. Consistency with what buyers see elsewhere. Compare outreach claims with your website, review profiles, and what AI Search Engines say about you. Any gap between them is something the buyer will find.

Fix problems in this order. If infrastructure is failing, nothing downstream can be measured. If the list is wrong, better copy only reaches the wrong people faster.

How to measure without overstating

Outbound measurement works best in three layers.

  • Layer 1: activity and infrastructure signals. Authentication status, bounce rate, spam complaints, reply rate, and positive reply rate. These are diagnostic.
  • Layer 2: conversion performance. Meetings held, show rate, and qualified opportunities.
  • Layer 3: business impact. Pipeline, new customers, and recurring revenue, attributed only where reliable data supports it.

Attribution has a specific trap here. 6sense reports that buyers usually have a favorite before first contact. When an outbound meeting becomes a deal, the deal may reflect a preference formed earlier through a referral, content, reviews, an event, or prior experience. Crediting the sequence alone overstates outbound. Crediting it with nothing understates it. Record the first touch and the last touch, and ask on the first call how the buyer found you.

Set a baseline before changing anything. Compare results over quarters, not weeks, because long sales cycles delay outcomes. Compare against your own history and your closed-won patterns, not against vendor averages from other senders in other conditions. No one can guarantee meetings, pipeline, or revenue from outbound.

Where this leaves cold outbound

The evidence supports a conditional yes. Outbound tends to work better when:

  • The list is narrow and resembles the accounts that have closed before.
  • The reason for contact is specific to the recipient's role and situation.
  • Sending infrastructure is authenticated and monitored.
  • Follow-up is coordinated across channels and has a stop rule.
  • Sales follows up quickly and tracks meetings held and qualified opportunities.
  • The website and other public sources back up the claims in the message.

It tends to work worse when:

  • The list is bought, stale, or broad.
  • Messaging is generic and volume is the strategy.
  • Teams measure replies and ignore opportunities.
  • Automation breaks platform or legal rules.
  • Outbound is asked to create recognition from zero in a market where buyers shortlist early.

Outbound also does not replace the rest of the system. Where buyers decide on a Day One shortlist, being known before the first message matters. That is an argument for coordinating outbound with Account-Based Marketing, content, reviews, and search presence. It is not an argument for abandoning outbound.

Practical takeaways

  • Define "works" at the opportunity. Replies and booked meetings are diagnostic. Qualified opportunities and pipeline are the result.
  • Do not borrow a benchmark as a target. Published reply rates range from under 0.5% to over 4% because definitions and populations differ.
  • Fix infrastructure first. Authenticate every sending domain, watch bounces and spam complaints, and remember that Postmaster Tools covers only personal Gmail accounts.
  • Shrink and sharpen the list. Match contacts to the accounts and roles that have closed before, and expect lower reply rates in larger companies.
  • Lead with relevance. Buyers name irrelevant outreach as the main problem, so state a specific reason for contact.
  • Know the rules where you send. CAN-SPAM, the UK regulations, Canada's anti-spam law, and LinkedIn's policies differ, and counsel should review the program.
  • Test AI inside your own motion. The public evidence is mixed and comes mostly from sellers of AI tools and from agencies.
  • Coordinate outbound with the rest of the system. Buyers form shortlists early, check your website, and compare what they find with what you say.
  • Measure in layers and set a baseline. Judge the motion on qualified opportunities, pipeline, and customers, and compare over quarters.

10spring works alongside internal marketing teams. If you want to pressure-test your outbound motion against the evidence in this article, Book A Call. Let's accelerate your pipeline together.

Sources and further reading

All sources accessed November 13, 2026.