More than half of full-time professionals throw away cold business messages without reading a word, and a further 30% glance at them before binning them, according to a survey of 1,009 workers that sales data company Firmable published on August 19, 2026. The same study found that 54% of respondents send cold outreach themselves.

In Short

A company that sells sales data asked 1,009 office workers what they do with unsolicited business messages, and most said they get rid of them almost straight away. That matters to anyone selling to businesses, because many of the people throwing these messages away send them too, and a third use AI to help write their own even though most distrust the AI-written messages they receive. What decides a reply, by their own account, is whether a message is relevant and useful, far more than whether a person typed it.

A reflex, not a decision

Firmable, a sales intelligence company selling to teams in the United States, Canada and Asia-Pacific, built its report around one question: what is a professional's most common reaction to a cold message? The answers fall into five bands. According to Firmable, 52% delete the message without reading it, 30% skim and then delete, 11% read but rarely reply, 6% reply if the message is relevant and 1% almost always reply. The five bands add up to exactly 100%.

The headline figure of 82% combines the first two groups. That distinction matters more than it might appear. When Firmable shared the study with PPC Land on October 2, 2026, its summary described all 82% as deleting cold outreach on sight, dropping the skim category - yet 30 points of that total involve at least a look at the content.

Women were more likely than men to discard messages on sight, at 85% against 77%. Among government and nonprofit workers, the share who bin outreach without opening it climbs to 61%, nine points above the overall figure.

Volume goes some way to explaining the behavior. In a typical week, 44% of respondents receive six or more cold messages and 16% field 16 or more, according to Firmable. Gen X carries the heaviest load: 60% of that cohort receive six or more a week, against 40% of millennials and 38% of Gen Z. Three quarters of respondents (75%) have blocked, muted or reported a sender for repeated outreach. And 38% have replied to a stranger purely to make the messages stop - a figure that sits awkwardly with any sales dashboard that counts every reply as engagement.

What gets a message binned

The survey asked which signals instantly mark a message as not worth the reader's time. Respondents could pick more than one answer; the six options in the report's chart total 203%. Looking obviously mass-sent was the most common trigger, cited by 63%. A generic "Hi {FirstName}" greeting followed at 37%, then getting the recipient's name, role or company wrong at 36%. A fake compliment put off 27%. Stock phrases such as "just following up" or "circling back" were a dealbreaker for 20%, the same share that objected to messages running too long.

The ordering is telling. A broken mail-merge field is a visible failure of automation, while looking mass-sent is a judgment about the message as a whole. Recipients penalize the impression of volume more heavily than any single technical slip.

The people who complain are often the senders

Here the report turns to its central theme. "The answer is a double standard," it states. More than half of respondents (54%) send cold outreach themselves, while 46% say they never do, according to Firmable.

The chart on self-assessment splits the full sample four ways: 46% never send cold messages, 28% hold themselves to a higher standard than they expect from others, 22% to about the same standard and 4% to a lower one. Those four bars sum to 100% of all respondents. The report's text, however, describes the 22% and 4% figures as shares "among those who send", which the chart does not support. Recalculated over the 54% who do send, about 41% of senders hold themselves to roughly the same standard and about 7% to a lower one. Put differently, close to half of all senders (48%) claim no higher bar for their own outreach than the one they apply to incoming messages. The report's key takeaways express the same data as 26% of all professionals.

A second inconsistency concerns who sends. Gen Z is described as the most active group, with 28% having cold-messaged a stranger in the past year for a job, vendor, introduction or favor, against 21% of Gen X and 18% of millennials. Weighted by the sample's generational mix, those figures imply that roughly one respondent in five cold-messaged a stranger in the past year - well short of the 54% who say they send cold outreach. The report does not define what the broader figure covers, whether a longer time frame, messages sent as part of a job, or other channels.

When outreach fails, recipients mostly blame the content. An irrelevant offer was the most common reason given for not replying, at 28%, followed by messages that were generic or templated (20%), not being the buyer (18%), a sender who did not understand the recipient (14%) and an aggressive or pushy tone (12%). Speed has a cost as well. Some 28% of respondents have turned down a cold message and later realized it was relevant, rising to 32% among Gen Z.

The AI double standard

The sharpest contrast in the data involves AI. According to Firmable, 61% of respondents distrust the AI outreach they receive, but 33% use AI to help write their own cold or professional messages. Use is higher in technology (47%) and financial services (43%), and men rely on AI for drafting more than women, at 39% against 30%.

That 33% covers "cold or professional messages", a broader category than cold outreach alone, so it cannot be read as the share of cold messages written with AI. Data from the seller's side runs in the same direction. A separate Firmable survey of 222 B2B sales professionals, covered on September 26, found 45% using AI to draft outreach and an estimated 32% of their CRM data flawed. The two samples differ - one made up of sellers, the other of professionals at large - and the figures are not directly comparable.

Fake, not machine-made

The report argues that hostility toward AI is a proxy for something else. "What they reject is outreach that feels fake, whether it was written by a human or a machine," it states.

AI-generated video is the clearest test case. A full 61% of respondents said an AI-generated "personalized" video - the report gives Loom as an example - feels fake and makes them less likely to engage; only 7% found it impressive. Gen Z was the most skeptical generation at 70%, followed by millennials at 61% and Gen X at 53%. That is the reverse of the pattern often assumed for younger, digitally native workers. Technology workers were the outlier: 20% found such videos impressive, nearly three times the overall rate, though 49% in that sector still called them fake.

Two-thirds of respondents (67%) said they can spot AI-written outreach and ignore it, rising to 75% in healthcare. The survey measured confidence, not accuracy, and no detection test is described. The gap matters. A Raptive study of 3,000 US adults, published in July 2025, found readers rated identical content worse when they merely believed it was AI-generated, with trust falling by half.

What, then, earns a reply? Only 13% said human authorship counts most when deciding whether to respond, against 58% who prioritize relevance or genuine value, according to Firmable. The chart on reply triggers ranks offering something genuinely useful - data, an introduction or a resource - first at 26%. Messages that clearly reflect real research on the recipient drew 10%, extreme brevity 9%, pointing out a problem or opportunity 9%, a reference to something recent about the recipient 8% and a clear discount or deal 8%. Price incentives sit at the bottom of the list.

Two things in that chart go unexplained. The six options sum to 70%, leaving 30% of responses unaccounted for. And the 58% relevance figure does not map cleanly onto any of the listed categories, so how it was derived - from a separate question or a grouping of answers - is not stated.

The pattern resembles findings outside B2B sales. Direct mail company Lob found in a survey of 2,000 US adults, covered in September, that 56% value useful offers over seeing their first name printed, while 62% said the growth of AI had made them warier of digital ads.

Thin on method

Firmable's methodology note runs to three sentences. It says the company surveyed 1,009 full-time professionals about cold B2B outreach in 2026 and gives two demographic splits: 62% women, 36% men and 2% who identified another way or preferred not to say; 17% Gen Z, 60% millennials, 20% Gen X and 2% baby boomers. The generational shares sum to 99%, presumably through rounding.

Much is missing. The report does not state when fieldwork took place, which countries respondents live in, how they were recruited, whether results were weighted, or the exact wording of questions beyond chart titles. No margin of error is given. A random sample of 1,009 would carry a margin of roughly plus or minus three percentage points at 95% confidence, but without recruitment details that figure cannot be assumed. Sector findings for healthcare, technology, financial services, and government and nonprofit come without subgroup sizes, so their reliability cannot be judged.

Sample composition shapes the headline numbers. Women make up 62% of respondents and were eight points more likely than men to discard outreach on sight, so the overall 82% leans toward their answers. Millennials account for six in ten respondents. The report does not say whether either skew was corrected. Similar gaps were flagged in Firmable's sales-team survey, which likewise gave no fieldwork dates, countries or margin of error.

Even the name varies. The published page is titled "The cold outreach pulse report", while the summary Firmable shared on October 2 referred to it as the Cold Outreach Mood Report.

The commercial context is plain. Firmable describes itself as "the AI-native sales intelligence platform for B2B sales teams," supplying prospect data, buying signals and agent-driven actions, and says its technology continuously refreshes a proprietary map of the market. The report, bylined to Chester Thompson, Firmable's vice president for the Americas, closes on a conclusion that fits that product: senders who still get replies are those who arrive with the right context about the recipient. "None of this means cold outreach is dead. It means the bar moved," the report states. The survey does not test that claim against reply rates. It measures stated attitudes.

Why the inbox numbers matter to marketers

The findings land in a market where both the volume of messages and the tools for producing them keep growing. Germany's digital industry association BVDW, in an email marketing guide published in March, cited Radicati Group estimates of more than 376 billion emails sent and received daily in 2025, with more than 393 billion projected by 2027. The same guide described AI agents increasingly sitting between sender and recipient, deciding whether a message is forwarded, prioritized or discarded. Firmable asked people about their own habits; it did not ask how much of the sorting is already delegated to filters or assistants.

Mailbox providers have tightened too. Since February 1, 2024, Gmail has required bulk senders to authenticate mail with SPF, DKIM and DMARC, offer one-click unsubscribe and keep spam rates below 0.3% as measured in Postmaster Tools. Spam reports are the input that threshold is built on. The survey's 75% figure includes reporting, though it does not separate reporting from blocking or muting, nor email from LinkedIn and other channels.

In the United States, the CAN-SPAM Act makes no exception for business-to-business email. An allegation in May that Google Ads representatives sent advertisers commercial emails without compliant opt-outs put the potential penalty at up to $53,088 per violating message.

In Europe, the pressure falls on the data that makes personalization possible. Italy's data protection authority fined contact-data provider Lusha 2 million euros in a decision dated July 14, 2026, gave it 60 days to erase the Italian contacts it holds and rejected its reliance on legitimate interest, noting that Italian law requires consent for third-party marketing use of contact details. Nine days before that decision, HubSpot withdrew terms for a Contact Discovery feature built to pool customer CRM records into a shared data enrichment set, four days after publishing them on July 1.

That is the tension the Firmable data exposes without resolving. Recipients reward research - 36% reject messages that get a name, role or company wrong, and 10% name visible research as the thing most likely to earn a reply. Yet the data that feeds such research is under legal and commercial scrutiny on both sides of the Atlantic.

The tooling race

AI tools for sellers are multiplying. Salesforce and Anthropic put 37 prebuilt sales skills inside Claude for select pilot customers on August 26, a week after Firmable published its findings. A banner on the copy of Firmable's report page supplied to PPC Land advertises an MCP connection that lets its product work inside Claude, ChatGPT and Copilot, which the company says takes two minutes to set up without a developer. B2B data company Channel99 connected its marketing data to ChatGPT, Copilot and Claude through the same protocol in February. The prospecting platforms themselves are contested: a trade association and software developers filed an antitrust suit against LinkedIn on September 14, alleging that Sales Navigator customers were pushed into pricier tiers with a 10-license minimum.

Disclosure and the authorship question

Research on advertising sentiment points the same way as Firmable's generational data. In January, IAB research found 39% of Gen Z held negative views of AI-generated ads, nearly twice the 20% recorded among millennials. And an NYU Stern study cited in the IAB's AI disclosure framework, detailed on August 18 - a day before Firmable published - found that labeling generative AI involvement cut ad click-through rates by 31.5%.

Those studies measure reactions to advertising; Firmable's measures stated preferences about one-to-one business messages. They converge on an awkward question. If only 13% of professionals say human authorship decides a reply, why do 61% distrust AI outreach and 67% claim to ignore it on sight? Firmable's own ranking offers a partial answer: the leading dealbreaker, at 63%, is the look of a mass-sent message, whoever or whatever produced it. The survey cannot say whether AI-assisted messages actually earn fewer replies. That would require behavioral data, which the report does not contain.

Timeline

Summary

Who: Firmable, a sales intelligence company serving the United States, Canada and Asia-Pacific, surveyed 1,009 full-time professionals; the report is bylined to Chester Thompson, its vice president for the Americas. The findings bear on B2B sellers, marketers and any organization that relies on unsolicited outreach.

What: 52% of respondents delete cold messages without reading them and 30% skim then delete, while 1% almost always reply. More than half (54%) send cold outreach themselves; 61% distrust AI outreach they receive but 33% use AI to help write their own messages. Only 13% say human authorship matters most for a reply, against 58% who prioritize relevance or value. The report gives no fieldwork dates, countries or margin of error, and contains internal inconsistencies on sender shares and reply drivers.

When: Firmable published the report on August 19, 2026, and shared the findings with PPC Land on October 2, 2026.

Where: The survey's geography is not disclosed. Firmable sells in the United States, Canada and Asia-Pacific, and the outreach discussed spans email, LinkedIn and other channels without separation.

Why: Cold outreach remains a core B2B acquisition channel at a time of rising email volume, stricter mailbox rules, legal action over contact data in Europe and a rapid spread of AI drafting tools. The survey suggests recipients judge messages on relevance and the appearance of mass production rather than on who wrote them - a claim the stated-preference data cannot verify against actual reply behavior.