Noise：为什么受众在指标察觉之前就已无视信息

Noise is anything that competes with, dilutes, or distracts from the message a marketer wants a person to receive. In practice it comes from two places: too much volume, and messages that do not fit the person on the other end.

Volume and fit are different problems, and fixing one does not fix the other. The receiver, not the sender, decides what counts as noise. And when people decide a channel has become noisy, the evidence below suggests they rarely complain. They mute, block, scroll past, or show up less often, and those exits are easy to miss in routine reporting.

Noise in Marketing Communication

Noise is a practitioner term, not a standardized metric. The useful question is which kind an audience is reacting to. The term borrows the everyday idea of interference between a sender and a receiver. For marketers, that interference includes competing ads, notifications, crowded feeds, full inboxes, and the brand's own messages when there are simply too many of them.

Two Sources of Noise: Volume and Fit

Volume noise is about quantity. The person receives more messages than they can or want to process, so even reasonable messages start to feel like too much. Fit noise is about mismatch. The message reaches someone for whom it is not useful or relevant, so it reads as interference no matter how few arrive.

The clearest public data separating the two comes from news, so it serves as context here, not as a measure of marketing. In a Pew Research Center survey of 3,560 U.S. adults conducted December 8 to 14, 2025, 52% said they were worn out by the amount of news (news, not advertising), 34% said they were not, and 14% were not sure. In the same survey, 48% said most of the news they come across is not relevant to their lives, while 35% said it mostly is.

Analysis: those are two different complaints, one about how much arrives and one about whether it applies at all. Reuters Institute data on news alerts points the same way: people who turned alerts off cited both getting too many and not finding them useful (news, not advertising). A quieter schedule will not rescue an irrelevant message. A sharper message will not rescue a schedule that floods the audience.

What Noise Is Not

Analysis: the word also appears in two other marketing conversations, and mixing them up leads to the wrong fix. Statistical noise in analytics and A/B testing is random variation in data that can make a difference between two versions look real when it is not, or hide one that is. The fix is better test design and enough data, not fewer messages.

Judgment noise is unwanted variation when different people, or the same person on different days, make different calls on the same case, such as two reviewers scoring an identical lead differently. The fix is clearer criteria. This entry covers communication noise only.

How Audiences Answer Noise

People tend to answer noise in three ways. They switch a channel off, they look past whatever reads as an ad, or they use the channel less. Advertising data leads, with news data as supporting context.

They Switch the Channel Off

In a UK government-commissioned survey, 36% of respondents said they use pop-up or ad-blocking software. Ipsos ran it for the Department for Culture, Media and Sport (DCMS) among 8,511 people aged 16 and over between April 4 and 10, 2024, using a random probability panel. Ad blocking is the most literal silent exit: nobody tells the advertiser.

News data shows the same behavior with notifications (news, not advertising). In the Reuters Institute's 2025 analysis of eight markets, including the US, UK, India, and Brazil, 79% of respondents received no news alerts in an average week. Among those not getting alerts, 43% said they had actively turned them off, citing reasons such as too many alerts or not finding them useful. That 43% applies only to non-recipients, not to everyone surveyed.

Pew's survey shows a broader version (news, not advertising): 60% of U.S. adults had at some point reduced the amount of news they get, and 67% had stopped getting news from a specific source. The survey did not establish why, so overload should not be treated as the only cause.

They Look Past What Reads as an Ad

A 2025 eye-tracking study in Frontiers in Psychology tested 152 participants in a university lab in Germany, using mock Instagram-style feeds that mixed sponsored and organic posts. Sponsored posts received fewer fixations on average than organic posts, 5.00 compared with 6.52, and significantly lower dwell time. Recruitment used convenience and snowball sampling, so this is an experimental observation, not a benchmark.

Analysis: this is the quietest exit. The impression may still be counted, yet attention moves on once a post reads as advertising. A delivered impression and a noticed message are not the same event.

They Use the Channel Less

Ofcom research found that 45% of UK podcast listeners say there are too many adverts on the podcasts they listen to, a share Ofcom says has grown over the past five years. That measures perception, not behavior.

The behavioral side comes from a long-run field experiment on Pandora Internet Radio by Ali Goli, Jason Huang, David Reiley, and Nickolai Riabov, published in 2025 using listener data from June 2014 to April 2016 across around 35 million Pandora listeners randomly assigned to different ad loads. Each additional audio ad per hour reduced total hours listened by about 2.08% and active listening days by about 1.91%. The long-run effect was roughly three times the short-run effect.

That finding is the basis for this entry's title, and it comes from one study in one channel. It suggests a short test can underestimate what a heavier ad load costs, because listeners drift away gradually. It does not show whether social feeds, email, or display ads behave the same way.

Why Relevance Alone Doesn't Cancel Noise

The standard advice is to be more relevant. Relevance helps with fit, but the evidence suggests it does not cancel noise on its own.

In the same DCMS and Ipsos survey, 72% agreed that online advertising often frustrates them, and 34% strongly agreed. Asked about ads based on their search history or other online activity, 69% said those ads frustrated them, with 36% strongly agreeing. Ads built from a person's own behavior are an attempt at relevance by design. Most respondents still found them frustrating.

An online experiment by Sieun Ha and Matthew S. Eastin, published in 2025 in the Journal of Digital & Social Media Marketing, studied advertising clutter on social media with 170 participants. The authors concluded that high advertisement relevance might not be sufficient to offset the negative effects of both physical and expected advertising clutter.

Analysis: relevance works on fit, while clutter is a volume problem. A well-targeted ad dropped into a crowded feed still arrives inside the noise, so a gain on one dial should not be read as progress on the other.

Measuring Noise Without a Standard Metric

No standardized noise rate appears in the government, academic, or industry research reviewed for this entry, and none of it established a universal count of ads a person sees per day. Noise has to be read through proxies, each showing part of the picture.

Exit signals show volume and fit problems after the fact: notification opt-outs, unsubscribes, hidden or reported ads, and ad-blocker use where visible. Attention signals, such as the fixation and dwell measures in the Frontiers study, show whether a delivered message was actually looked at.

Usage over time is the signal most likely to be missed. The Pandora experiment tracked listening over many months, and the long-run effect turned out larger than the short-run one. Analysis: a team that judges ad load or send frequency from a two-week test may not see a gradual loss of audience until well after the decision. Pairing any volume increase with a holdout group or a longer observation window, and tracking exits next to clicks and conversions, helps close that gap.

What the Evidence Covers in Indonesia

Direct measurement of advertising noise in Indonesia did not appear in the sources reviewed, so the evidence is indirect. The Reuters Institute reported that CNN Indonesia typically sends up to 50 news alerts a day, based on the Project Push tool, which tracked alerts from more than 100 news providers; this is a publisher practice reported by the Reuters Institute, news not advertising, and not the number of alerts any one person receives. In the Reuters Institute's 2026 online survey, 37% of Indonesian respondents said they sometimes or often avoid the news, down 2 points from the year before (news, not advertising; an online sample, not a census). A 2026 Societies study by Sabrina Oktaria Sihombing and Juliana Juliana (Universitas Pelita Harapan) found that perceived information overload was positively associated with intention to do a digital detox (path coefficient 0.336, p < .001) among 198 university students in Greater Jakarta, chosen through purposive sampling and each holding at least three active social media accounts. That is an association, not proof that overload causes disengagement, and it does not measure advertising.

Reducing Noise: Volume Levers and Fit Levers

Analysis: because noise has two sources, it helps to sort the fixes into two groups and check which one a problem calls for.

Volume levers control how much reaches each person. The most direct is frequency capping, which limits how many times the same person sees a message within a set period. Closely related is watching average frequency, the number of exposures per person, instead of total impressions alone. Volume levers often trade against reach: cutting repeat exposures can free budget to reach more people once, a different goal from pressing the same people harder.

Fit levers control whether what arrives is worth receiving. They include tighter audience definitions, better timing, letting people choose what they receive, and dropping messages that serve the brand's calendar more than the reader's need. Fit levers do not replace volume levers, as the 69% frustration with search-history-based ads shows.

A simple check before adding another message: if the audience received nothing from this campaign this week, would anyone notice? If not, the message is probably adding to the noise.

Frequently Asked Questions

What causes noise in marketing?

Usually two things at once: volume, meaning more messages than a person can or wants to process, and poor fit, meaning messages that are not useful or relevant to the person receiving them. Pew's December 2025 survey of U.S. adults shows these as separate complaints about news, not advertising: 52% felt worn out by the amount, and 48% said most of it was not relevant.

Is noise the same as ad clutter?

Ad clutter is one form of volume noise, where many ads compete in the same space, such as a feed or a page. Noise is broader. It also covers notifications and poorly matched messages, even when few arrive.

Does personalization or better targeting remove noise?

Not on its own. In a 2024 UK survey for DCMS, 69% of people aged 16 and over said they were frustrated by online ads based on their search history or other online activity. A 2025 experiment on social media found that high ad relevance might not be sufficient to offset the negative effects of advertising clutter. Relevance improves fit. It does not solve volume.

How can you tell if an audience is experiencing noise?

There is no standard noise metric, so the answer comes from proxies: opt-outs from notifications or email, hidden or reported ads, ad-blocker use where visible, attention measures such as dwell time, and changes in usage over time.

Can noise hurt results even when short-term numbers look stable?

One long-run experiment suggests it can. On Pandora, with listener data from 2014 to 2016, the long-run effect of extra audio ads on listening was roughly three times the short-run effect. That is a single study in one channel, so it supports longer test windows rather than proving a general rule.

Related Terms

• Frequency
• Frequency Capping
• Reach

Sources

Pew Research Center, Americans' Complicated Relationship With News, February 11, 2026. U.S. adults, n=3,560, survey December 8 to 14, 2025. Worn out by news, relevance, reduced intake, stopped a source. News, not advertising. VERIFIED. https://www.pewresearch.org/journalism/2026/02/11/americans-complicated-relationship-with-news/.

Reuters Institute for the Study of Journalism, University of Oxford. Nic Newman, Walking the notification tightrope, Digital News Report 2025, June 17, 2025. Eight markets. Alerts received and turned off; CNN Indonesia alert volume via Project Push. News, not advertising. VERIFIED. https://reutersinstitute.politics.ox.ac.uk/digital-news-report/2025/walking-notification-tightrope-how-engage-audiences-while-avoiding.

Reuters Institute for the Study of Journalism. Digital News Report 2026, Indonesia page, June 16, 2026. Indonesia, online sample. News avoidance. News, not advertising. VERIFIED. https://reutersinstitute.politics.ox.ac.uk/digital-news-report/2026/indonesia.

Ipsos for the UK Department for Culture, Media and Sport, Online Advertising Experiences Survey. UK, aged 16 and over, n=8,511, April 4 to 10, 2024, probability panel. Frustration with online ads and behavior-based ads; ad-blocker use. VERIFIED. https://www.gov.uk/government/publications/online-advertising-experiences-survey-research-report.

Ofcom, International Podcast Day: taking a look at our listening habits. UK podcast listeners. Share saying podcasts carry too many adverts. VERIFIED. https://www.ofcom.org.uk/media-use-and-attitudes/media-habits-adults/international-podcast-day-taking-a-look-at-our-listening-habits.

Hübner, Thalmann, and Henseler, Blending in or standing out? The disclosure dilemma of ad cues of social media native advertising, Frontiers in Psychology 16, August 13, 2025. Lab eye-tracking experiment, Germany, n=152. Fixations and dwell time. VERIFIED. https://doi.org/10.3389/fpsyg.2025.1636910.

Goli, Huang, Reiley, and Riabov, Measuring Consumer Sensitivity to Audio Advertising: A Long-Run Field Experiment on Pandora Internet Radio, arXiv 2412.05516, published 2025 in Quantitative Marketing and Economics. Around 35 million listeners, data June 2014 to April 2016. Hours and days listened per extra ad. VERIFIED (results); REPORTED (journal citation). https://arxiv.org/abs/2412.05516v1 and https://ideas.repec.org/a/kap/qmktec/v23y2025i3d10.1007_s11129-025-09294-7.html.

Ha and Eastin, Not another ad! Expectancy violation effects of advertising clutter on social media, Journal of Digital & Social Media Marketing 13(1), 2025. Professional journal, Henry Stewart Publications. Online experiment, n=170. Clutter and ad relevance. VERIFIED (abstract). https://econpapers.repec.org/RePEc:aza:jdsmm0:y:2025:v:13:i:1:p:78-94.

Sihombing and Juliana, Digital Detox Intention Among Indonesian Generation Z, Societies 16(2), 54, February 10, 2026. Greater Jakarta students, n=198, purposive. Information overload and digital detox intention (association). VERIFIED. https://www.mdpi.com/2075-4698/16/2/54.