attraction.lab

Breadcrumbing in Dating: Where the 30% Figure Comes From, and What Was Actually Measured

It comes from 626 Spaniards surveyed in 2019, 86% of whom had never heard the word. In the same table, more said they had done it than had it done to them.

Research-backed writing on attraction, dating and relationships. Drafted by AI against a named bibliography, published only when every claim links to its source.
28 min read
Editorial illustration for: Breadcrumbing in Dating: Where the 30% Figure Comes From, and What Was Actually Measured

Before the article. Being kept on a string is not the same as being controlled, but the two sit close enough together that it is worth saying which is which before anything else. If what is happening also involves someone monitoring where you are, demanding access to your phone, isolating you from people, or alternating punishment with affection until you are managing your own behaviour to avoid the punishment, that is not ambiguity and nothing on this page addresses it. In the United States the National Domestic Violence Hotline is 1-800-799-7233, or text START to 88788, 24 hours a day. Find A Helpline, in the sources below, lists free and confidential services by country. A section near the top returns to this in more detail.

With that said. Almost every page about breadcrumbing opens with the same statistic: that over 30% of people have experienced it in the past year. The figure is real, it comes from a real study, and it is reported at roughly the right size — which makes this a different kind of page from the ones here that take apart a withdrawn number or a rule invented in a film.

What the figure has lost on its way around the internet is its context, and the context is more interesting than the number. It comes from 626 adults in Spain surveyed in 2019, of whom 86% had never heard the word before the survey explained it to them. In the same table, slightly more people said they had breadcrumbed someone than said it had happened to them. The single item that produced it has been reused, unchanged, by the only other large study in the field. And the only validated questionnaire for measuring the behaviour was built on 247 secondary-school pupils and, by its own authors, cannot measure being on the receiving end at all.

What follows is fourteen sections tracing the figure to its source, setting out the one finding that genuinely holds up — and it is not the one you would expect — and marking, at normal size, the large amount that nobody has measured. Our editorial policy explains why every empirical claim here carries a linked source.

#1

Where the 30% figure comes from

Search for breadcrumbing and a number arrives early. The Attachment Project's page on it opens by saying that over 30% of us have likely experienced breadcrumbing in the last year alone. PsychCentral tells you that research from 2021 indicates around 30% of dating adults have been breadcrumbed in the last 12 months. The figure turns up in roughly that form across the subject, and it is doing a particular job: it tells you the thing is common, so the thing is real, so read on.

It traces to one study. Raúl Navarro, Elisa Larrañaga, Santiago Yubero and Beatriz Víllora published it in Escritos de Psicología in 2021, and the sample is 626 adults in Spain aged 18 to 40. Recruitment was a survey link sent to doctoral students at one medium-sized university in central Spain, who were asked to forward it to family members and acquaintances. The exact number is 35.6% — 223 of 626 — who said that someone they considered a dating partner had breadcrumbed them at least once in the previous twelve months.

So the figure is not invented, and the pages quoting it are not misquoting it. That is worth saying plainly, because this site spends a lot of its time on numbers that were withdrawn by the person who produced them or traced back to a film script. This is not one of those. It is a real measurement from a real paper, reported at about the right size.

What happens to it is subtler. It travels without the four or five facts that tell you what it is a measurement of — including a second number from the very same table that points in an awkward direction, and the fact that the great majority of the people being counted had never heard the word until the survey explained it to them. This page is about those facts, and about what the rest of the small literature can and cannot support.

#2

Before the rest of this page: where this stops being a dating annoyance

This article is about what has been measured. There is a situation the same search sometimes comes from, and it needs answering first, because the answer does not depend on any of the statistics below.

Being kept on a string is not, by itself, abuse. It is inconsiderate, it is common, and most of the time it is someone managing their own uncertainty badly rather than running a strategy on you. But the same small literature that has studied breadcrumbing has also found it sitting alongside other things. In the one United Kingdom sample, exposure to breadcrumbing correlated with exposure to gaslighting at r = .30 — people reporting one tended to report the other. That does not mean a breadcrumber is an abuser. It does mean the categories blur in real life, and it is worth knowing which side of the line a particular situation is on.

The things that mark the line are not about message frequency. The US National Domestic Violence Hotline's warning-sign list is concrete: someone monitoring where you are and who you are with, demanding passwords or access to your phone, deciding how you spend money, pressuring you about sex, isolating you from the people in your life, or alternating between punishment and affection in a way that makes you manage your own behaviour to avoid the punishment. If a relationship that gets described as breadcrumbing also has those in it, the ambiguity is not the problem and reading about response rates will not help.

In the United States the National Domestic Violence Hotline is 1-800-799-7233, or text START to 88788, 24 hours a day. Find A Helpline, in the sources at the foot of this page, lists free and confidential services by country.

One more thing belongs here rather than lower down. Intermittent contact that you cannot predict, from someone you cannot leave, is the mechanism in the literature on trauma bonding — and that literature is about situations with a power imbalance and abuse in them, not about someone who is slow to reply. The two get conflated constantly in this genre. They are not the same, and the difference matters more than any figure below.

#3

The number in the same table that nobody quotes

The 2021 paper reports two prevalence figures for breadcrumbing, not one, and they sit in the same table.

35.6% (223 of 626) said they had been breadcrumbed at least once in the previous twelve months. 36.7% (230 of 626) said that they had breadcrumbed somebody else at least once in the previous twelve months.

Slightly more people reported doing it than reported having it done to them. The same inversion holds at the heavy end: 5.1% (32 people) said they had been breadcrumbed more than five times in the year, and 6.9% (43 people) said they had done it more than five times.

This is not a scandal and it does not mean anyone is lying. In a convenience sample there is no reason the two figures should match, and the pattern is familiar from the neighbouring literature — the same study found 19.3% had been ghosted and 23.2% had ghosted someone. People are not a closed system; the person who breadcrumbed you need not be in the survey.

What it does do is put a dent in the frame. Breadcrumbing is written about almost exclusively as something that is done to you, by a type of person, for a reason you are invited to speculate about. The genre's standard section is “why do they do it,” answered from the outside. But in the only dataset that asked both questions, the people who said they had done it were a slightly larger group than the people who said it had happened to them — and they were drawn from the same ordinary pool of adults, most of them with jobs and partners and university degrees.

That makes the honest version of this subject less like a profile of a predator and more like a description of a widely shared behaviour that most people on both sides find unsatisfying. The same thing turned out to be true of keeping someone in reserve, where the study that counted found the back-burner list to be close to universal rather than the mark of a particular character. It is a less satisfying story. It is the one the data supports.

#4

86% of the sample had never heard the word

Here is the detail that changes how to read every prevalence figure in this subject, and it is reported in the first paragraph of the 2021 results section.

Before being asked anything else, participants were asked whether they knew the terms. 539 of the 626 — 86.1% — said they were unfamiliar with “breadcrumbing.” For “ghosting,” 398 (63.6%) said they were unfamiliar with it. The study was run in Spain in 2019, in Spanish, and both words are English loans, which explains a good deal of that.

The researchers handled it in the obvious way: they supplied a definition. Participants read a paragraph explaining that breadcrumbers do not stop talking on WhatsApp, send random DMs or the occasional like so as not to discard the other person entirely, but the relationship does not progress; that it can happen after a break-up when the initiator does not want to let go; and that it is a way of keeping a date on hold. Then they were asked how often it had happened to them.

This is a reasonable method and the paper is open about it. But it means the 35.6% is not a count of people who recognised an experience they already had a name for. It is a count of people who, having been handed a fairly broad description of a fairly common way of behaving, said yes, that has happened. That is the same gap between a word and a measurement that runs through most of the vocabulary of modern dating. The abstract of the paper puts it in a single clause that no page quoting the figure carries over: half the participants were unfamiliar with the terms.

The breadth of the supplied definition matters too. As written, it covers the ex who still likes your photos, the person who texts every few weeks without suggesting anything, and the match who keeps the conversation alive for a month. Those are different situations with different stakes, and the survey collapses them into one yes. A prevalence estimate is only ever as specific as the question that produced it, and this question was deliberately wide — because it had to be, for a sample that mostly did not know the word.

#5

What the measure actually is, and how many studies share it

Prevalence figures in this area rest on a single survey item.

In the 2021 study, after the definition paragraph, participants answered one question — had a dating partner breadcrumbed them in the last year — on a five-point scale: never; not in the last year but before; once or twice; three to five times; more than five times. One question, one scale. The authors list this among their own limitations in the companion paper, noting that they measured both ghosting and breadcrumbing by asking only one question each, and did not ask how long it had been going on or what kind of relationship it was.

A single item is not automatically a bad measure. For something concrete and countable it is often the right tool. The difficulty is that breadcrumbing as defined here is neither concrete nor countable: the respondent has to decide what counts as a crumb, over what period, from whom, and whether the relationship “progressed.” The research on what message patterns actually predict ran into the same wall from the other side: the studies that found anything were the ones that stopped counting messages and measured what share of the relationship was happening on a screen.

The part that is easy to miss is how far that one item has travelled. The most recent study in the field — Rusi Jaspal and Barbara Lopes, published in Psychology International in 2025, with 544 adults recruited in the United Kingdom through Prolific — used the same single item developed by Navarro and colleagues, with the same definition paragraph translated across. Their methods section says so explicitly.

So the two largest quantitative samples on breadcrumbing, five years and two countries apart, are not independent measurements that happen to agree. They are the same instrument, written by one research group, applied twice. When a second study appears to corroborate a first, the useful question is whether it could have disagreed — and a shared single-item measure with a shared definition limits how much the second one was ever in a position to tell you. There is a separate, properly validated questionnaire in this field, and it has a problem of its own; that comes further down.

#6

The second estimate, and why it is not the same kind of number

The 2025 United Kingdom study gives a second prevalence figure, and it is much larger: of 544 adults, 272 (50%) reported ever having experienced breadcrumbing, and 324 (59.6%) reported ever having experienced ghosting.

Fifty per cent and thirty-five point six per cent look like a disagreement, or like evidence that the problem is growing. Neither reading survives a look at the two questions. The Spanish figure counts people who had been breadcrumbed in the last twelve months. The British figure counts people who had ever been breadcrumbed. Those are different denominators over different windows, and a lifetime figure will always be the larger one. The samples differ as well: the UK study recruited through Prolific, a paid participant platform, with a mean age of 30.7, 74.8% of participants in a relationship and 47.2% identifying as non-heterosexual — a very different composition from the Spanish snowball sample.

The same trap sits in the ghosting literature, and it is worth seeing once because it explains why these numbers range so widely. Published estimates for having been ghosted run from about 13% in a 2014 YouGov poll of 1,000 US adults up to 72% in a 2019 study of 333 US adults, as the Navarro group summarise them. That top figure is not evidence of an epidemic: the study in question recruited people who had recently experienced a break-up, which its own abstract states. Ask people who just broke up how a relationship ended and you will find a lot of ghosting. The number is about the sampling frame, not about the world.

There is no probability-sample estimate of breadcrumbing anywhere — no national survey, no census-weighted panel, nothing drawn to be representative of a country. Every figure in this subject comes from a convenience sample or a paid platform. That is normal for a young research area and it is not a scandal. It does mean that “30% of people” is a sentence the evidence cannot support, and “35.6% of 626 Spanish adults who answered a forwarded survey link in 2019” is a sentence it can.

#7

What the word meant before the research took it

The term reached the literature through a route worth tracing, because the definition the papers use is a quotation from an anonymous internet post.

The earliest substantial treatment is journalism: Jessica Bennett's column “The Agony of the Digital Tease,” published in the New York Times on 8 July 2016. Her breadcrumbers are not only romantic. One is a recruiter who sent “sup” every few days and never set up a meeting. One is an editor who repeatedly viewed a LinkedIn profile without messaging. One is a graduate student's brother, who “sends me random links on Reddit and then disappears.” The piece catalogues link-baiters, creepers, zombies and text-bubblers, and attributes the metaphor, hedged, to Urban Dictionary and to Hansel and Gretel — noting the irony that the breadcrumbs programmers named navigation trails after lead you home, and these lead you astray.

That 2016 usage is broad: a general name for digital teasing, in dating, work and families alike. The research that followed narrowed it to what one instrument calls affective-sexual relationships, which is a defensible scoping decision, but it means the word in the papers is not quite the word in circulation.

The definition the papers actually quote is narrower still, and its provenance is unusual. Navarro and colleagues define breadcrumbing in 2020 partly by quoting, verbatim and with a citation, the Urban Dictionary entry: “when the ‘crush’ has no intentions of taking things further, but they like the attention. So they flirt here or there, send DMs/texts just to keep the person interested, knowing damn well they're staying single.” That entry was posted by a user called Briness on 11 May 2018 and became Urban Dictionary's Word of the Day on 3 June 2021.

None of this makes the behaviour unreal. Pocketing came out of a tabloid column and cushioning has no research behind the word at all, and in both cases something real sits underneath. But it does set the expectation correctly: this is a folk term that research adopted, not a construct that research built and then named.

And on that scale, breadcrumbing is the lucky one. Europe PMC returns 19 records for it and one for micro-cheating, which arrived the same way — from a media interview at the end of 2017 — and which no study has since defined, measured or counted. The research that adopted this word is thin. For some of the others there was never any to adopt.

Further illustration for: Breadcrumbing in Dating: Where the 30% Figure Comes From, and What Was Actually Measured
#8

The finding that holds up, and the one that did not appear

The strongest thing in this literature is a 2020 paper by the same Spanish group, in the International Journal of Environmental Research and Public Health, using the same 626 participants. It asked whether being on the receiving end of these behaviours goes with anything measurable.

Participants completed three established scales: the Satisfaction With Life Scale, an eight-item version of the UCLA Loneliness Scale, and the helplessness subscale of a multidimensional fatalism scale. Internal consistency was good on all three (McDonald's omega .88, .91 and .87). Using a deliberately strict cut-off — three or more incidents in the year — the researchers sorted people into four groups: 30 ghosting victims (4.8%), 88 breadcrumbing victims (14.1%), 15 who had experienced both (2.4%), and 493 not involved (78.8%).

The result has two halves and the second is the interesting one.

Breadcrumbing showed up. In the group comparisons, the breadcrumbed and the combined group were less satisfied with life and reported more loneliness and more helplessness than the other two. In regressions adjusting for gender, age, sexual orientation and relationship status, breadcrumbing remained associated with all three (standardised betas of −.081 for life satisfaction, .143 for helplessness and .078 for loneliness, each at p < .01).

Ghosting did not. Not on any of the three. The people who had been ghosted three or more times in the year scored essentially the same as the people who had experienced neither. The authors call this surprising in their own discussion and spend several paragraphs on why it might be: ghosting ends a thing, so recovery can begin, whereas breadcrumbing keeps someone in a standby state the authors compare to ostracism.

That explanation is a hypothesis offered for future research, not a result — they say as much. But the null itself is a finding, it is the opposite of what the genre assumes, and it is the single most useful thing this literature has produced. The slow drip appears to be the part that correlates with feeling worse. The clean disappearance did not.

#9

The biggest number in that table is not breadcrumbing

Having given that result its due, it needs its size putting on it, because the regression table says something the summaries never carry.

The largest coefficients in all three models are not for breadcrumbing. They are for relationship status. Being single rather than partnered was associated with lower life satisfaction at a standardised beta of .340, with loneliness at −.335, and with helplessness at −.123. Breadcrumbing's betas, as above, were −.081, .078 and .143. Being single was associated with the outcomes roughly four times as strongly as being breadcrumbed, on the same scales in the same sample. The whole models explained 12.0%, 14.0% and 3.7% of the variance after adjustment, and relationship status is carrying most of that.

That matters because of a second line in the same paper: single participants reported significantly more breadcrumbing than partnered ones (χ² = 22.17, p < .001). So the predictor and the biggest confounder move together. People who are single are more exposed to breadcrumbing and also, on these measures, lonelier and less satisfied — and the design cannot separate which way any of it runs. Data collected at one moment cannot tell you whether being breadcrumbed makes people lonely, whether lonelier people are more exposed to it, or whether being single produces both.

Two smaller things, for anyone who opens the paper. The regression table is headed N = 643 where the sample is 626, and the loneliness coefficient for breadcrumbing is printed with a negative sign although the unstandardised coefficient beside it and the text both describe a positive association. The reported F statistics also do not correspond to the reported R² values: with R² = .129, six predictors and this sample size, the standard formula gives F ≈ 15.3, not the 150.2 printed. These look like typesetting slips rather than analytic problems, and the direction of the findings is stated consistently in the prose. They are mentioned because this page is asking you to trust a table, and you should know what is in it.

#10

Attachment, and the prediction that broke

The most interesting study in this literature is the one that connects breadcrumbing to attachment, and it is interesting because its authors were wrong in a specific, declared way.

Vivek Khattar, Nuria Huete and Raúl Navarro surveyed 682 adults in India and Spain, analysing 304 and 334 after attention checks, and published in BMC Psychology in 2023. Attachment was measured with the ECR-RD8, an eight-item version of the Experiences in Close Relationships scale giving an anxiety and an avoidance score. Breadcrumbing was measured by a validated questionnaire — of which more in the next section.

Their hypothesis was stated in advance and it was reasonable. Avoidant attachment should predict breadcrumbing, since keeping someone at a manageable distance is what avoidance does. Anxious attachment should predict it negatively, since anxiously attached people pursue closeness rather than withhold it.

The first half held. The second half inverted. Both dimensions were positively associated with breadcrumbing in both countries. In India, anxious attachment was the single strongest term in the model (β = .58, p < .001) with avoidance barely significant (β = .09, p = .027). In Spain, avoidance dominated (β = .46) and anxiety was smaller but clearly present (β = .17). The authors write that the result “was different than expected” and offer the push-and-pull reading: alternating contact and withdrawal is also what ambivalence looks like from outside.

This is a genuinely useful corrective to the standard framing, which treats inconsistent contact as diagnostic of avoidance. It is not. Both insecure patterns can produce it, and from the receiving end they are indistinguishable — which is exactly why supporting an avoidant partner and supporting an anxious one call for different things and the behaviour alone will not tell you which you are looking at.

Three caveats, all from the paper. It measured breadcrumbing engagement — doing it — not being on the receiving end, because no validated victimisation measure exists. The cross-country contrast is confounded: the Indian sample was 81% women with a mean age of 22, the Spanish sample half men with a mean age of 28, and the authors say this probably explains the demographic differences. And the avoidance subscale was the least reliable instrument in the study, at α = .604 in Spain — the subscale carrying the headline Spanish finding.

#11

The one validated instrument was built on schoolchildren

The questionnaire used in that attachment study is the BREAD-ASR, and it is the only validated instrument for breadcrumbing in existence. It is worth knowing what it was built on, because nothing in the subject mentions it.

It was developed by María del Carmen Rodríguez-García and colleagues in the nursing faculty at the University of Almería, and published in the same journal in December 2020. The validation sample was 247 high-school students, recruited by convenience from a single public secondary school in south-eastern Spain between March and June 2019, who completed a paper-and-pencil survey. Inclusion criteria were being 12 to 18 years old, owning a mobile phone, having had affective-sexual relationships, and having a parent or guardian's consent. The pilot that checked the wording was 30 volunteers aged 12 to 18.

The psychometrics are reported honestly. Exploratory factor analysis produced four factors — communication, avoidance, commitment and dependence — explaining 40% of the variance, with an ordinal alpha of .83 after one of the seventeen items was dropped. There was no test-retest, so temporal stability is unknown.

The authors' own limitations section is unusually direct, and two of its statements are the ones to keep. First: “The instrument developed only perpetration measures; therefore, it would not be appropriate to use it to explore breadcrumbing victimization.” Second: “This study was conducted in the Spanish language; therefore, its current application is limited to the Spanish-speaking population.”

Three years later the scale was applied to adults aged 18 to 40 in India. The 2023 team did the responsible thing and ran confirmatory factor analyses in both samples, and reported the output rather than burying it — but the output is mixed. The Indian fit was reasonable. The Spanish fit was not: a Kaiser–Meyer–Olkin value of 0.58, below the 0.60 usually treated as the floor, with NFI .77 and CFI .819, both under the conventional .90 threshold.

None of this makes the attachment finding worthless. Scales get extended beyond their validation samples all the time and the field could not move otherwise. It does mean that the measurement foundation of the quantitative work here is one perpetration scale built on secondary-school pupils, which the field is using on adults in two languages because there is nothing else.

#12

What the recipients' side of this rests on

If the quantitative work measures the people who breadcrumb, something has to speak for the people on the other end. One study does, and it is small enough that the number should travel with it.

Vivek Khattar, Shivangi Upadhyay and Raúl Navarro published a qualitative study in Societies in 2023 using focus groups with young adults who had been breadcrumbed. Five themes emerged from the thematic analysis — charm, leading on, incongruence, avoiding emotional investment, and commitment uncertainty — and the conversations also touched on effects on later relationships, emotional disturbance, self-concept and signs of depression.

Those five themes are a good description. They are also the entire recipient-side evidence base, and they come from two focus groups with nine participants in total: one man and eight women.

Nine people is a legitimate qualitative sample. Focus groups are for generating categories, not estimating frequencies, and on that basis the study did its job — it produced vocabulary for an experience that had none. The problem is what happens downstream, where those themes get restated as what breadcrumbing does to people, in the confident register of a finding. Pages citing this work describe focus-group interviews identifying five recurring themes without ever reaching the figure nine, which is the one number a reader needs in order to weigh it.

It is also worth noticing what the sex ratio does. Eight of nine participants were women, and the genre's assumption that breadcrumbing is something men do to women has a tendency to come out of studies like this looking confirmed. The quantitative data does not support it. In the 2021 Spanish sample there was no significant gender difference in breadcrumbing, in either direction, for receiving it or for doing it. In the 2025 UK sample there was no significant difference by sex either. The one place gender did appear was the Spanish half of the 2023 study, where men reported more engagement — and the authors decline to draw a conclusion from it, because it did not replicate in their Indian sample.

Searched on 31 October 2026, PubMed returns three records for breadcrumbing. All three appear on this page.

#13

What the evidence will support you concluding

Pulled together, the defensible list is short, and each item is tied to what produced it.

The behaviour is real and the hurt is plausible. The association between being breadcrumbed and lower life satisfaction, more loneliness and more helplessness is the most replicated-looking thing here, and it was found with established scales. It is correlational and the effect sizes are small, but it is not nothing, and nobody needs a study to be told that being strung along feels bad.

The sharpest finding is the comparison, not the correlation. Ghosting, in the same sample, was associated with none of the three outcomes. Whatever is happening, it appears to be about the open loop rather than the rejection. If you are deciding which situation to get out of, the evidence points at the ambiguous one — which is also what the research on situationships and on staying in a relationship that is not working keeps finding.

It travels with other things. In the UK sample, breadcrumbing exposure correlated with ghosting exposure (r = .30) and gaslighting exposure (r = .30), and with lower social support (r = −.18) and younger age (r = −.17). When ghosting and gaslighting entered the model, social support stopped being a significant predictor. The authors read this as cumulative victimisation eroding the protection that support normally offers. A flatter reading is available and should be said: all four were self-reported by the same person at the same sitting, and people who report one bad thing tend to report others.

Being single is the strongest demographic signal, in every sample. Spain 2021, Spain and India 2023, and the UK 2025 all found it, with the largest demographic effect sizes in each. The plain explanation — single people are doing more dating, so they meet more of it — is both the most likely and entirely untested.

Nothing here supports a profile. No study has found a kind of person who breadcrumbs. What has been found is that both insecure attachment dimensions go with doing it, that roughly as many people report doing it as having it done to them, and that the genre's demographic assumptions about gender do not appear in the data at all.

#14

What nobody has measured

The limits belong in the same type size as the findings, and on this subject they take up more room than usual.

Nobody has followed anyone over time. Every quantitative study here is cross-sectional — one questionnaire, one sitting. There is no study in which people were measured before and after being breadcrumbed, which means the central causal claim of the entire genre, that this does something to you, has never been tested. The 2023 authors say it plainly: no study so far has tracked people over time to establish which comes first.

There is no instrument for the receiving end. The only validated scale measures perpetration and its authors say it should not be used for victimisation. Everything quantitative about being breadcrumbed therefore rests on one survey item, shared across the two largest studies.

No sample has been representative of anywhere. Snowball sampling in Spain, a paid platform in the UK, social media and WhatsApp in India. No probability sample, no national estimate, and no United States data at all — which is where most of the pages quoting the 30% figure are written and read.

The mechanism is a speculation, and the authors label it as one. The intermittent-reinforcement account — unpredictable reward keeps you hooked — is offered in the 2020 discussion as a possible explanation that future research should test, not as a result. The experimental work behind it is about resistance to extinction in animals, and the place it has been studied in human relationships is trauma bonding, where the conditions are abuse and a power imbalance. Nobody has shown that it is what makes a quiet phone compelling. The nearest thing to a measured version of an open loop kept open on purpose sits in the research on post-breakup friendship, where staying in touch out of unresolved romantic desire was the one motive associated with worse outcomes: what staying friends with an ex has actually been measured to do.

And there is no evidence about what to do. Every page on this subject, including the ones citing the studies properly, ends with advice: set boundaries, call it out, do not take it personally, walk away. None of that has been tested on this behaviour. It may well be good advice. It is not a finding, and anything on this page that sounds like guidance is an inference, offered as one. The one thing with support behind it is the same thing the ambiguity research keeps arriving at: asking directly resolves it faster than watching does, and it is rejected less often than people expect.

Pulling it together

The honest summary of this subject is that the behaviour is ordinary, the experience is unpleasant, and the evidence is thin in a way that the confident tone of most writing about it conceals.

Three things are worth carrying away. The first is that the comparison beats the correlation: in the one study that measured both, being breadcrumbed went with lower life satisfaction, more loneliness and more helplessness, and being ghosted went with none of them. The open loop looks like the part that costs something, which is an argument for resolving ambiguity rather than enduring it.

The second is that there is no profile here. Both anxious and avoidant attachment go with doing it, roughly as many people report doing it as having it done to them, and the genre's assumptions about who does this to whom do not appear in the data. The person leaving you on read is considerably more likely to be someone handling their own uncertainty badly than someone running a technique.

The third is the limit. Nobody has followed anyone over time, no sample has been representative of any country, there is no instrument for the receiving end, and the intermittent-reinforcement explanation everybody reaches for is a hypothesis its own authors flagged for future testing.

Which leaves the only move with real support behind it, and it is the same one that ends most pages on this site: ask. It resolves the ambiguity faster than watching does, and people say no less often, and more gently, than anyone expects them to.

Frequently asked questions

Is it true that 30% of people have been breadcrumbed?

The figure is real but it is narrower than it sounds. It comes from Navarro, Larrañaga, Yubero and Víllora's 2021 study of 626 adults aged 18 to 40 in Spain, recruited by forwarding a survey link from doctoral students at one university to their families and acquaintances. In that sample, 35.6% — 223 people — said a dating partner had breadcrumbed them at least once in the previous twelve months. Two things do not travel with the number. 86.1% of the sample said they had never heard the word before the survey supplied a definition, so this is not a count of people recognising something they already had a name for. And in the same table, 36.7% said they had breadcrumbed somebody else. There is no nationally representative estimate of breadcrumbing anywhere, in any country, so “30% of people” is a claim the evidence cannot support.

Does breadcrumbing hurt more than being ghosted?

On the three things one study measured, the people who had been breadcrumbed scored worse and the people who had been ghosted did not score worse at all. In the 2020 paper using the same 626 Spaniards, those breadcrumbed three or more times in a year reported lower life satisfaction, more loneliness and more helplessness than the uninvolved, while those ghosted three or more times were statistically indistinguishable from people who had experienced neither. The authors call the ghosting null surprising and suggest that a clean ending allows recovery to begin while breadcrumbing keeps someone in a standby state — but they present that as a hypothesis for future research, not a result. The effects are small, the design is cross-sectional, and the strongest predictor in the same table was being single rather than being breadcrumbed.

What kind of person breadcrumbs people?

No study has found a type. The closest thing is Khattar, Huete and Navarro's 2023 survey of 638 analysed adults in India and Spain, which found that both anxious and avoidant attachment were positively associated with reporting breadcrumbing behaviour. The authors had predicted that anxious attachment would be associated negatively, and say in the paper that the result was different from what they expected: alternating contact and withdrawal is what ambivalence looks like from the outside as well as what distance-keeping looks like. On demographics, there was no significant gender difference in the Spanish 2021 sample or the UK 2025 sample. The study measured people's own reported behaviour, not what was done to them, because no validated measure of being breadcrumbed exists.

Is breadcrumbing a form of emotional abuse?

By itself, usually not, and the distinction is worth keeping sharp. Inconsistent contact from someone who has not decided what they want is common, unpleasant and not abuse. What makes something abusive is a pattern of control: monitoring, restricting access to money or people, pressure about sex, demands for passwords, or alternating punishment and affection in a way that makes you manage your own behaviour. The one caution the data supports is that these things co-occur — in the 2025 UK study of 544 adults, breadcrumbing exposure correlated with gaslighting exposure at r = .30. That is an association between self-reports collected at one sitting, not evidence that one causes the other. If the warning signs above are present, the National Domestic Violence Hotline is 1-800-799-7233, or text START to 88788.

Why do people breadcrumb instead of just saying they are not interested?

The honest answer is that this has not been measured for breadcrumbing specifically, and anyone giving you a confident motive is inferring it, usually from what recipients believe. The adjacent research is more useful. People are consistently worse at delivering a refusal than they predict: Joel, Teper and MacDonald found that participants rejected others far less often when the situation was real than when it was hypothetical, out of consideration for the other person's feelings, and the friend zone article sets that study out in full. That produces exactly this behaviour — warm, vague, non-committal contact that is easier to send than a clear no. The 2023 attachment data is consistent with it being about the sender's discomfort rather than a strategy aimed at you, since both insecure attachment patterns predicted it.

Does ignoring a breadcrumber make them come back?

Nobody has tested it, and that is the whole answer for breadcrumbing specifically. What exists nearby is not encouraging for the tactic. The one experiment on waiting before replying after a date — Teichmann and colleagues, preregistered, 543 participants, set out in full in our article on the three-day rule — found that replying the next morning produced slightly higher relationship intentions than replying immediately, and that waiting two days was worse than both on every measure. Deliberately engineered unavailability has not been shown to increase anyone's interest. What the evidence does support is that ambiguity itself is costly to the person sitting in it, and that asking directly resolves it faster than waiting does.

How is breadcrumbing different from cushioning or pocketing?

They describe different arrangements and they have different evidence behind them. Breadcrumbing is sporadic, non-committal contact that keeps a connection alive without letting it progress. Cushioning is keeping other people in reserve while in a relationship, and the behaviour underneath it has been studied since 2014 under the name back burners, where the striking finding is how ordinary it turned out to be. Pocketing is being kept out of someone's visible life, and its research sits in the literature on relationship concealment. Of the three, breadcrumbing has the most direct research under its own name — three records in PubMed as of 31 October 2026 — and all of it is cross-sectional.

Sources

What this article draws on. For the broader bibliography see our sources page.

Keep reading