You are eleven weeks in. You have sent more applications than you want to count, you have had three conversations that went nowhere, and the question underneath every evening is whether this is normal or whether something is wrong with you.
There is a number that appears to answer it. A job search takes about six months, people say, so relax, you are early. It does come from real data. What that data actually records is something much narrower, and seeing what it is changes what you should be planning for.
The number everyone quotes is not the number you want
The US Bureau of Labor Statistics publishes the length of time unemployed people have been looking for work, every month, in Table A-12 of its Employment Situation release. In the figures for July 2026, published on 7 August, the average duration was 24.9 weeks and the median was 10.5.1 Those are July's numbers, and that table always carries the newest month — so if you are reading this later it will not say 24.9 and 10.5. The July release is archived under its own date.8
Those are two averages of the same thing, and one is nearly two and a half times the other. That gap is the whole story. When a mean sits that far above a median, the distribution has a long tail: a minority of exceptionally long spells pulling the average away from the typical case. The mean works out at just under six months, which is presumably where the familiar figure comes from. What the gap shows is that the mean is being pulled by the long spells, so it describes the tail's influence rather than the typical case — which is a different thing from the number people reach for when they say a search takes about six months.
Then there is the caveat that should be attached to the number every time it is quoted, and never is. From the BLS's own definitions page:2
"These measures should not be interpreted as the length of time it takes someone to find a job, or how long they look for work before giving up their job search. The Current Population Survey does not ask how long it took someone to find a job, and the duration measures do not provide that information."
The same page explains why: the figures "reflect the still-in-progress spells of unemployment, not completed spells."2 The survey stops people in the middle of searching and asks how long they have been at it so far. Anyone who found work in June is not in the July number.
What the distribution does say
The same table breaks the unemployed into bands, which is more useful than either average. Seasonally adjusted, in July 2026:18
- 28.2% had been looking for less than five weeks
- 29.5% for five to fourteen weeks
- 16.7% for fifteen to twenty-six weeks
- 25.5% for twenty-seven weeks or more
So among everyone still searching in a given week, most are early, and a quarter are past six months. With the caveat above in mind, that is not a prediction of how your search will go. It is a description of who is out there looking alongside you, and the shape of it is the part to hold on to: this is not a bell curve centred on six months. It is heavily front-weighted with a substantial tail, and planning against "the average" means planning against a number that is reporting the tail's pull rather than the middle of the distribution.
How long the tail runs depends enormously on where you are
Europe measures a different thing, which is helpful, because it puts a number on the tail directly. Eurostat counts long-term unemployment: people aged 15 to 74 who are out of work and have been actively seeking employment for at least a year, following the International Labour Organization definition.4 Expressed as a share of everyone unemployed, in 2025 that was 31.5% across the EU27, down slightly from 32.4% the year before.3
The average conceals the interesting part. In the same year that share was 13.7% in the Netherlands and 14.1% in Denmark, 27.9% in Germany, 50.3% in Italy and 63.4% in Slovakia.3
These are shares of the people currently unemployed, not the odds that any given search runs a year, and the same length bias applies: someone stuck for two years is in far more of these annual counts than someone who found work in March. Read as composition rather than as risk, though, they are stark. Among Danes who are out of work, roughly one in seven has been looking a year or more. Among Slovaks, closer to two in three.
The table shows the gap; it does not explain it, and it holds nothing else constant — the unemployed populations differ in age, education and occupation too. But nothing about a CV changes as it crosses those borders, and the candidates for a gap that size are things like vacancy rates, sector mix, and benefit and hiring rules. None of them is a fact about the person searching. If you have been at this for eight months, the state of the market you are standing in is a more informative thing to know than any amount of advice about your CV format.
Two limits on those figures. They are annual data for 2025, last updated in June 2026, so they describe last year rather than this month. And they count people who are out of work, which leaves out everyone searching from inside a job — a group no duration statistic on this page covers at all.
What the length of a search does to how you are read
There is a fear underneath the planning question: that the longer this goes on, the worse it gets, and that the gap itself is now the problem. The best evidence on that comes from audit studies, where researchers send fictitious but realistic applications to real vacancies and vary one detail at random. That design isolates the effect of the detail, because nothing else about the candidate differs.
Kroft, Lange and Notowidigdo sent roughly 12,000 such résumés to about 3,000 real postings across the 100 largest US cities, in sales, customer service, administrative support and clerical roles. The average callback rate was 4.7%. Callbacks fell as the stated length of unemployment rose, "with the majority of this decline occurring during the first eight months".5 The working paper appeared in September 2012 and the article in the Quarterly Journal of Economics the following year, so the fieldwork necessarily predates that.
Eriksson and Rooth ran a comparable experiment in Sweden, sending 8,466 applications from fictitious young workers to 3,786 employers across thirteen occupations between March and November 2007.7 Their findings are more encouraging than the American ones, and more specific. Employers did not treat short current spells differently, which the authors read as employers understanding "that worker/firm matching takes time". They did attach a negative value to a current spell lasting at least nine months. And past long-term unemployment did not affect hiring decisions at all, which they attribute to subsequent work experience erasing the signal.6
The two studies disagree about when the penalty starts, and both are old. Treat them as evidence about how employers read a gap rather than as a timetable. What they agree on is more useful than where they differ: it is the current spell that is being read, it is read as a signal rather than a disqualification, and the reading is not linear in time.
If the Swedish result holds, the most reliable way to remove the signal is to be working again in some form, because the penalty attaches to the spell rather than to your history. That is not free advice. Contract, part-time or freelance work takes the hours a search needs, and it can lower the number you next anchor to. It is also a completely different calculation at month three than at month ten, which is rather the point.
Planning against a distribution instead of an average
A plan cannot remove the uncertainty. It can be built to survive it, and the shape of the data says what it has to survive.
Set the runway to the tail, not the middle. Note what the snapshot cannot do here: it gives no odds that your search runs long, and reading it that way is the exact mistake this article started by pulling apart. What it does establish is that long searches are ordinary. At any moment a quarter of the people still looking have been at it more than six months, and length bias means those months are, if anything, undercounted for each of them. A plan that breaks at month four is betting against a well-populated tail. That is mostly a question about money, and it is far better answered in week one than in week sixteen.
Choose a review date rather than a volume target. The reflex at month three is to apply to more things, which treats a strategy problem as an effort problem. Decide now that on a specific date you will change something structural — the roles, the level, the geography, the story you are telling — and then hold the date instead of renegotiating it every Sunday.
Set a floor you can keep, not a peak you cannot. Front-loading has real value; the evidence above suggests early weeks are the cheap ones. But a fortnight of twelve-hour days followed by a month of avoidance is worse than a smaller, steady week, and the tail is where long searches are actually decided.
Count your own funnel. Since no published statistic can tell you how long yours will take, the only diagnostic that means anything is your own: applications sent, replies, first conversations, final stages. Fifty applications and no replies is a different problem from ten first interviews and no offer, and only one of them is solved by applying to more. A tracker is easy to start and easy to let slide, and it earns its keep later than it feels like it should. If you would rather not maintain one, JobCraftly can keep the record as you go.
What the numbers cannot tell you
There is no forecast in these figures. The people who publish them say so outright. That is a statement about this data rather than a claim that nothing better exists anywhere — longitudinal and administrative datasets can follow spells to their end, and none of them is the number people reach for when they tell you a search takes about six months.
What the figures do support is narrower and still worth carrying: long searches make up a large share of the people looking at any moment, and that share differs enormously between countries. None of it is evidence about you in particular, which is worth holding onto on the evenings when it feels like the only thing it could be.
