Large wage surveys describe a reference period well before their publication date, so a figure released this spring typically describes pay from about a year earlier — before anybody copies it onto a second site with a fresh timestamp. That lag is fine for understanding the shape of a market and actively harmful in a live negotiation, where a year of wage growth sits between the number you are quoting and the number the employer is working from.
The date that matters is not the one displayed
Every wage figure describes a period, and that period is almost never the date printed at the top of the page. A major occupational wage survey collects data over a defined window, processes it for months, and publishes well afterwards, so the number you are reading describes what people were paid roughly a year before you read it. Nothing on the page you are reading says so anywhere.
Then the copying starts, and the drift gets worse. A figure gets quoted in an article, which is dated the day it was written. That article gets summarized somewhere else with a fresh timestamp. Within two steps a number describing last year’s pay is sitting on a page dated this month, with nothing on the page that is untrue and nothing that tells you what you are actually looking at.
None of this is deception on anybody’s part. It is what happens when a figure with a reference period travels through a medium that only records publication dates. The reader’s job is to find the reference period, and it is usually findable in one click.
How much drift to expect
A year of wage growth is the thing sitting between the figure and today, and how much that matters depends entirely on the period. In a flat market it is a rounding error you can ignore. In a period of rapid wage growth it can be several percent, which is more than most negotiations move.
It also varies by occupation rather than applying evenly. Fields with tight labor markets move faster than the average, and fields where pay is set by a schedule or a bargaining agreement barely move between revisions. So the correct adjustment is not a single national number applied to everything.
The practical approach is to treat the published figure as a floor rather than a current market rate, and to check a wage growth series for the period between the reference date and today if the number is going to decide anything. That is a two-minute check and it is the difference between quoting a figure and understanding one. It is also the difference between being checkable and being caught out.
Where stale figures do the most damage
The damage concentrates in one place: a live negotiation. If you quote a figure describing last year to an employer who is working from current market data, you are arguing for a number the market has already moved past, and you will not be corrected. An employer offered a benchmark that favors them has no reason to point out that it is a year old.
The second place is career decisions with a long horizon — choosing a field, deciding whether a qualification is worth pursuing, weighing a move. There the drift matters less because you are reading the shape rather than the level, and shapes change slowly. A year of drift will not reverse an occupational comparison.
The third is comparison across sources, where mixing a figure describing one period with a figure describing another produces a difference that is entirely artificial. Two sources can agree perfectly and appear to disagree by four percent purely because one is a year ahead of the other. That gap is entirely artificial and entirely avoidable.
How to find the real date
Go to the original source rather than the page that quoted it. Statistical agencies state the reference period prominently, usually in the title of the dataset or the first line of the technical notes, and it takes one click from any article that cites them honestly. Articles that do not link the source are telling you something.
Where a page does not name its source at all, that is the finding. A salary figure with no stated origin and no reference period is not a data point, and the appropriate weight to give it is very little, however confidently it is presented and however well the page ranks in search. Search ranking measures popularity rather than provenance or accuracy.
Self-reported sites present a different problem, because they accumulate submissions over time rather than describing a single period. A page may be mixing figures from three years without saying so, and the aggregate has no reference period at all. Check whether the site lets you filter by recency, and prefer the recent slice even though the sample is smaller.
The trap in the other direction
Fresher is not automatically better, and this is where people overcorrect. A very recent figure from a small sample is less reliable than an older figure from a large one, because the sampling error on a thin sample is frequently larger than a year of wage drift. Sample size and recency are a trade-off rather than a hierarchy.
Posting data is the clearest example of that trade-off. It is genuinely current, which is its main virtue, and it describes what employers advertise rather than what they pay. It also skews toward roles that are hard to fill, since those get posted repeatedly, and toward employers who advertise externally rather than promoting internally.
So the choice between a large stale figure and a small fresh one is a real trade-off rather than an obvious call. The best answer is usually to use both and notice when they disagree, because the disagreement tells you whether the market has moved since the survey period. Two imperfect readings are worth more than one confident one.
When a year-old figure is fine
Most of the time it is genuinely fine, and it is worth saying so. If you are trying to establish whether a role pays around $60,000 or around $95,000, a year of drift changes nothing about the answer. If you are comparing two occupations, both figures come from the same vintage and the comparison is unaffected.
Percentile spreads are also stable in a way that levels are not. The ratio between the tenth and ninetieth percentile in an occupation describes its structure, and structure changes slowly even when levels move. So a year-old spread is nearly as good as a current one.
The rule that follows is worth carrying: staleness matters for levels and matters much less for shapes, ratios and comparisons. Most of the useful things you do with wage data are shape questions rather than level questions. That is worth remembering before discarding an older source.
What to do in a live negotiation
Use the freshest defensible source you have, and say what it is. “The published range for this occupation in this metro is X to Y as of the most recent survey, and posted ranges for comparable roles here are currently running at Z” is a sentence an employer can check, and checkability is what makes a benchmark persuasive rather than merely assertive. An employer can verify it in a minute.
Adjust upward for the lag rather than quoting the raw figure, and say that you have done so. Naming the adjustment is more credible than hiding it, because an employer who works with this data daily already knows the reference period and will notice if you appear not to. Naming the adjustment is what makes you credible rather than optimistic.
And prefer posted ranges from jurisdictions with disclosure requirements when they exist for your role. They are current, they are the employer’s own statement of what they expect to pay, and nobody can argue that they describe a period that has passed.
Common questions
How old is a typical published wage figure?
A large survey describes a reference period well before publication, so a figure released this spring typically describes pay from about a year earlier — before anybody copies it onto a page with a fresh date.
Why do dates drift further?
Because copying resets the timestamp. A figure quoted in an article and summarized elsewhere ends up on a page dated this month while describing last year, with nothing on the page untrue.
Where does staleness cause real damage?
In a live negotiation. Quoting a year-old benchmark to an employer working from current data means arguing for a number the market has passed, and nobody will correct you.
How do I find the real date?
Go to the original source. Statistical agencies state the reference period in the dataset title or the technical notes, one click from any article citing them honestly.
What if a page names no source?
That is the finding. A salary figure with no stated origin and no reference period deserves very little weight, however confidently presented.
Is fresher always better?
No. A very recent figure from a small sample can be less reliable than an older figure from a large one, because sampling error on a thin sample often exceeds a year of wage drift.
When is a year-old figure fine?
Most of the time. Staleness affects levels and matters much less for shapes, ratios and comparisons — and percentile spreads describe structure, which changes slowly.
What should I quote in a negotiation?
The freshest defensible source, named, with any lag adjustment stated openly. Posted ranges in disclosure jurisdictions are strongest — current, and the employer's own statement of what they expect to pay.