B2B contact data goes stale at roughly 20% to 25% a year. The figure on most vendor pages, 22.5% a year or 2.1% a month, is one number written two ways, and it leads back to a HubSpot calculator that credits MarketingSherpa research with no year, sample or method attached.
The strongest current evidence is not a decay study at all. It is US labor data: in January 2026, 20.6% of wage and salary workers had been with their current employer for a year or less, according to the Bureau of Labor Statistics.
So plan on about a fifth of any prospect list going bad every 12 months, and faster when your buyers are younger. Every figure I found above 40% a year has no method you can check. Below is each number I could trace, graded, then the math on a 10,000-contact list and a re-verification schedule.
Where does the 22.5% a year figure come from?
Follow the citations on most data decay articles and they end on the same page: HubSpot's Database Decay Simulation. It is an interactive calculator, built to show marketers how fast an email database shrinks, with one footnote for a source.
The page's headline number: an email marketing list loses roughly 22.5% of its records in a year. The footnote gives the only source: MarketingSherpa research showing that B2B data decays at 2.1% per month, which HubSpot annualizes to that same 22.5% figure.
There is no link to that research. No year, no sample size, no word on how anyone measured it. The other material the page cites is HubSpot's own 2013 and 2014 State of Inbound reports, which places the calculator in roughly that era. The page itself carries no publish date (checked September 28, 2026).
Two things follow.
First, 2.1% a month and 22.5% a year are one data point quoted twice. Compound 2.1% over 12 months and you get 22.48%.
Second, the original is hard to find. A search of MarketingSherpa's own site on September 28, 2026 did not turn up a page carrying the 2.1% figure. So a number from around 2013, whose underlying study nobody links to, still gets published in 2026 articles as the current benchmark.
It may still be roughly right. But it is old and thin, and you should know that before you plan a quarter of sending around it.
Which decay numbers hold up when you trace them?
I took every decay figure that shows up on the pages ranking for this question and followed each one to its origin. The grades:
- A: primary source, dated, method stated
- B: named source with a stated sample, but vendor data or a measure that sits next to decay
- C: traced to a real origin, no method published
- D: no source you can open
| Figure | Where you see it | Where it actually leads | Method you can check | Grade |
|---|---|---|---|---|
| 20.6% of workers had a year or less with their employer (January 2026); median tenure 4.1 years | The BLS release itself | US Bureau of Labor Statistics tenure release, published September 24, 2026 | Yes: national survey, dated | A (measures job changes, one cause of decay) |
| 23% of addresses invalid in 2025, following 22% in 2022, 23% in 2021, 25% in 2023 and 28% in 2024 | ZeroBounce's 2026 list decay report | More than 11 billion addresses ZeroBounce verified from January to December 2025, all industries and company sizes | Sample and period stated; counts addresses found invalid when checked, not one list followed over time; no B2B-only split | B |
| 22.5% a year, 2.1% a month | Most vendor pages on the topic | HubSpot's calculator, crediting MarketingSherpa, around 2013 to 2014 | None published | C |
| 22.5% a year, credited to a "Dun & Bradstreet B2B Data Benchmark" | Cleanlist's decay statistics page, updated July 2026 | A Dun & Bradstreet URL that returned a 404 error on September 28, 2026 | No year, sample or method given | D |
| "Between 22.5% and 70.3% annually" | Landbase blog post, April 2026, later repeated as "Landbase field-level analysis" on ZoomInfo's decay page | No source named in the post | None | D |
| Email about 3.6% a month, "about 43% a year" | ZoomInfo's decay page, updated June 2026 | Credited to a Landbase and SMARTe field-level analysis | None; and 43% is 3.6 x 12, while 3.6% compounded monthly comes to about 35.6% | D |
| Job titles 25% to 35% a year, phone numbers 15% to 25% a year | Landbase post | No source named | None | D |
| Startups 30% to 40% a year, government 8% to 12% a year | Cleanlist's industry table | No study or sample given for the rows | None | D |
The ZeroBounce row comes from a company that sells email verification, so read it as vendor data. It is still the only decay-style figure here with a sample size and a date range attached.
Read the grades together. Only one row is primary, and what it counts is people changing employers. The one vendor figure with a stated sample moved between 22% and 28% across five years. The old 22.5% sits inside that same band. Three weak signals pointing at one range are worth more than any single one of them.
The 43% and 70% figures stand alone with nothing underneath. Both come from companies that sell contact data, which is worth remembering when a page leads with them.
Why do published decay rates disagree so much?
None of the pages I read explains the spread. I see four reasons.
They measure different things. A verification check asks whether a mailbox still accepts mail. An accuracy check asks whether any field on the record is wrong. Someone promoted inside the same company keeps a working inbox and a wrong title. ZeroBounce's 23% is the share of addresses that failed when customers uploaded them, which mixes every kind of list, across all industries and company sizes.
Fields age at different speeds. An email address breaks when a person leaves. A title changes on promotion, even when the person stays put. Company name and industry barely move. The vendor pages I checked put work email at or near the top for speed, and the one that lists company details puts them at the bottom. None shows how it measured any of it.
The arithmetic slips. Multiply a monthly rate by 12 and you overstate it, because each month's loss comes out of a smaller list. That is how 3.6% a month becomes "43% a year" on one page when the compounded answer is closer to 36%.
Pages cite each other. One vendor states 70.3% with no source. A larger data provider repeats it as "field-level analysis." Another page gives the 22.5% a different parent entirely. After a few rounds, a number looks confirmed because many pages carry it.
What actually makes B2B contact data go stale?
People change jobs. The BLS release published on September 24, 2026 is the cleanest number I found. Median tenure with the current employer was 4.1 years in January 2026, up from 3.9 years in January 2024. The share with a year or less fell to 20.6%, from 22.2%.
The split by age matters more for a prospect list. Workers aged 25 to 34 had a median of 3.0 years with their employer. Workers aged 55 to 64 had 9.6 years. Management, professional and related occupations had the highest median tenure of the major occupation groups, 4.9 years.
If you sell to senior decision-makers, your records last longer than a list full of people under 35. That alone is a reason to distrust any single decay rate applied to every list.
Be honest about what the BLS number leaves out. It is US data only, so it says nothing direct about a list of German or Czech buyers. It misses promotions inside the same company, rebrands, email domain changes and acquisitions. It also counts people who just joined the workforce.
It is still the best sanity check I found. Roughly one worker in five being new to their employer fits a 20% to 25% annual decay range. It does not fit 70%.
What does 2.1% a month do to a 10,000-contact list?
The table below compounds the traced figure month by month, next to what the 70% claim would mean for the same list. The 70% column works out to about 9.5% lost every month.
| Months since the list was built | Still accurate at 2.1% a month | Changed | Still accurate if 70% a year were true |
|---|---|---|---|
| 1 | 9,790 | 210 | 9,045 |
| 3 | 9,383 | 617 | 7,401 |
| 6 | 8,804 | 1,196 | 5,477 |
| 12 | 7,752 | 2,248 | 3,000 |
| 24 | 6,009 | 3,991 | 900 |
| 36 | 4,658 | 5,342 | 270 |
Look at the 24-month row. Under the 70% claim, a two-year-old list would have 900 usable people left out of 10,000. Yet half of US wage and salary workers had been with their employer for 4.1 years or more in January 2026. A list cannot lose 91% of its people to job changes in 2 years when half the workforce has not moved in 4.
Now put the traced column next to a bounce ceiling: a 0.5% cap allows 50 bad addresses per 10,000 sends, no more. One month after the list is built, 210 records have changed. Not every changed record bounces: some mailboxes forward, and some domains accept mail for any address. But nothing on the list tells you which 210 they are until you verify again.
Many of these pages also quote a Gartner dollar figure for the yearly cost of bad data. As quoted, it covers poor data quality across a whole organization, which is a much wider problem than a stale prospect list. The list math above is closer to yours.
How often should you re-verify a B2B prospect list?
Smirnov Consulting Group is a Prague-based B2B outbound lead generation agency that runs cold email and LinkedIn campaigns for founder-led B2B companies and books qualified sales calls. On the lists I send to, the bounce rate sets the schedule.
My rule is short: verify every address before sending, and use NeverBounce or DeBounce to keep bounces below 0.5%.
A bounce rate climbing past that 0.5% mark tells you a list has gone bad while there is still time to stop. I explain the reasoning, and why I stop a list well before the looser ceilings some guides allow, in my weekly sender reputation routine.
This is the schedule I would hand a founder, with a note on how solid the reasoning behind each line is.
| Situation | What to do | Why, and how solid it is |
|---|---|---|
| New list, first send within days | Verify every address before it goes out | My own standard; firm |
| Contacts not yet sent, list older than 30 days | Re-verify them before the send | About 210 changes per 10,000 in a month at the traced rate, more than 4 times a 50-bounce budget if every change bounced; based on a C-grade number |
| Contacts sitting in a sequence for 3 months or more | Re-verify before the next step goes out | About 617 changes per 10,000 at the traced rate; C-grade number |
| CRM export 6 to 12 months old | Treat it as a new list: verify every address and re-check titles for the accounts you care most about | 12% to 22% changed at the traced rate; the BLS data puts one worker in five as new to their employer within a year |
| Anything older than a year, or a list bought cheap | Re-verify all of it, or rebuild it, before a single send | Decay plus unknown quality at the start |
| Any live campaign with a bounce rate above 0.5% | Stop sending from that list and re-verify, whatever the calendar says | 0.5% is the ceiling I hold every list to, and it is the only trigger you measure on your own list; firm |
The cheap-list row deserves one more line, because no decay table covers it. A cheap list from a freelancer brings low opens and a high bounce rate. A list like that can be stale before it ever reaches your inbox, so the monthly decay rate is the least of its problems.
Before paying for one, read my take on buying a ready-made email list, and before sending to any list, audit it line by line first.
If the export fails that audit, rebuild instead of patching. A rebuild starts with deciding which companies qualify, and I cover that step in how I pick the market a list is built for.
The 22.5% is a fair planning number that nobody can fully source. Use it to size how much re-verification you need, and let your own bounce rate decide when.
FAQ
Is the 22.5% a year B2B data decay figure still accurate?
Nobody can say for sure. It traces to a HubSpot calculator from around 2013 to 2014 that credits MarketingSherpa but links no study and gives no method. Newer evidence, the January 2026 BLS tenure data and one verification vendor's 22% to 28% for 2021 to 2025, lands in the same range.
How often should I clean a B2B email list?
Verify every address right before it is sent, whatever the list's age. Beyond that, re-verify anything left unsent for more than 30 days, treat a CRM export older than 6 months as a new list, and pause any campaign once bounces climb past the 0.5% mark, until the list is checked again.
Do some industries or roles decay faster than others?
Yes, but most published industry tables name no study behind the rows. The signal you can check is age: in January 2026, US workers aged 25 to 34 had a median of 3.0 years with their employer, against 9.6 years for ages 55 to 64. Lists of younger staff go stale faster.
Does email verification catch every decayed contact?
No. Verification tells you whether a mailbox accepts mail. It cannot tell you whether the person behind it still holds the title you targeted, and a domain that accepts mail for any address can make a departed contact look deliverable. For lists older than 6 months, re-check titles on your most important accounts too.