How Fast Does B2B Contact Data Decay? What the Evidence Shows

That range applies to US-heavy B2B lists of named contacts. It rises with junior, sales and services contacts and falls with executive and finance contacts. No public cohort study measures B2B decay by field and segment, so the range is built from labour and email-validity data, not from a direct measurement of any vendor's database.

How Fast Does B2B Contact Data Decay?

About one in five US workers is new to their employer in any given year, as the Bureau of Labor Statistics found 20.6% of wage and salary workers had a year or less of tenure in January 2026, which is the floor for how fast B2B contact records go out of date.

Employer change is the main driver of decay in B2B contact databases. When a person leaves a company, their work email, direct dial, title and company fields all go wrong at once. Tenure data is the cleanest public view of that churn because it counts people, not events, and the BLS sample is large and dated.

Tenure is a floor, not the full rate. It captures employer changes only. A contact promoted inside the same company still counts as a long-tenure worker, yet the title on their record is now wrong. The real share of business contact records that need an update each year is higher than the share of people who changed employer.

Email validity gives a second view. ZeroBounce, an email verification company, reported that 23% of the more than 11 billion email addresses it verified in 2025 were invalid. Its earlier editions put the invalid share at 23% for 2021, 22% for 2022, 25% for 2023 and 28% for 2024.

Read that number with its owner in mind. ZeroBounce measures lists its customers chose to submit for cleaning, from solo businesses to enterprises, and B2B and consumer addresses are mixed. Lists sent to a verifier skew toward older, dirtier files, so the 23% describes submitted lists, not a typical CRM.

Monthly labour churn is the third signal. The BLS Job Openings and Labor Turnover Survey (JOLTS) counted 5.3 million separations in September 2026, a monthly rate of 3.3%, with quits at 2.1% and layoffs and discharges at 1.0%. JOLTS counts separation events, not unique people. One person can leave two jobs in a year, so a monthly rate cannot be multiplied into an annual share of people who moved.

Put together, the evidence supports a planning range rather than a single number.

• Employer change: About a fifth of US workers are in their first year with an employer, per BLS.
• Role change: Title changes inside one company add to employer change, and they never show up as a bounce.
• Email validity: Around a quarter of submitted lists fail verification, per ZeroBounce, on files that skew old.

For budgeting and list hygiene, plan for roughly a fifth of contacts to go stale each year, then adjust by segment. A list of sales reps and junior marketers will decay faster than a list of CFOs.

Where Does The 22.5% Decay Figure Come From?

The 22.5% annual decay rate repeated across B2B data articles comes from HubSpot's Database Decay Simulation page, which credits 'Marketing Sherpas research' for a 2.1% monthly rate but links no report, sample or date. Most ranking articles on contact decay cite this one figure as their benchmark.

The page says email marketing databases "naturally degrade by about 22.5% every year". A footnote attributes that to research showing B2B data decays at 2.1% a month. There is no link to the underlying study, no description of how the decay was measured, no sample size and no year.

For a buyer, that changes what the number can do. A figure with no method cannot tell you whether it measured bounces, job changes or unsubscribes. It cannot tell you whether the list was B2B, consumer or mixed, or which year it describes. It is a secondhand estimate, not a measured cohort result, and it should not anchor a data budget or a vendor negotiation.

Compare it with the two dated sources in the previous section.

• BLS tenure: A named survey, about 60,000 households, January 2026, a stated definition of tenure.
• ZeroBounce: More than 11 billion addresses, a stated period of January to December 2025, a stated owner.
• HubSpot page: A single rate, credited to research that the page does not link or describe.

The 22.5% figure sits close to the BLS and ZeroBounce numbers, but being close is not the same as being sourced. Quote the dated figures, and treat 22.5% as folklore.

Which Contact Fields Decay Fastest?

Employer change breaks the most fields at once, because a contact who leaves a company takes the employer, work email, direct dial and title on their record out of date together. A promotion inside the same company breaks only the title and never shows up as a bounce, a change Apollo treats as its own job change alert.

Employer. BLS tenure data covers employer change. When someone changes employer, the company, domain, work email and direct dial all fail together. The BLS figure of 20.6% of workers with a year or less at their employer is a good proxy for how often that happens across a US workforce.

Job title. Title decay is invisible to bounce checks. Apollo's job change alerts flag saved contacts who move company and also fire when a contact changes title or position without changing company. That second trigger is the point. A promoted contact never bounces, but a sequence written for their old role misses. Apollo charges a credit to update a contact with a job change, and alert access depends on plan.

Email address. Verifiers such as ZeroBounce report on email validity. It still lags the real change. An inbox at a former employer can stay open, forward or turn catch-all for a while, so a list can pass verification and still hold people who left. ZeroBounce reported more than 9% of the addresses it checked in 2025 as catch-all, which a verifier cannot confirm either way.

Phone. A direct dial belongs to a desk and goes stale when the person leaves it. A mobile follows the person across employers, so it lasts longer but tells you less about their current company. Cognism and SalesIntel both verify phone numbers by hand. On Lusha, revealing a phone number costs five times the credits of a verified email.

RocketReach builds this layered view into its records. Its reliability score factors in job title changes, employer changes and company updates, shown as colour-coded letter grades. That is a useful model for your own data, since it scores a record on more than whether the email delivers.

Which Roles And Industries Decay Fastest?

Sales contacts go stale faster than contacts in management occupations, since median tenure in January 2026 was 3.3 years for sales and related occupations against 6.1 years for management occupations, according to the Bureau of Labor Statistics. Shorter tenure means more employer changes per year, so lists built on those roles need more frequent checks.

Three patterns matter for a contact list.

• Age and seniority. Workers aged 25 to 34 have a median of 3.0 years with their employer, against 9.6 years for ages 55 to 64. Younger workers have far shorter tenure with their current employer.
• Function. Sales and related roles sit at 3.3 years, well below management occupations at 6.1 years. Computer and mathematical roles sit in between at 4.6 years.
• Industry. Professional and business services, at 3.8 years, turns over faster than information at 4.9 years or financial activities at 5.0 years.

Monthly turnover data points the same way. JOLTS put the total separations rate for professional and business services at 4.2% in September 2026, against 3.3% for all industries. Those are monthly event counts, so they show which sectors churn more, not what share of people leave in a year.

Decision rule. Lists heavy in junior, sales and professional-services contacts need shorter re-check cycles than executive or finance lists. If two lists are the same size, the one built on SDRs and agency staff will lose accuracy faster.

Vendor coverage varies along the same lines. Lusha documentation says its own data quality varies by region, industry and seniority, and is stronger in English-speaking markets and technology. That variation is one reason large B2B databases can still miss your ICP, a problem that sits alongside decay rather than replacing it.

What Do Data Vendors Say About Refresh Cycles?

A vendor's refresh cycle is a re-check schedule, not an accuracy rate. Cognism says it refreshes 95% of VP-and-above contacts every 60 days, while SalesIntel says its team verifies every contact every 90 days. Several cycles apply only to part of the database, so read the scope before the number.

Apollo. Apollo states it runs monthly checks on its database and refreshes a record in real time when its system captures a signal such as a job change. Its sources are a contributor network of over two million contributors, its engagement suite, public crawling and third-party providers, with over 200 million records processed a month. Before relying on it, confirm your plan includes job change alerts, since that is where title decay gets caught.

Cognism. Cognism states it manually phone-verifies mobile numbers and refreshes 95% of VP-and-above contacts every 60 days. Its help centre calls a phone-verified contact a Diamond and describes other contacts as continuously improved and verified, with no set schedule. Diamonds-on-Demand, its requested verification, is expected to succeed for about 15 to 20% of US requests and 10 to 15% of UK requests. If most of your list sits below VP level, the 60-day cycle does not describe it.

SalesIntel. SalesIntel states every contact is verified every 90 days by a human research team layered on a proprietary data feed, with 1.75 million contacts verified a week. It names direct dial, mobile, business email and title as the human-verified fields, and it aims for 95% accuracy. That is a schedule and a target, not an audited accuracy figure. For high-churn segments, 90 days is a long gap.

RocketReach. RocketReach states its top 600 million most popular profiles refresh in real time whenever a user opens them, and all other contacts refresh on a rotating schedule as often as once a month. New data arrives daily. Niche or less-viewed contacts fall outside the real-time group, so check where your segment sits.

Lusha. Lusha says its data is refreshed continuously, with no specific schedule. Sources are public professional profiles, company websites, business directories and contributed data, and Lusha says it does not buy from third-party data brokers. Emails carry High, Medium or Low confidence labels. Compare those labels with your own bounce results before trusting them.

The pattern across all five is the same. Four of the five state a re-check schedule, sometimes for a subset, and Lusha states none. None of those statements is a measured accuracy rate for your segment. Outdated contacts are a recurring complaint across the category in review-site feedback. If you are weighing the best B2B contact databases against each other, ask each one for decay or accuracy results on a sample of your own ICP, not its refresh schedule.

How To Measure Contact Data Decay In Your Own CRM

To measure contact data decay, freeze a dated cohort of contacts you are permitted to process, re-check every record at 30, 90, 180 and 365 days, and divide the changed records by the cohort's starting size, never by the records that remain. That gives a rate for your list, your segments and your sources.

Public figures cannot do that job. BLS tenure counts employer changes among US workers. ZeroBounce counts invalid emails in lists customers submitted. Neither is a cohort measurement of your CRM, and neither sees title changes inside one company. A cohort on your own data does.

• Freeze a cohort of contacts you have a lawful basis to process, and record the baseline date.
• Record each contact's segment at baseline, covering seniority, function, industry, region and company size.
• Define the denominator once, as every contact in the cohort at baseline, never the ones still active.
• Re-check every record at 30, 90, 180 and 365 days against a public profile, company site or verifier.
• Code each change by cause, using one code per change and the date it was found.
• Report decay per interval and per segment, with counts beside every percentage.
• Publish the dates, sample size, denominator, check method and limitations with the result.

Freeze the cohort. Pick a sample large enough to split by segment, from records you are permitted to process. The frozen-sample idea is the same one used to evaluate B2B database coverage before buying. A sample chosen and dated before you look at outcomes stops you from cherry-picking records that happen to look clean.

Fix the denominator. Dividing by the records that survive hides decay. If 1,000 contacts start the cohort and you divide changes by the records still active at day 365, every departure shrinks the base and hides decay. Divide by the starting count every time.

Use cause codes. A single "bad record" flag cannot tell you what to fix. Code each change so the report shows which field is failing.

Re-check the same way each time. Use one method per interval, and note it. A record checked against a public profile at day 30 and a verifier at day 90 mixes two measures. Apollo's alerts show why the title check matters, since a new title at the same employer never shows up as a bounce.

Report counts, not only percentages. A 30% decay rate in a segment of ten contacts is three people. Show the count, the denominator and the rate together, so a reader can judge the size behind each number.

Publish the limits. State the cohort dates, the sample size, the source of the records, the check method and what the cohort cannot see. A rate published that way can be compared with next year's cohort and with any vendor's claim. Processing contact data for a study still needs a lawful basis where you operate. This is general information, not legal advice.

How Often Should You Re-Verify B2B Contact Data?

Re-verify a B2B contact right before it enters a sequence, and review junior, sales and professional-services records more often than executive lists, because median tenure for workers ages 25 to 34 was 3.0 years in January 2026, per BLS. A fixed calendar cadence wastes credits on stable records and misses fast-moving ones.

Before every send or call sequence. A record verified months ago may have changed since. Checking at the point of use catches the decay that matters, on the contacts you are about to spend effort on.

By segment, not by list. Set a shorter review cycle for high-churn segments and a longer one for stable ones.

• Shorter cycles suit sales roles, ages 25 to 34 and professional and business services, where median tenure runs 3.0 to 3.8 years.
• Longer cycles suit management, finance and older senior contacts, where median tenure runs 5.0 to 9.6 years.

Match the vendor's own cycle. If your vendor states it re-checks a record every 90 days, as SalesIntel does, re-revealing that record at day 30 pays for data the vendor has not refreshed. The same applies to Cognism's 60-day cycle for VP-and-above contacts and RocketReach's rotation of as often as monthly for profiles outside its top tier. Line your review dates up with the source's stated refresh, or re-check against an independent source instead.

Refresh phones selectively. On Lusha, revealing a phone number costs 5 credits against 1 credit for a verified email, so a full phone refresh of a large list costs five times an email-only refresh. Refresh phones for contacts you plan to call, not for the whole database.

Fix what you hold before buying more. When the CRM holds the records you will contact, re-verify them before buying a fresh list that starts decaying on day one. When the CRM is the larger asset, enrich the CRM records you already hold before adding new ones.

What The Evidence Cannot Tell You

No single public dataset measures B2B contact decay directly, since BLS tenure data tracks employer changes among US workers, while ZeroBounce's 2026 report covers more than 11 billion emails its customers submitted for verification in 2025. Each source measures one part of decay, and the planning range in this post is built from those parts.

• BLS tenure. It covers US wage and salary workers and counts time with the current employer. It does not track title changes, email changes or workers outside the US.
• BLS JOLTS. It counts separation events in a month, not people. A monthly rate cannot be turned into the share of people who changed jobs in a year.
• ZeroBounce. It reports on lists customers submitted for verification, mixing B2B and consumer addresses and skewing toward lists that needed cleaning.
• The 22.5% figure. HubSpot's page gives no method, sample or date for it, so it cannot be compared with anything.
• Vendor refresh cycles. They are re-check schedules stated by each vendor, often for a subset of records, not measured accuracy.

No public cohort study measures field-level decay by segment with a published method and dates. That is why the method in the measurement section is the most reliable route to a rate you can defend for your own list.

FAQ

Sources