Every RevOps team eventually asks the same question: how stale is our database, really? The answer you find depends entirely on which 2018 vendor blog you land on first. Some say 22%. Some say 30%. A newer wave of 2026 studies says the real number, measured properly, is closer to 67% a year. All three can be true at once because they're measuring different things, at different intervals, with different definitions of "decayed." This article untangles that, gives you a method to measure your own database this week, and ends with a refresh schedule you can actually implement.
Data decay isn't one number it's three failure modes
When people say a contact "decayed," they usually mean one of three unrelated things:
- Contactability decay: the email bounces, the phone is disconnected, the LinkedIn profile is inactive. The channel is dead even if the person still works there.
- Identity decay: the person changed jobs, got promoted, or left the company entirely. The channel might still work, but you're now emailing the wrong human for the wrong role.
- Firmographic decay: the company itself changed acquired, renamed, downsized, moved HQ, changed its email domain format after a rebrand. Your targeting criteria (headcount, industry, tech stack) no longer match reality.
The 22% vs 67% debate, resolved
The "30% of B2B contact data decays annually" stat traces back to a 2017-2018 HBR/Gartner-adjacent citation that has been copy-pasted into hundreds of vendor blogs since, almost never re-tested. It measured a narrow slice: mid-to-senior job-title changes over 12 months in US enterprise data.
A newer 2026 methodology weekly re-verification of the same contact list over a full year rather than a single annual snapshot found something important: decay isn't linear, it compounds. If you check a list only once a year, you catch roughly 22-30% of contacts that have changed since your last check. But if you check weekly, you catch small decay events continuously, and when you sum the compounding weekly loss across 52 weeks, the effective annual decay rate for contactability plus identity combined lands closer to 60-67%.
Put simply: the old number measures decay at one point in time. The new number measures how much of your list has actually gone bad by the time you'd normally get around to re-checking it. Both are "correct" they're just answering different questions. For anyone still prospecting off a list purchased or exported once and never refreshed, 67% is the more honest planning assumption.
What actually drives the decay
- Job changes and promotions: LinkedIn's own economic graph data puts average U.S. white-collar job tenure at roughly 3-4 years, meaning in any given month a meaningful slice of your ICP is mid-transition.
- M&A and restructuring: acquisitions, rebrands, and layoffs don't just change one contact they invalidate entire company records, domains, and reporting structures at once.
- Email format shifts: companies migrating from
first.last@toflast@after a rebrand or email security overhaul silently break every previously-verified address. - Role consolidation: titles like "Head of Growth" get merged into "VP Marketing" during reorgs, breaking title-based segmentation even when the person hasn't left.
Three decay types, three KPIs, three fixes
| Decay type | What breaks | KPI to track | Primary fix |
|---|---|---|---|
| Contactability | Email bounces, disconnected numbers | Hard bounce rate, deliverability rate | Real-time email/phone verification pre-send |
| Identity | Person changed role or company | Reply-rate mismatch, auto-responder "I've moved" replies | Job-change monitoring, re-enrichment on trigger |
| Firmographic | Company details are outdated | Firmographic match rate on re-crawl | Periodic company re-crawl and domain re-verification |
How to measure your own database's decay rate this week
You don't need a data science team for this. Here's a method that takes one afternoon:
- Pull a random sample of 300-500 contacts added or last-verified 9-12 months ago random, not your best leads, or you'll bias the result.
- Run the emails through a verification API and record the hard bounce percentage. This is your baseline contactability decay.
- Spot-check 100 of those records manually (or via a LinkedIn/Sales Navigator lookup) to confirm current title and employer. Count how many no longer match your CRM record. This is your identity decay.
- Re-crawl 50 of the associated company domains and check headcount, industry tag, and tech stack against what's in your CRM. Count mismatches. This is your firmographic decay.
- Combine: decay rate = (bounced + identity-mismatched + firmographic-mismatched, deduplicated) / total sample.
Refresh cadence by outbound velocity
| Outbound volume | Recommended refresh cadence | Verification method |
|---|---|---|
| Under 500 contacts/month | Quarterly full re-verify | Batch email verification before each campaign |
| 500-5,000 contacts/month | Monthly full re-verify, weekly bounce monitoring | API-based verification integrated into sequencer |
| 5,000+ contacts/month or continuous prospecting | Continuous, per-send | Real-time verification at point of send, not batch |
Why waterfall enrichment now beats single-source data
Any single data provider no matter how good has coverage gaps and its own refresh lag. Waterfall enrichment (querying multiple providers in sequence and taking the first or best verified match) closes those gaps. Recent 2026 benchmark comparisons put single-source accuracy in the 65-75% range depending on provider and vertical, while waterfall approaches combining 3-4 sources land 10-15 points higher, typically 80-90%, because each provider's blind spots get covered by the next one in the chain. The tradeoff is cost and latency per lookup, which is why waterfall enrichment tends to make the most sense for high-value accounts and continuous prospecting rather than blanket one-time list buys.
Build a quarantine queue, not a delete button
Don't just purge bounced contacts quarantine them. A quarantine queue is a holding segment for any record that fails verification, shows a job-change signal, or hasn't been re-checked past your cadence threshold. Records sit there, blocked from active sequences, until they're re-verified or re-enriched. This does two things a simple delete doesn't: it prevents a single bad batch from tanking your sender reputation mid-campaign, and it preserves the record for re-activation once the person's new role or company is confirmed, instead of losing the lead entirely.
Continuous prospecting sidesteps the problem structurally
Every cadence table above is a workaround for the same root issue: a static list purchased or exported once starts decaying the moment it's created, and every day after that the gap between what's in your CRM and what's true in the world gets wider. The only way to actually eliminate decay, rather than manage it, is to stop treating prospecting as a one-time export and start treating it as a continuously running process search, crawl, verify, and score happening on a rolling basis instead of a quarterly refresh.
This is the structural argument for continuous or always-on prospecting over static list purchases: a list that's re-generated and re-verified on every run is never more than hours old, which is why [AutoReach](/register) runs lead discovery, site crawling, and contact verification as one continuous loop rather than a batch you refresh manually. Data doesn't get stale if it's never allowed to sit still long enough to decay.
FAQ
How often should I re-verify emails in my B2B database?
For low-volume outbound (under 500 contacts/month), quarterly batch verification before each campaign is sufficient. For anything above 5,000 contacts a month, verify at the point of send rather than on a calendar real-time verification catches decay that occurred since your last batch check, which at high send volume can otherwise cost you meaningful deliverability.
What's a good bounce rate for a B2B cold email list?
Under 2% hard bounce is the standard deliverability-safe threshold; above 5% risks inbox provider throttling or blacklisting. If your bounce rate is consistently above 3-4% on lists you believe are "recently verified," your verification method or refresh cadence is the problem, not the list itself.
Does GDPR or similar regulation require ongoing data accuracy upkeep?
Yes GDPR's accuracy principle (Article 5(1)(d)) requires personal data to be kept accurate and up to date, and CCPA/CPRA impose similar expectations around correcting outdated records on request. Treat your quarantine queue and refresh cadence as compliance infrastructure, not just a deliverability tactic, since a stale record that misidentifies someone's employer or role can itself become a data-accuracy complaint.
Is the 67% annual decay figure realistic for every industry?
No it's an upper bound observed in fast-moving sectors like SaaS, AI, and venture-backed startups where job tenure and company changes are frequent. Slower-moving verticals like manufacturing, government, and healthcare administration typically show decay closer to the older 20-30% range, so run the DIY measurement on your own list rather than assuming either number applies to you.
How is identity decay different from contactability decay in practice?
Contactability decay means the channel itself is broken the email bounces or the number is disconnected and it's caught by verification tools. Identity decay means the channel still works, but the person is no longer in the role or company you're targeting, which verification tools miss entirely and only job-change monitoring or manual spot-checks will catch. A list can have a 1% bounce rate and still have 20% identity decay, which is why bounce rate alone is a misleading health metric.